pszemraj commited on
Commit
a614e98
1 Parent(s): 75588fc

Upload trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +4237 -0
trainer_state.json ADDED
@@ -0,0 +1,4237 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 4.0,
5
+ "global_step": 1404,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 1.8604651162790697e-05,
13
+ "loss": 1.2577,
14
+ "step": 2
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 3.7209302325581394e-05,
19
+ "loss": 1.2575,
20
+ "step": 4
21
+ },
22
+ {
23
+ "epoch": 0.02,
24
+ "learning_rate": 5.5813953488372095e-05,
25
+ "loss": 1.2283,
26
+ "step": 6
27
+ },
28
+ {
29
+ "epoch": 0.02,
30
+ "learning_rate": 7.441860465116279e-05,
31
+ "loss": 1.2528,
32
+ "step": 8
33
+ },
34
+ {
35
+ "epoch": 0.03,
36
+ "learning_rate": 9.30232558139535e-05,
37
+ "loss": 1.2608,
38
+ "step": 10
39
+ },
40
+ {
41
+ "epoch": 0.03,
42
+ "learning_rate": 0.00011162790697674419,
43
+ "loss": 1.2019,
44
+ "step": 12
45
+ },
46
+ {
47
+ "epoch": 0.04,
48
+ "learning_rate": 0.0001302325581395349,
49
+ "loss": 1.2713,
50
+ "step": 14
51
+ },
52
+ {
53
+ "epoch": 0.05,
54
+ "learning_rate": 0.00014883720930232558,
55
+ "loss": 1.236,
56
+ "step": 16
57
+ },
58
+ {
59
+ "epoch": 0.05,
60
+ "learning_rate": 0.00016744186046511629,
61
+ "loss": 1.2529,
62
+ "step": 18
63
+ },
64
+ {
65
+ "epoch": 0.06,
66
+ "learning_rate": 0.000186046511627907,
67
+ "loss": 1.2585,
68
+ "step": 20
69
+ },
70
+ {
71
+ "epoch": 0.06,
72
+ "learning_rate": 0.0002046511627906977,
73
+ "loss": 1.2354,
74
+ "step": 22
75
+ },
76
+ {
77
+ "epoch": 0.07,
78
+ "learning_rate": 0.00022325581395348838,
79
+ "loss": 1.2448,
80
+ "step": 24
81
+ },
82
+ {
83
+ "epoch": 0.07,
84
+ "learning_rate": 0.00024186046511627906,
85
+ "loss": 1.2007,
86
+ "step": 26
87
+ },
88
+ {
89
+ "epoch": 0.08,
90
+ "learning_rate": 0.0002604651162790698,
91
+ "loss": 1.2464,
92
+ "step": 28
93
+ },
94
+ {
95
+ "epoch": 0.09,
96
+ "learning_rate": 0.0002790697674418605,
97
+ "loss": 1.2414,
98
+ "step": 30
99
+ },
100
+ {
101
+ "epoch": 0.09,
102
+ "learning_rate": 0.00029767441860465116,
103
+ "loss": 1.2117,
104
+ "step": 32
105
+ },
106
+ {
107
+ "epoch": 0.1,
108
+ "learning_rate": 0.00031627906976744186,
109
+ "loss": 1.2624,
110
+ "step": 34
111
+ },
112
+ {
113
+ "epoch": 0.1,
114
+ "learning_rate": 0.00033488372093023257,
115
+ "loss": 1.2599,
116
+ "step": 36
117
+ },
118
+ {
119
+ "epoch": 0.11,
120
+ "learning_rate": 0.0003534883720930233,
121
+ "loss": 1.2381,
122
+ "step": 38
123
+ },
124
+ {
125
+ "epoch": 0.11,
126
+ "learning_rate": 0.000372093023255814,
127
+ "loss": 1.2065,
128
+ "step": 40
129
+ },
130
+ {
131
+ "epoch": 0.12,
132
+ "learning_rate": 0.00039069767441860464,
133
+ "loss": 1.1707,
134
+ "step": 42
135
+ },
136
+ {
137
+ "epoch": 0.13,
138
+ "learning_rate": 0.0003999994671765845,
139
+ "loss": 1.1943,
140
+ "step": 44
141
+ },
142
+ {
143
+ "epoch": 0.13,
144
+ "learning_rate": 0.00039999520460629425,
145
+ "loss": 1.1769,
146
+ "step": 46
147
+ },
148
+ {
149
+ "epoch": 0.14,
150
+ "learning_rate": 0.0003999866795565616,
151
+ "loss": 1.1891,
152
+ "step": 48
153
+ },
154
+ {
155
+ "epoch": 0.14,
156
+ "learning_rate": 0.0003999738922090801,
157
+ "loss": 1.1709,
158
+ "step": 50
159
+ },
160
+ {
161
+ "epoch": 0.15,
162
+ "learning_rate": 0.00039995684283638535,
163
+ "loss": 1.1636,
164
+ "step": 52
165
+ },
166
+ {
167
+ "epoch": 0.15,
168
+ "learning_rate": 0.00039993553180184906,
169
+ "loss": 1.1606,
170
+ "step": 54
171
+ },
172
+ {
173
+ "epoch": 0.16,
174
+ "learning_rate": 0.00039990995955967125,
175
+ "loss": 1.1322,
176
+ "step": 56
177
+ },
178
+ {
179
+ "epoch": 0.17,
180
+ "learning_rate": 0.0003998801266548709,
181
+ "loss": 1.1243,
182
+ "step": 58
183
+ },
184
+ {
185
+ "epoch": 0.17,
186
+ "learning_rate": 0.0003998460337232739,
187
+ "loss": 1.1555,
188
+ "step": 60
189
+ },
190
+ {
191
+ "epoch": 0.18,
192
+ "learning_rate": 0.0003998076814914998,
193
+ "loss": 1.0912,
194
+ "step": 62
195
+ },
196
+ {
197
+ "epoch": 0.18,
198
+ "learning_rate": 0.0003997650707769462,
199
+ "loss": 1.1373,
200
+ "step": 64
201
+ },
202
+ {
203
+ "epoch": 0.19,
204
+ "learning_rate": 0.00039971820248777137,
205
+ "loss": 1.1136,
206
+ "step": 66
207
+ },
208
+ {
209
+ "epoch": 0.19,
210
+ "learning_rate": 0.0003996670776228748,
211
+ "loss": 1.1003,
212
+ "step": 68
213
+ },
214
+ {
215
+ "epoch": 0.2,
216
+ "learning_rate": 0.00039961169727187605,
217
+ "loss": 1.1401,
218
+ "step": 70
219
+ },
220
+ {
221
+ "epoch": 0.21,
222
+ "learning_rate": 0.0003995520626150915,
223
+ "loss": 1.0842,
224
+ "step": 72
225
+ },
226
+ {
227
+ "epoch": 0.21,
228
+ "learning_rate": 0.00039948817492350914,
229
+ "loss": 1.1078,
230
+ "step": 74
231
+ },
232
+ {
233
+ "epoch": 0.22,
234
+ "learning_rate": 0.0003994200355587614,
235
+ "loss": 1.0637,
236
+ "step": 76
237
+ },
238
+ {
239
+ "epoch": 0.22,
240
+ "learning_rate": 0.00039934764597309634,
241
+ "loss": 1.1324,
242
+ "step": 78
243
+ },
244
+ {
245
+ "epoch": 0.23,
246
+ "learning_rate": 0.0003992710077093466,
247
+ "loss": 1.0891,
248
+ "step": 80
249
+ },
250
+ {
251
+ "epoch": 0.23,
252
+ "learning_rate": 0.0003991901224008964,
253
+ "loss": 1.0628,
254
+ "step": 82
255
+ },
256
+ {
257
+ "epoch": 0.24,
258
+ "learning_rate": 0.00039910499177164686,
259
+ "loss": 1.0904,
260
+ "step": 84
261
+ },
262
+ {
263
+ "epoch": 0.25,
264
+ "learning_rate": 0.0003990156176359794,
265
+ "loss": 1.0331,
266
+ "step": 86
267
+ },
268
+ {
269
+ "epoch": 0.25,
270
+ "learning_rate": 0.00039892200189871666,
271
+ "loss": 1.06,
272
+ "step": 88
273
+ },
274
+ {
275
+ "epoch": 0.26,
276
+ "learning_rate": 0.00039882414655508235,
277
+ "loss": 1.0381,
278
+ "step": 90
279
+ },
280
+ {
281
+ "epoch": 0.26,
282
+ "learning_rate": 0.0003987220536906583,
283
+ "loss": 1.0759,
284
+ "step": 92
285
+ },
286
+ {
287
+ "epoch": 0.27,
288
+ "learning_rate": 0.0003986157254813405,
289
+ "loss": 1.032,
290
+ "step": 94
291
+ },
292
+ {
293
+ "epoch": 0.27,
294
+ "learning_rate": 0.00039850516419329227,
295
+ "loss": 1.0713,
296
+ "step": 96
297
+ },
298
+ {
299
+ "epoch": 0.28,
300
+ "learning_rate": 0.0003983903721828961,
301
+ "loss": 1.063,
302
+ "step": 98
303
+ },
304
+ {
305
+ "epoch": 0.28,
306
+ "learning_rate": 0.0003982713518967037,
307
+ "loss": 1.0361,
308
+ "step": 100
309
+ },
310
+ {
311
+ "epoch": 0.29,
312
+ "learning_rate": 0.00039814810587138346,
313
+ "loss": 1.0341,
314
+ "step": 102
315
+ },
316
+ {
317
+ "epoch": 0.3,
318
+ "learning_rate": 0.0003980206367336666,
319
+ "loss": 1.01,
320
+ "step": 104
321
+ },
322
+ {
323
+ "epoch": 0.3,
324
+ "learning_rate": 0.0003978889472002911,
325
+ "loss": 1.0464,
326
+ "step": 106
327
+ },
328
+ {
329
+ "epoch": 0.31,
330
+ "learning_rate": 0.0003977530400779441,
331
+ "loss": 1.0334,
332
+ "step": 108
333
+ },
334
+ {
335
+ "epoch": 0.31,
336
+ "learning_rate": 0.0003976129182632014,
337
+ "loss": 1.0365,
338
+ "step": 110
339
+ },
340
+ {
341
+ "epoch": 0.32,
342
+ "learning_rate": 0.00039746858474246647,
343
+ "loss": 1.0019,
344
+ "step": 112
345
+ },
346
+ {
347
+ "epoch": 0.32,
348
+ "learning_rate": 0.0003973200425919064,
349
+ "loss": 1.0059,
350
+ "step": 114
351
+ },
352
+ {
353
+ "epoch": 0.33,
354
+ "learning_rate": 0.0003971672949773864,
355
+ "loss": 1.06,
356
+ "step": 116
357
+ },
358
+ {
359
+ "epoch": 0.34,
360
+ "learning_rate": 0.00039701034515440234,
361
+ "loss": 0.9708,
362
+ "step": 118
363
+ },
364
+ {
365
+ "epoch": 0.34,
366
+ "learning_rate": 0.00039684919646801146,
367
+ "loss": 0.9961,
368
+ "step": 120
369
+ },
370
+ {
371
+ "epoch": 0.35,
372
+ "learning_rate": 0.00039668385235276085,
373
+ "loss": 1.0084,
374
+ "step": 122
375
+ },
376
+ {
377
+ "epoch": 0.35,
378
+ "learning_rate": 0.0003965143163326145,
379
+ "loss": 0.982,
380
+ "step": 124
381
+ },
382
+ {
383
+ "epoch": 0.36,
384
+ "learning_rate": 0.00039634059202087814,
385
+ "loss": 1.0116,
386
+ "step": 126
387
+ },
388
+ {
389
+ "epoch": 0.36,
390
+ "learning_rate": 0.0003961626831201219,
391
+ "loss": 0.9486,
392
+ "step": 128
393
+ },
394
+ {
395
+ "epoch": 0.37,
396
+ "learning_rate": 0.0003959805934221019,
397
+ "loss": 0.9435,
398
+ "step": 130
399
+ },
400
+ {
401
+ "epoch": 0.38,
402
+ "learning_rate": 0.0003957943268076793,
403
+ "loss": 0.9654,
404
+ "step": 132
405
+ },
406
+ {
407
+ "epoch": 0.38,
408
+ "learning_rate": 0.00039560388724673725,
409
+ "loss": 0.9505,
410
+ "step": 134
411
+ },
412
+ {
413
+ "epoch": 0.39,
414
+ "learning_rate": 0.00039540927879809664,
415
+ "loss": 0.9705,
416
+ "step": 136
417
+ },
418
+ {
419
+ "epoch": 0.39,
420
+ "learning_rate": 0.0003952105056094295,
421
+ "loss": 0.9316,
422
+ "step": 138
423
+ },
424
+ {
425
+ "epoch": 0.4,
426
+ "learning_rate": 0.00039500757191717053,
427
+ "loss": 0.9281,
428
+ "step": 140
429
+ },
430
+ {
431
+ "epoch": 0.4,
432
+ "learning_rate": 0.0003948004820464269,
433
+ "loss": 0.9292,
434
+ "step": 142
435
+ },
436
+ {
437
+ "epoch": 0.41,
438
+ "learning_rate": 0.00039458924041088606,
439
+ "loss": 0.9307,
440
+ "step": 144
441
+ },
442
+ {
443
+ "epoch": 0.42,
444
+ "learning_rate": 0.00039437385151272155,
445
+ "loss": 0.9195,
446
+ "step": 146
447
+ },
448
+ {
449
+ "epoch": 0.42,
450
+ "learning_rate": 0.0003941543199424972,
451
+ "loss": 0.9232,
452
+ "step": 148
453
+ },
454
+ {
455
+ "epoch": 0.43,
456
+ "learning_rate": 0.0003939306503790693,
457
+ "loss": 0.9023,
458
+ "step": 150
459
+ },
460
+ {
461
+ "epoch": 0.43,
462
+ "learning_rate": 0.00039370284758948667,
463
+ "loss": 0.9184,
464
+ "step": 152
465
+ },
466
+ {
467
+ "epoch": 0.44,
468
+ "learning_rate": 0.0003934709164288892,
469
+ "loss": 0.8961,
470
+ "step": 154
471
+ },
472
+ {
473
+ "epoch": 0.44,
474
+ "learning_rate": 0.0003932348618404045,
475
+ "loss": 0.9495,
476
+ "step": 156
477
+ },
478
+ {
479
+ "epoch": 0.45,
480
+ "learning_rate": 0.0003929946888550423,
481
+ "loss": 0.9092,
482
+ "step": 158
483
+ },
484
+ {
485
+ "epoch": 0.46,
486
+ "learning_rate": 0.0003927504025915875,
487
+ "loss": 0.9123,
488
+ "step": 160
489
+ },
490
+ {
491
+ "epoch": 0.46,
492
+ "learning_rate": 0.0003925020082564907,
493
+ "loss": 0.8893,
494
+ "step": 162
495
+ },
496
+ {
497
+ "epoch": 0.47,
498
+ "learning_rate": 0.00039224951114375764,
499
+ "loss": 0.8835,
500
+ "step": 164
501
+ },
502
+ {
503
+ "epoch": 0.47,
504
+ "learning_rate": 0.0003919929166348361,
505
+ "loss": 0.9182,
506
+ "step": 166
507
+ },
508
+ {
509
+ "epoch": 0.48,
510
+ "learning_rate": 0.00039173223019850126,
511
+ "loss": 0.8915,
512
+ "step": 168
513
+ },
514
+ {
515
+ "epoch": 0.48,
516
+ "learning_rate": 0.00039146745739073927,
517
+ "loss": 0.9062,
518
+ "step": 170
519
+ },
520
+ {
521
+ "epoch": 0.49,
522
+ "learning_rate": 0.00039119860385462867,
523
+ "loss": 0.8893,
524
+ "step": 172
525
+ },
526
+ {
527
+ "epoch": 0.5,
528
+ "learning_rate": 0.00039092567532022024,
529
+ "loss": 0.8561,
530
+ "step": 174
531
+ },
532
+ {
533
+ "epoch": 0.5,
534
+ "learning_rate": 0.00039064867760441464,
535
+ "loss": 0.8803,
536
+ "step": 176
537
+ },
538
+ {
539
+ "epoch": 0.51,
540
+ "learning_rate": 0.0003903676166108389,
541
+ "loss": 0.874,
542
+ "step": 178
543
+ },
544
+ {
545
+ "epoch": 0.51,
546
+ "learning_rate": 0.0003900824983297202,
547
+ "loss": 0.852,
548
+ "step": 180
549
+ },
550
+ {
551
+ "epoch": 0.52,
552
+ "learning_rate": 0.00038979332883775817,
553
+ "loss": 0.8907,
554
+ "step": 182
555
+ },
556
+ {
557
+ "epoch": 0.52,
558
+ "learning_rate": 0.00038950011429799574,
559
+ "loss": 0.8538,
560
+ "step": 184
561
+ },
562
+ {
563
+ "epoch": 0.53,
564
+ "learning_rate": 0.00038920286095968755,
565
+ "loss": 0.8809,
566
+ "step": 186
567
+ },
568
+ {
569
+ "epoch": 0.54,
570
+ "learning_rate": 0.0003889015751581666,
571
+ "loss": 0.8811,
572
+ "step": 188
573
+ },
574
+ {
575
+ "epoch": 0.54,
576
+ "learning_rate": 0.00038859626331470967,
577
+ "loss": 0.8574,
578
+ "step": 190
579
+ },
580
+ {
581
+ "epoch": 0.55,
582
+ "learning_rate": 0.0003882869319363999,
583
+ "loss": 0.8522,
584
+ "step": 192
585
+ },
586
+ {
587
+ "epoch": 0.55,
588
+ "learning_rate": 0.0003879735876159887,
589
+ "loss": 0.889,
590
+ "step": 194
591
+ },
592
+ {
593
+ "epoch": 0.56,
594
+ "learning_rate": 0.00038765623703175474,
595
+ "loss": 0.8101,
596
+ "step": 196
597
+ },
598
+ {
599
+ "epoch": 0.56,
600
+ "learning_rate": 0.0003873348869473619,
601
+ "loss": 0.8479,
602
+ "step": 198
603
+ },
604
+ {
605
+ "epoch": 0.57,
606
+ "learning_rate": 0.00038700954421171503,
607
+ "loss": 0.797,
608
+ "step": 200
609
+ },
610
+ {
611
+ "epoch": 0.58,
612
+ "learning_rate": 0.00038668021575881406,
613
+ "loss": 0.7963,
614
+ "step": 202
615
+ },
616
+ {
617
+ "epoch": 0.58,
618
+ "learning_rate": 0.00038634690860760607,
619
+ "loss": 0.8253,
620
+ "step": 204
621
+ },
622
+ {
623
+ "epoch": 0.59,
624
+ "learning_rate": 0.00038600962986183574,
625
+ "loss": 0.8512,
626
+ "step": 206
627
+ },
628
+ {
629
+ "epoch": 0.59,
630
+ "learning_rate": 0.00038566838670989407,
631
+ "loss": 0.8082,
632
+ "step": 208
633
+ },
634
+ {
635
+ "epoch": 0.6,
636
+ "learning_rate": 0.00038532318642466506,
637
+ "loss": 0.7988,
638
+ "step": 210
639
+ },
640
+ {
641
+ "epoch": 0.6,
642
+ "learning_rate": 0.0003849740363633706,
643
+ "loss": 0.8149,
644
+ "step": 212
645
+ },
646
+ {
647
+ "epoch": 0.61,
648
+ "learning_rate": 0.00038462094396741407,
649
+ "loss": 0.8348,
650
+ "step": 214
651
+ },
652
+ {
653
+ "epoch": 0.62,
654
+ "learning_rate": 0.0003842639167622212,
655
+ "loss": 0.787,
656
+ "step": 216
657
+ },
658
+ {
659
+ "epoch": 0.62,
660
+ "learning_rate": 0.00038390296235707997,
661
+ "loss": 0.8334,
662
+ "step": 218
663
+ },
664
+ {
665
+ "epoch": 0.63,
666
+ "learning_rate": 0.00038353808844497866,
667
+ "loss": 0.8091,
668
+ "step": 220
669
+ },
670
+ {
671
+ "epoch": 0.63,
672
+ "learning_rate": 0.0003831693028024414,
673
+ "loss": 0.7977,
674
+ "step": 222
675
+ },
676
+ {
677
+ "epoch": 0.64,
678
+ "learning_rate": 0.00038279661328936273,
679
+ "loss": 0.7975,
680
+ "step": 224
681
+ },
682
+ {
683
+ "epoch": 0.64,
684
+ "learning_rate": 0.00038242002784884006,
685
+ "loss": 0.8279,
686
+ "step": 226
687
+ },
688
+ {
689
+ "epoch": 0.65,
690
+ "learning_rate": 0.0003820395545070044,
691
+ "loss": 0.8101,
692
+ "step": 228
693
+ },
694
+ {
695
+ "epoch": 0.66,
696
+ "learning_rate": 0.000381655201372849,
697
+ "loss": 0.8091,
698
+ "step": 230
699
+ },
700
+ {
701
+ "epoch": 0.66,
702
+ "learning_rate": 0.00038126697663805705,
703
+ "loss": 0.7846,
704
+ "step": 232
705
+ },
706
+ {
707
+ "epoch": 0.67,
708
+ "learning_rate": 0.0003808748885768267,
709
+ "loss": 0.7447,
710
+ "step": 234
711
+ },
712
+ {
713
+ "epoch": 0.67,
714
+ "learning_rate": 0.0003804789455456948,
715
+ "loss": 0.7934,
716
+ "step": 236
717
+ },
718
+ {
719
+ "epoch": 0.68,
720
+ "learning_rate": 0.0003800791559833588,
721
+ "loss": 0.7674,
722
+ "step": 238
723
+ },
724
+ {
725
+ "epoch": 0.68,
726
+ "learning_rate": 0.000379675528410497,
727
+ "loss": 0.8027,
728
+ "step": 240
729
+ },
730
+ {
731
+ "epoch": 0.69,
732
+ "learning_rate": 0.0003792680714295868,
733
+ "loss": 0.7721,
734
+ "step": 242
735
+ },
736
+ {
737
+ "epoch": 0.7,
738
+ "learning_rate": 0.0003788567937247215,
739
+ "loss": 0.7928,
740
+ "step": 244
741
+ },
742
+ {
743
+ "epoch": 0.7,
744
+ "learning_rate": 0.0003784417040614249,
745
+ "loss": 0.7528,
746
+ "step": 246
747
+ },
748
+ {
749
+ "epoch": 0.71,
750
+ "learning_rate": 0.00037802281128646514,
751
+ "loss": 0.7643,
752
+ "step": 248
753
+ },
754
+ {
755
+ "epoch": 0.71,
756
+ "learning_rate": 0.0003776001243276653,
757
+ "loss": 0.7411,
758
+ "step": 250
759
+ },
760
+ {
761
+ "epoch": 0.72,
762
+ "learning_rate": 0.00037717365219371387,
763
+ "loss": 0.7973,
764
+ "step": 252
765
+ },
766
+ {
767
+ "epoch": 0.72,
768
+ "learning_rate": 0.00037674340397397224,
769
+ "loss": 0.7675,
770
+ "step": 254
771
+ },
772
+ {
773
+ "epoch": 0.73,
774
+ "learning_rate": 0.0003763093888382813,
775
+ "loss": 0.7273,
776
+ "step": 256
777
+ },
778
+ {
779
+ "epoch": 0.74,
780
+ "learning_rate": 0.00037587161603676577,
781
+ "loss": 0.7177,
782
+ "step": 258
783
+ },
784
+ {
785
+ "epoch": 0.74,
786
+ "learning_rate": 0.00037543009489963714,
787
+ "loss": 0.7306,
788
+ "step": 260
789
+ },
790
+ {
791
+ "epoch": 0.75,
792
+ "learning_rate": 0.0003749848348369949,
793
+ "loss": 0.776,
794
+ "step": 262
795
+ },
796
+ {
797
+ "epoch": 0.75,
798
+ "learning_rate": 0.000374535845338626,
799
+ "loss": 0.7211,
800
+ "step": 264
801
+ },
802
+ {
803
+ "epoch": 0.76,
804
+ "learning_rate": 0.0003740831359738023,
805
+ "loss": 0.755,
806
+ "step": 266
807
+ },
808
+ {
809
+ "epoch": 0.76,
810
+ "learning_rate": 0.00037362671639107705,
811
+ "loss": 0.7184,
812
+ "step": 268
813
+ },
814
+ {
815
+ "epoch": 0.77,
816
+ "learning_rate": 0.00037316659631807887,
817
+ "loss": 0.7762,
818
+ "step": 270
819
+ },
820
+ {
821
+ "epoch": 0.77,
822
+ "learning_rate": 0.00037270278556130463,
823
+ "loss": 0.7511,
824
+ "step": 272
825
+ },
826
+ {
827
+ "epoch": 0.78,
828
+ "learning_rate": 0.00037223529400591055,
829
+ "loss": 0.6934,
830
+ "step": 274
831
+ },
832
+ {
833
+ "epoch": 0.79,
834
+ "learning_rate": 0.000371764131615501,
835
+ "loss": 0.734,
836
+ "step": 276
837
+ },
838
+ {
839
+ "epoch": 0.79,
840
+ "learning_rate": 0.00037128930843191707,
841
+ "loss": 0.6909,
842
+ "step": 278
843
+ },
844
+ {
845
+ "epoch": 0.8,
846
+ "learning_rate": 0.00037081083457502146,
847
+ "loss": 0.7328,
848
+ "step": 280
849
+ },
850
+ {
851
+ "epoch": 0.8,
852
+ "learning_rate": 0.00037032872024248377,
853
+ "loss": 0.7237,
854
+ "step": 282
855
+ },
856
+ {
857
+ "epoch": 0.81,
858
+ "learning_rate": 0.0003698429757095623,
859
+ "loss": 0.695,
860
+ "step": 284
861
+ },
862
+ {
863
+ "epoch": 0.81,
864
+ "learning_rate": 0.0003693536113288859,
865
+ "loss": 0.7213,
866
+ "step": 286
867
+ },
868
+ {
869
+ "epoch": 0.82,
870
+ "learning_rate": 0.00036886063753023264,
871
+ "loss": 0.7162,
872
+ "step": 288
873
+ },
874
+ {
875
+ "epoch": 0.83,
876
+ "learning_rate": 0.0003683640648203078,
877
+ "loss": 0.7244,
878
+ "step": 290
879
+ },
880
+ {
881
+ "epoch": 0.83,
882
+ "learning_rate": 0.00036786390378252,
883
+ "loss": 0.7237,
884
+ "step": 292
885
+ },
886
+ {
887
+ "epoch": 0.84,
888
+ "learning_rate": 0.00036736016507675566,
889
+ "loss": 0.7422,
890
+ "step": 294
891
+ },
892
+ {
893
+ "epoch": 0.84,
894
+ "learning_rate": 0.00036685285943915144,
895
+ "loss": 0.6849,
896
+ "step": 296
897
+ },
898
+ {
899
+ "epoch": 0.85,
900
+ "learning_rate": 0.0003663419976818659,
901
+ "loss": 0.6817,
902
+ "step": 298
903
+ },
904
+ {
905
+ "epoch": 0.85,
906
+ "learning_rate": 0.0003658275906928488,
907
+ "loss": 0.6784,
908
+ "step": 300
909
+ },
910
+ {
911
+ "epoch": 0.86,
912
+ "learning_rate": 0.000365309649435609,
913
+ "loss": 0.7205,
914
+ "step": 302
915
+ },
916
+ {
917
+ "epoch": 0.87,
918
+ "learning_rate": 0.0003647881849489811,
919
+ "loss": 0.7011,
920
+ "step": 304
921
+ },
922
+ {
923
+ "epoch": 0.87,
924
+ "learning_rate": 0.0003642632083468898,
925
+ "loss": 0.6943,
926
+ "step": 306
927
+ },
928
+ {
929
+ "epoch": 0.88,
930
+ "learning_rate": 0.00036373473081811327,
931
+ "loss": 0.7404,
932
+ "step": 308
933
+ },
934
+ {
935
+ "epoch": 0.88,
936
+ "learning_rate": 0.00036320276362604453,
937
+ "loss": 0.6986,
938
+ "step": 310
939
+ },
940
+ {
941
+ "epoch": 0.89,
942
+ "learning_rate": 0.0003626673181084516,
943
+ "loss": 0.6867,
944
+ "step": 312
945
+ },
946
+ {
947
+ "epoch": 0.89,
948
+ "learning_rate": 0.00036212840567723554,
949
+ "loss": 0.6806,
950
+ "step": 314
951
+ },
952
+ {
953
+ "epoch": 0.9,
954
+ "learning_rate": 0.00036158603781818766,
955
+ "loss": 0.6978,
956
+ "step": 316
957
+ },
958
+ {
959
+ "epoch": 0.91,
960
+ "learning_rate": 0.00036104022609074435,
961
+ "loss": 0.6866,
962
+ "step": 318
963
+ },
964
+ {
965
+ "epoch": 0.91,
966
+ "learning_rate": 0.00036049098212774083,
967
+ "loss": 0.6515,
968
+ "step": 320
969
+ },
970
+ {
971
+ "epoch": 0.92,
972
+ "learning_rate": 0.0003599383176351633,
973
+ "loss": 0.6801,
974
+ "step": 322
975
+ },
976
+ {
977
+ "epoch": 0.92,
978
+ "learning_rate": 0.00035938224439189935,
979
+ "loss": 0.6599,
980
+ "step": 324
981
+ },
982
+ {
983
+ "epoch": 0.93,
984
+ "learning_rate": 0.000358822774249487,
985
+ "loss": 0.6567,
986
+ "step": 326
987
+ },
988
+ {
989
+ "epoch": 0.93,
990
+ "learning_rate": 0.0003582599191318621,
991
+ "loss": 0.7001,
992
+ "step": 328
993
+ },
994
+ {
995
+ "epoch": 0.94,
996
+ "learning_rate": 0.000357693691035104,
997
+ "loss": 0.6653,
998
+ "step": 330
999
+ },
1000
+ {
1001
+ "epoch": 0.95,
1002
+ "learning_rate": 0.00035712410202718033,
1003
+ "loss": 0.6879,
1004
+ "step": 332
1005
+ },
1006
+ {
1007
+ "epoch": 0.95,
1008
+ "learning_rate": 0.00035655116424768916,
1009
+ "loss": 0.6308,
1010
+ "step": 334
1011
+ },
1012
+ {
1013
+ "epoch": 0.96,
1014
+ "learning_rate": 0.0003559748899076009,
1015
+ "loss": 0.6673,
1016
+ "step": 336
1017
+ },
1018
+ {
1019
+ "epoch": 0.96,
1020
+ "learning_rate": 0.0003553952912889978,
1021
+ "loss": 0.6634,
1022
+ "step": 338
1023
+ },
1024
+ {
1025
+ "epoch": 0.97,
1026
+ "learning_rate": 0.0003548123807448118,
1027
+ "loss": 0.6576,
1028
+ "step": 340
1029
+ },
1030
+ {
1031
+ "epoch": 0.97,
1032
+ "learning_rate": 0.00035422617069856197,
1033
+ "loss": 0.6552,
1034
+ "step": 342
1035
+ },
1036
+ {
1037
+ "epoch": 0.98,
1038
+ "learning_rate": 0.0003536366736440892,
1039
+ "loss": 0.6709,
1040
+ "step": 344
1041
+ },
1042
+ {
1043
+ "epoch": 0.99,
1044
+ "learning_rate": 0.00035304390214529015,
1045
+ "loss": 0.6434,
1046
+ "step": 346
1047
+ },
1048
+ {
1049
+ "epoch": 0.99,
1050
+ "learning_rate": 0.0003524478688358492,
1051
+ "loss": 0.6508,
1052
+ "step": 348
1053
+ },
1054
+ {
1055
+ "epoch": 1.0,
1056
+ "learning_rate": 0.0003518485864189699,
1057
+ "loss": 0.6753,
1058
+ "step": 350
1059
+ },
1060
+ {
1061
+ "epoch": 1.0,
1062
+ "learning_rate": 0.00035124606766710314,
1063
+ "loss": 0.6838,
1064
+ "step": 352
1065
+ },
1066
+ {
1067
+ "epoch": 1.01,
1068
+ "learning_rate": 0.00035064032542167587,
1069
+ "loss": 0.5915,
1070
+ "step": 354
1071
+ },
1072
+ {
1073
+ "epoch": 1.01,
1074
+ "learning_rate": 0.00035003137259281695,
1075
+ "loss": 0.6328,
1076
+ "step": 356
1077
+ },
1078
+ {
1079
+ "epoch": 1.02,
1080
+ "learning_rate": 0.0003494192221590822,
1081
+ "loss": 0.6565,
1082
+ "step": 358
1083
+ },
1084
+ {
1085
+ "epoch": 1.03,
1086
+ "learning_rate": 0.0003488038871671776,
1087
+ "loss": 0.656,
1088
+ "step": 360
1089
+ },
1090
+ {
1091
+ "epoch": 1.03,
1092
+ "learning_rate": 0.00034818538073168136,
1093
+ "loss": 0.6013,
1094
+ "step": 362
1095
+ },
1096
+ {
1097
+ "epoch": 1.04,
1098
+ "learning_rate": 0.0003475637160347645,
1099
+ "loss": 0.6271,
1100
+ "step": 364
1101
+ },
1102
+ {
1103
+ "epoch": 1.04,
1104
+ "learning_rate": 0.0003469389063259094,
1105
+ "loss": 0.6441,
1106
+ "step": 366
1107
+ },
1108
+ {
1109
+ "epoch": 1.05,
1110
+ "learning_rate": 0.0003463109649216282,
1111
+ "loss": 0.6319,
1112
+ "step": 368
1113
+ },
1114
+ {
1115
+ "epoch": 1.05,
1116
+ "learning_rate": 0.00034567990520517844,
1117
+ "loss": 0.6292,
1118
+ "step": 370
1119
+ },
1120
+ {
1121
+ "epoch": 1.06,
1122
+ "learning_rate": 0.0003450457406262779,
1123
+ "loss": 0.5924,
1124
+ "step": 372
1125
+ },
1126
+ {
1127
+ "epoch": 1.07,
1128
+ "learning_rate": 0.0003444084847008181,
1129
+ "loss": 0.635,
1130
+ "step": 374
1131
+ },
1132
+ {
1133
+ "epoch": 1.07,
1134
+ "learning_rate": 0.0003437681510105761,
1135
+ "loss": 0.6007,
1136
+ "step": 376
1137
+ },
1138
+ {
1139
+ "epoch": 1.08,
1140
+ "learning_rate": 0.000343124753202925,
1141
+ "loss": 0.6321,
1142
+ "step": 378
1143
+ },
1144
+ {
1145
+ "epoch": 1.08,
1146
+ "learning_rate": 0.00034247830499054336,
1147
+ "loss": 0.615,
1148
+ "step": 380
1149
+ },
1150
+ {
1151
+ "epoch": 1.09,
1152
+ "learning_rate": 0.00034182882015112244,
1153
+ "loss": 0.6328,
1154
+ "step": 382
1155
+ },
1156
+ {
1157
+ "epoch": 1.09,
1158
+ "learning_rate": 0.0003411763125270731,
1159
+ "loss": 0.5974,
1160
+ "step": 384
1161
+ },
1162
+ {
1163
+ "epoch": 1.1,
1164
+ "learning_rate": 0.0003405207960252305,
1165
+ "loss": 0.618,
1166
+ "step": 386
1167
+ },
1168
+ {
1169
+ "epoch": 1.11,
1170
+ "learning_rate": 0.00033986228461655766,
1171
+ "loss": 0.6112,
1172
+ "step": 388
1173
+ },
1174
+ {
1175
+ "epoch": 1.11,
1176
+ "learning_rate": 0.00033920079233584776,
1177
+ "loss": 0.6346,
1178
+ "step": 390
1179
+ },
1180
+ {
1181
+ "epoch": 1.12,
1182
+ "learning_rate": 0.00033853633328142515,
1183
+ "loss": 0.594,
1184
+ "step": 392
1185
+ },
1186
+ {
1187
+ "epoch": 1.12,
1188
+ "learning_rate": 0.00033786892161484456,
1189
+ "loss": 0.576,
1190
+ "step": 394
1191
+ },
1192
+ {
1193
+ "epoch": 1.13,
1194
+ "learning_rate": 0.0003371985715605897,
1195
+ "loss": 0.6082,
1196
+ "step": 396
1197
+ },
1198
+ {
1199
+ "epoch": 1.13,
1200
+ "learning_rate": 0.0003365252974057698,
1201
+ "loss": 0.6075,
1202
+ "step": 398
1203
+ },
1204
+ {
1205
+ "epoch": 1.14,
1206
+ "learning_rate": 0.00033584911349981495,
1207
+ "loss": 0.5957,
1208
+ "step": 400
1209
+ },
1210
+ {
1211
+ "epoch": 1.15,
1212
+ "learning_rate": 0.0003351700342541708,
1213
+ "loss": 0.5885,
1214
+ "step": 402
1215
+ },
1216
+ {
1217
+ "epoch": 1.15,
1218
+ "learning_rate": 0.0003344880741419911,
1219
+ "loss": 0.6087,
1220
+ "step": 404
1221
+ },
1222
+ {
1223
+ "epoch": 1.16,
1224
+ "learning_rate": 0.0003338032476978289,
1225
+ "loss": 0.5699,
1226
+ "step": 406
1227
+ },
1228
+ {
1229
+ "epoch": 1.16,
1230
+ "learning_rate": 0.00033311556951732754,
1231
+ "loss": 0.6346,
1232
+ "step": 408
1233
+ },
1234
+ {
1235
+ "epoch": 1.17,
1236
+ "learning_rate": 0.00033242505425690893,
1237
+ "loss": 0.5821,
1238
+ "step": 410
1239
+ },
1240
+ {
1241
+ "epoch": 1.17,
1242
+ "learning_rate": 0.0003317317166334615,
1243
+ "loss": 0.5686,
1244
+ "step": 412
1245
+ },
1246
+ {
1247
+ "epoch": 1.18,
1248
+ "learning_rate": 0.00033103557142402624,
1249
+ "loss": 0.5742,
1250
+ "step": 414
1251
+ },
1252
+ {
1253
+ "epoch": 1.19,
1254
+ "learning_rate": 0.0003303366334654822,
1255
+ "loss": 0.6053,
1256
+ "step": 416
1257
+ },
1258
+ {
1259
+ "epoch": 1.19,
1260
+ "learning_rate": 0.00032963491765423,
1261
+ "loss": 0.6033,
1262
+ "step": 418
1263
+ },
1264
+ {
1265
+ "epoch": 1.2,
1266
+ "learning_rate": 0.00032893043894587423,
1267
+ "loss": 0.5794,
1268
+ "step": 420
1269
+ },
1270
+ {
1271
+ "epoch": 1.2,
1272
+ "learning_rate": 0.000328223212354905,
1273
+ "loss": 0.6071,
1274
+ "step": 422
1275
+ },
1276
+ {
1277
+ "epoch": 1.21,
1278
+ "learning_rate": 0.0003275132529543778,
1279
+ "loss": 0.5719,
1280
+ "step": 424
1281
+ },
1282
+ {
1283
+ "epoch": 1.21,
1284
+ "learning_rate": 0.0003268005758755921,
1285
+ "loss": 0.5795,
1286
+ "step": 426
1287
+ },
1288
+ {
1289
+ "epoch": 1.22,
1290
+ "learning_rate": 0.00032608519630776913,
1291
+ "loss": 0.5689,
1292
+ "step": 428
1293
+ },
1294
+ {
1295
+ "epoch": 1.23,
1296
+ "learning_rate": 0.00032536712949772793,
1297
+ "loss": 0.5605,
1298
+ "step": 430
1299
+ },
1300
+ {
1301
+ "epoch": 1.23,
1302
+ "learning_rate": 0.0003246463907495605,
1303
+ "loss": 0.5789,
1304
+ "step": 432
1305
+ },
1306
+ {
1307
+ "epoch": 1.24,
1308
+ "learning_rate": 0.0003239229954243058,
1309
+ "loss": 0.6234,
1310
+ "step": 434
1311
+ },
1312
+ {
1313
+ "epoch": 1.24,
1314
+ "learning_rate": 0.0003231969589396218,
1315
+ "loss": 0.5473,
1316
+ "step": 436
1317
+ },
1318
+ {
1319
+ "epoch": 1.25,
1320
+ "learning_rate": 0.0003224682967694576,
1321
+ "loss": 0.5777,
1322
+ "step": 438
1323
+ },
1324
+ {
1325
+ "epoch": 1.25,
1326
+ "learning_rate": 0.0003217370244437231,
1327
+ "loss": 0.5794,
1328
+ "step": 440
1329
+ },
1330
+ {
1331
+ "epoch": 1.26,
1332
+ "learning_rate": 0.0003210031575479583,
1333
+ "loss": 0.5729,
1334
+ "step": 442
1335
+ },
1336
+ {
1337
+ "epoch": 1.26,
1338
+ "learning_rate": 0.000320266711723001,
1339
+ "loss": 0.5354,
1340
+ "step": 444
1341
+ },
1342
+ {
1343
+ "epoch": 1.27,
1344
+ "learning_rate": 0.00031952770266465346,
1345
+ "loss": 0.5975,
1346
+ "step": 446
1347
+ },
1348
+ {
1349
+ "epoch": 1.28,
1350
+ "learning_rate": 0.00031878614612334806,
1351
+ "loss": 0.5648,
1352
+ "step": 448
1353
+ },
1354
+ {
1355
+ "epoch": 1.28,
1356
+ "learning_rate": 0.00031804205790381113,
1357
+ "loss": 0.5529,
1358
+ "step": 450
1359
+ },
1360
+ {
1361
+ "epoch": 1.29,
1362
+ "learning_rate": 0.00031729545386472666,
1363
+ "loss": 0.5645,
1364
+ "step": 452
1365
+ },
1366
+ {
1367
+ "epoch": 1.29,
1368
+ "learning_rate": 0.00031654634991839804,
1369
+ "loss": 0.5496,
1370
+ "step": 454
1371
+ },
1372
+ {
1373
+ "epoch": 1.3,
1374
+ "learning_rate": 0.000315794762030409,
1375
+ "loss": 0.5262,
1376
+ "step": 456
1377
+ },
1378
+ {
1379
+ "epoch": 1.3,
1380
+ "learning_rate": 0.0003150407062192831,
1381
+ "loss": 0.5253,
1382
+ "step": 458
1383
+ },
1384
+ {
1385
+ "epoch": 1.31,
1386
+ "learning_rate": 0.0003142841985561427,
1387
+ "loss": 0.5519,
1388
+ "step": 460
1389
+ },
1390
+ {
1391
+ "epoch": 1.32,
1392
+ "learning_rate": 0.0003135252551643662,
1393
+ "loss": 0.5331,
1394
+ "step": 462
1395
+ },
1396
+ {
1397
+ "epoch": 1.32,
1398
+ "learning_rate": 0.0003127638922192445,
1399
+ "loss": 0.5431,
1400
+ "step": 464
1401
+ },
1402
+ {
1403
+ "epoch": 1.33,
1404
+ "learning_rate": 0.0003120001259476361,
1405
+ "loss": 0.5293,
1406
+ "step": 466
1407
+ },
1408
+ {
1409
+ "epoch": 1.33,
1410
+ "learning_rate": 0.00031123397262762157,
1411
+ "loss": 0.5695,
1412
+ "step": 468
1413
+ },
1414
+ {
1415
+ "epoch": 1.34,
1416
+ "learning_rate": 0.0003104654485881562,
1417
+ "loss": 0.5577,
1418
+ "step": 470
1419
+ },
1420
+ {
1421
+ "epoch": 1.34,
1422
+ "learning_rate": 0.0003096945702087223,
1423
+ "loss": 0.5517,
1424
+ "step": 472
1425
+ },
1426
+ {
1427
+ "epoch": 1.35,
1428
+ "learning_rate": 0.00030892135391898015,
1429
+ "loss": 0.5678,
1430
+ "step": 474
1431
+ },
1432
+ {
1433
+ "epoch": 1.36,
1434
+ "learning_rate": 0.0003081458161984175,
1435
+ "loss": 0.5548,
1436
+ "step": 476
1437
+ },
1438
+ {
1439
+ "epoch": 1.36,
1440
+ "learning_rate": 0.0003073679735759987,
1441
+ "loss": 0.5325,
1442
+ "step": 478
1443
+ },
1444
+ {
1445
+ "epoch": 1.37,
1446
+ "learning_rate": 0.000306587842629812,
1447
+ "loss": 0.5451,
1448
+ "step": 480
1449
+ },
1450
+ {
1451
+ "epoch": 1.37,
1452
+ "learning_rate": 0.00030580543998671685,
1453
+ "loss": 0.597,
1454
+ "step": 482
1455
+ },
1456
+ {
1457
+ "epoch": 1.38,
1458
+ "learning_rate": 0.00030502078232198885,
1459
+ "loss": 0.5763,
1460
+ "step": 484
1461
+ },
1462
+ {
1463
+ "epoch": 1.38,
1464
+ "learning_rate": 0.0003042338863589649,
1465
+ "loss": 0.5401,
1466
+ "step": 486
1467
+ },
1468
+ {
1469
+ "epoch": 1.39,
1470
+ "learning_rate": 0.00030344476886868643,
1471
+ "loss": 0.5923,
1472
+ "step": 488
1473
+ },
1474
+ {
1475
+ "epoch": 1.4,
1476
+ "learning_rate": 0.000302653446669542,
1477
+ "loss": 0.5462,
1478
+ "step": 490
1479
+ },
1480
+ {
1481
+ "epoch": 1.4,
1482
+ "learning_rate": 0.0003018599366269091,
1483
+ "loss": 0.5558,
1484
+ "step": 492
1485
+ },
1486
+ {
1487
+ "epoch": 1.41,
1488
+ "learning_rate": 0.0003010642556527945,
1489
+ "loss": 0.5426,
1490
+ "step": 494
1491
+ },
1492
+ {
1493
+ "epoch": 1.41,
1494
+ "learning_rate": 0.00030026642070547365,
1495
+ "loss": 0.5492,
1496
+ "step": 496
1497
+ },
1498
+ {
1499
+ "epoch": 1.42,
1500
+ "learning_rate": 0.0002994664487891297,
1501
+ "loss": 0.5124,
1502
+ "step": 498
1503
+ },
1504
+ {
1505
+ "epoch": 1.42,
1506
+ "learning_rate": 0.0002986643569534906,
1507
+ "loss": 0.5248,
1508
+ "step": 500
1509
+ },
1510
+ {
1511
+ "epoch": 1.43,
1512
+ "learning_rate": 0.0002978601622934661,
1513
+ "loss": 0.5365,
1514
+ "step": 502
1515
+ },
1516
+ {
1517
+ "epoch": 1.44,
1518
+ "learning_rate": 0.0002970538819487831,
1519
+ "loss": 0.5602,
1520
+ "step": 504
1521
+ },
1522
+ {
1523
+ "epoch": 1.44,
1524
+ "learning_rate": 0.00029624553310362065,
1525
+ "loss": 0.5596,
1526
+ "step": 506
1527
+ },
1528
+ {
1529
+ "epoch": 1.45,
1530
+ "learning_rate": 0.0002954351329862435,
1531
+ "loss": 0.5358,
1532
+ "step": 508
1533
+ },
1534
+ {
1535
+ "epoch": 1.45,
1536
+ "learning_rate": 0.000294622698868635,
1537
+ "loss": 0.5489,
1538
+ "step": 510
1539
+ },
1540
+ {
1541
+ "epoch": 1.46,
1542
+ "learning_rate": 0.00029380824806612895,
1543
+ "loss": 0.5386,
1544
+ "step": 512
1545
+ },
1546
+ {
1547
+ "epoch": 1.46,
1548
+ "learning_rate": 0.0002929917979370406,
1549
+ "loss": 0.47,
1550
+ "step": 514
1551
+ },
1552
+ {
1553
+ "epoch": 1.47,
1554
+ "learning_rate": 0.0002921733658822965,
1555
+ "loss": 0.533,
1556
+ "step": 516
1557
+ },
1558
+ {
1559
+ "epoch": 1.48,
1560
+ "learning_rate": 0.000291352969345064,
1561
+ "loss": 0.5462,
1562
+ "step": 518
1563
+ },
1564
+ {
1565
+ "epoch": 1.48,
1566
+ "learning_rate": 0.00029053062581037916,
1567
+ "loss": 0.5371,
1568
+ "step": 520
1569
+ },
1570
+ {
1571
+ "epoch": 1.49,
1572
+ "learning_rate": 0.00028970635280477436,
1573
+ "loss": 0.5198,
1574
+ "step": 522
1575
+ },
1576
+ {
1577
+ "epoch": 1.49,
1578
+ "learning_rate": 0.0002888801678959045,
1579
+ "loss": 0.537,
1580
+ "step": 524
1581
+ },
1582
+ {
1583
+ "epoch": 1.5,
1584
+ "learning_rate": 0.0002880520886921727,
1585
+ "loss": 0.5412,
1586
+ "step": 526
1587
+ },
1588
+ {
1589
+ "epoch": 1.5,
1590
+ "learning_rate": 0.000287222132842355,
1591
+ "loss": 0.5242,
1592
+ "step": 528
1593
+ },
1594
+ {
1595
+ "epoch": 1.51,
1596
+ "learning_rate": 0.0002863903180352244,
1597
+ "loss": 0.522,
1598
+ "step": 530
1599
+ },
1600
+ {
1601
+ "epoch": 1.52,
1602
+ "learning_rate": 0.0002855566619991733,
1603
+ "loss": 0.5238,
1604
+ "step": 532
1605
+ },
1606
+ {
1607
+ "epoch": 1.52,
1608
+ "learning_rate": 0.0002847211825018366,
1609
+ "loss": 0.5555,
1610
+ "step": 534
1611
+ },
1612
+ {
1613
+ "epoch": 1.53,
1614
+ "learning_rate": 0.00028388389734971197,
1615
+ "loss": 0.5228,
1616
+ "step": 536
1617
+ },
1618
+ {
1619
+ "epoch": 1.53,
1620
+ "learning_rate": 0.000283044824387781,
1621
+ "loss": 0.5319,
1622
+ "step": 538
1623
+ },
1624
+ {
1625
+ "epoch": 1.54,
1626
+ "learning_rate": 0.0002822039814991288,
1627
+ "loss": 0.5281,
1628
+ "step": 540
1629
+ },
1630
+ {
1631
+ "epoch": 1.54,
1632
+ "learning_rate": 0.00028136138660456265,
1633
+ "loss": 0.4977,
1634
+ "step": 542
1635
+ },
1636
+ {
1637
+ "epoch": 1.55,
1638
+ "learning_rate": 0.00028051705766223034,
1639
+ "loss": 0.4931,
1640
+ "step": 544
1641
+ },
1642
+ {
1643
+ "epoch": 1.56,
1644
+ "learning_rate": 0.00027967101266723694,
1645
+ "loss": 0.5028,
1646
+ "step": 546
1647
+ },
1648
+ {
1649
+ "epoch": 1.56,
1650
+ "learning_rate": 0.0002788232696512619,
1651
+ "loss": 0.5225,
1652
+ "step": 548
1653
+ },
1654
+ {
1655
+ "epoch": 1.57,
1656
+ "learning_rate": 0.00027797384668217434,
1657
+ "loss": 0.4941,
1658
+ "step": 550
1659
+ },
1660
+ {
1661
+ "epoch": 1.57,
1662
+ "learning_rate": 0.00027712276186364794,
1663
+ "loss": 0.5071,
1664
+ "step": 552
1665
+ },
1666
+ {
1667
+ "epoch": 1.58,
1668
+ "learning_rate": 0.00027627003333477543,
1669
+ "loss": 0.5207,
1670
+ "step": 554
1671
+ },
1672
+ {
1673
+ "epoch": 1.58,
1674
+ "learning_rate": 0.0002754156792696817,
1675
+ "loss": 0.5227,
1676
+ "step": 556
1677
+ },
1678
+ {
1679
+ "epoch": 1.59,
1680
+ "learning_rate": 0.0002745597178771365,
1681
+ "loss": 0.5112,
1682
+ "step": 558
1683
+ },
1684
+ {
1685
+ "epoch": 1.6,
1686
+ "learning_rate": 0.0002737021674001665,
1687
+ "loss": 0.5078,
1688
+ "step": 560
1689
+ },
1690
+ {
1691
+ "epoch": 1.6,
1692
+ "learning_rate": 0.0002728430461156662,
1693
+ "loss": 0.5128,
1694
+ "step": 562
1695
+ },
1696
+ {
1697
+ "epoch": 1.61,
1698
+ "learning_rate": 0.0002719823723340087,
1699
+ "loss": 0.513,
1700
+ "step": 564
1701
+ },
1702
+ {
1703
+ "epoch": 1.61,
1704
+ "learning_rate": 0.00027112016439865546,
1705
+ "loss": 0.5119,
1706
+ "step": 566
1707
+ },
1708
+ {
1709
+ "epoch": 1.62,
1710
+ "learning_rate": 0.00027025644068576503,
1711
+ "loss": 0.5165,
1712
+ "step": 568
1713
+ },
1714
+ {
1715
+ "epoch": 1.62,
1716
+ "learning_rate": 0.00026939121960380157,
1717
+ "loss": 0.5215,
1718
+ "step": 570
1719
+ },
1720
+ {
1721
+ "epoch": 1.63,
1722
+ "learning_rate": 0.0002685245195931427,
1723
+ "loss": 0.5127,
1724
+ "step": 572
1725
+ },
1726
+ {
1727
+ "epoch": 1.64,
1728
+ "learning_rate": 0.0002676563591256861,
1729
+ "loss": 0.5072,
1730
+ "step": 574
1731
+ },
1732
+ {
1733
+ "epoch": 1.64,
1734
+ "learning_rate": 0.0002667867567044562,
1735
+ "loss": 0.4924,
1736
+ "step": 576
1737
+ },
1738
+ {
1739
+ "epoch": 1.65,
1740
+ "learning_rate": 0.0002659157308632096,
1741
+ "loss": 0.4821,
1742
+ "step": 578
1743
+ },
1744
+ {
1745
+ "epoch": 1.65,
1746
+ "learning_rate": 0.00026504330016604015,
1747
+ "loss": 0.4862,
1748
+ "step": 580
1749
+ },
1750
+ {
1751
+ "epoch": 1.66,
1752
+ "learning_rate": 0.0002641694832069832,
1753
+ "loss": 0.4877,
1754
+ "step": 582
1755
+ },
1756
+ {
1757
+ "epoch": 1.66,
1758
+ "learning_rate": 0.00026329429860961947,
1759
+ "loss": 0.4702,
1760
+ "step": 584
1761
+ },
1762
+ {
1763
+ "epoch": 1.67,
1764
+ "learning_rate": 0.0002624177650266779,
1765
+ "loss": 0.5067,
1766
+ "step": 586
1767
+ },
1768
+ {
1769
+ "epoch": 1.68,
1770
+ "learning_rate": 0.00026153990113963826,
1771
+ "loss": 0.4797,
1772
+ "step": 588
1773
+ },
1774
+ {
1775
+ "epoch": 1.68,
1776
+ "learning_rate": 0.0002606607256583331,
1777
+ "loss": 0.5172,
1778
+ "step": 590
1779
+ },
1780
+ {
1781
+ "epoch": 1.69,
1782
+ "learning_rate": 0.00025978025732054875,
1783
+ "loss": 0.5014,
1784
+ "step": 592
1785
+ },
1786
+ {
1787
+ "epoch": 1.69,
1788
+ "learning_rate": 0.0002588985148916261,
1789
+ "loss": 0.4776,
1790
+ "step": 594
1791
+ },
1792
+ {
1793
+ "epoch": 1.7,
1794
+ "learning_rate": 0.0002580155171640606,
1795
+ "loss": 0.4959,
1796
+ "step": 596
1797
+ },
1798
+ {
1799
+ "epoch": 1.7,
1800
+ "learning_rate": 0.0002571312829571019,
1801
+ "loss": 0.457,
1802
+ "step": 598
1803
+ },
1804
+ {
1805
+ "epoch": 1.71,
1806
+ "learning_rate": 0.0002562458311163523,
1807
+ "loss": 0.5027,
1808
+ "step": 600
1809
+ },
1810
+ {
1811
+ "epoch": 1.72,
1812
+ "learning_rate": 0.00025535918051336574,
1813
+ "loss": 0.4833,
1814
+ "step": 602
1815
+ },
1816
+ {
1817
+ "epoch": 1.72,
1818
+ "learning_rate": 0.00025447135004524516,
1819
+ "loss": 0.4837,
1820
+ "step": 604
1821
+ },
1822
+ {
1823
+ "epoch": 1.73,
1824
+ "learning_rate": 0.00025358235863423984,
1825
+ "loss": 0.4873,
1826
+ "step": 606
1827
+ },
1828
+ {
1829
+ "epoch": 1.73,
1830
+ "learning_rate": 0.00025269222522734215,
1831
+ "loss": 0.478,
1832
+ "step": 608
1833
+ },
1834
+ {
1835
+ "epoch": 1.74,
1836
+ "learning_rate": 0.00025180096879588373,
1837
+ "loss": 0.509,
1838
+ "step": 610
1839
+ },
1840
+ {
1841
+ "epoch": 1.74,
1842
+ "learning_rate": 0.000250908608335131,
1843
+ "loss": 0.5005,
1844
+ "step": 612
1845
+ },
1846
+ {
1847
+ "epoch": 1.75,
1848
+ "learning_rate": 0.0002500151628638807,
1849
+ "loss": 0.4677,
1850
+ "step": 614
1851
+ },
1852
+ {
1853
+ "epoch": 1.75,
1854
+ "learning_rate": 0.0002491206514240541,
1855
+ "loss": 0.482,
1856
+ "step": 616
1857
+ },
1858
+ {
1859
+ "epoch": 1.76,
1860
+ "learning_rate": 0.0002482250930802915,
1861
+ "loss": 0.4808,
1862
+ "step": 618
1863
+ },
1864
+ {
1865
+ "epoch": 1.77,
1866
+ "learning_rate": 0.00024732850691954557,
1867
+ "loss": 0.4488,
1868
+ "step": 620
1869
+ },
1870
+ {
1871
+ "epoch": 1.77,
1872
+ "learning_rate": 0.00024643091205067496,
1873
+ "loss": 0.4671,
1874
+ "step": 622
1875
+ },
1876
+ {
1877
+ "epoch": 1.78,
1878
+ "learning_rate": 0.0002455323276040367,
1879
+ "loss": 0.4483,
1880
+ "step": 624
1881
+ },
1882
+ {
1883
+ "epoch": 1.78,
1884
+ "learning_rate": 0.0002446327727310787,
1885
+ "loss": 0.49,
1886
+ "step": 626
1887
+ },
1888
+ {
1889
+ "epoch": 1.79,
1890
+ "learning_rate": 0.00024373226660393134,
1891
+ "loss": 0.4837,
1892
+ "step": 628
1893
+ },
1894
+ {
1895
+ "epoch": 1.79,
1896
+ "learning_rate": 0.0002428308284149992,
1897
+ "loss": 0.4797,
1898
+ "step": 630
1899
+ },
1900
+ {
1901
+ "epoch": 1.8,
1902
+ "learning_rate": 0.00024192847737655152,
1903
+ "loss": 0.4832,
1904
+ "step": 632
1905
+ },
1906
+ {
1907
+ "epoch": 1.81,
1908
+ "learning_rate": 0.0002410252327203133,
1909
+ "loss": 0.4833,
1910
+ "step": 634
1911
+ },
1912
+ {
1913
+ "epoch": 1.81,
1914
+ "learning_rate": 0.00024012111369705488,
1915
+ "loss": 0.445,
1916
+ "step": 636
1917
+ },
1918
+ {
1919
+ "epoch": 1.82,
1920
+ "learning_rate": 0.00023921613957618215,
1921
+ "loss": 0.4898,
1922
+ "step": 638
1923
+ },
1924
+ {
1925
+ "epoch": 1.82,
1926
+ "learning_rate": 0.00023831032964532543,
1927
+ "loss": 0.4769,
1928
+ "step": 640
1929
+ },
1930
+ {
1931
+ "epoch": 1.83,
1932
+ "learning_rate": 0.00023740370320992865,
1933
+ "loss": 0.4595,
1934
+ "step": 642
1935
+ },
1936
+ {
1937
+ "epoch": 1.83,
1938
+ "learning_rate": 0.00023649627959283792,
1939
+ "loss": 0.4658,
1940
+ "step": 644
1941
+ },
1942
+ {
1943
+ "epoch": 1.84,
1944
+ "learning_rate": 0.00023558807813388937,
1945
+ "loss": 0.4783,
1946
+ "step": 646
1947
+ },
1948
+ {
1949
+ "epoch": 1.85,
1950
+ "learning_rate": 0.00023467911818949753,
1951
+ "loss": 0.4618,
1952
+ "step": 648
1953
+ },
1954
+ {
1955
+ "epoch": 1.85,
1956
+ "learning_rate": 0.00023376941913224206,
1957
+ "loss": 0.4526,
1958
+ "step": 650
1959
+ },
1960
+ {
1961
+ "epoch": 1.86,
1962
+ "learning_rate": 0.0002328590003504557,
1963
+ "loss": 0.4492,
1964
+ "step": 652
1965
+ },
1966
+ {
1967
+ "epoch": 1.86,
1968
+ "learning_rate": 0.00023194788124781027,
1969
+ "loss": 0.4576,
1970
+ "step": 654
1971
+ },
1972
+ {
1973
+ "epoch": 1.87,
1974
+ "learning_rate": 0.00023103608124290363,
1975
+ "loss": 0.4899,
1976
+ "step": 656
1977
+ },
1978
+ {
1979
+ "epoch": 1.87,
1980
+ "learning_rate": 0.00023012361976884553,
1981
+ "loss": 0.4477,
1982
+ "step": 658
1983
+ },
1984
+ {
1985
+ "epoch": 1.88,
1986
+ "learning_rate": 0.00022921051627284382,
1987
+ "loss": 0.4781,
1988
+ "step": 660
1989
+ },
1990
+ {
1991
+ "epoch": 1.89,
1992
+ "learning_rate": 0.00022829679021578935,
1993
+ "loss": 0.4776,
1994
+ "step": 662
1995
+ },
1996
+ {
1997
+ "epoch": 1.89,
1998
+ "learning_rate": 0.0002273824610718418,
1999
+ "loss": 0.4792,
2000
+ "step": 664
2001
+ },
2002
+ {
2003
+ "epoch": 1.9,
2004
+ "learning_rate": 0.00022646754832801426,
2005
+ "loss": 0.4764,
2006
+ "step": 666
2007
+ },
2008
+ {
2009
+ "epoch": 1.9,
2010
+ "learning_rate": 0.00022555207148375813,
2011
+ "loss": 0.4788,
2012
+ "step": 668
2013
+ },
2014
+ {
2015
+ "epoch": 1.91,
2016
+ "learning_rate": 0.0002246360500505473,
2017
+ "loss": 0.4535,
2018
+ "step": 670
2019
+ },
2020
+ {
2021
+ "epoch": 1.91,
2022
+ "learning_rate": 0.00022371950355146262,
2023
+ "loss": 0.47,
2024
+ "step": 672
2025
+ },
2026
+ {
2027
+ "epoch": 1.92,
2028
+ "learning_rate": 0.00022280245152077538,
2029
+ "loss": 0.4924,
2030
+ "step": 674
2031
+ },
2032
+ {
2033
+ "epoch": 1.93,
2034
+ "learning_rate": 0.00022188491350353152,
2035
+ "loss": 0.4373,
2036
+ "step": 676
2037
+ },
2038
+ {
2039
+ "epoch": 1.93,
2040
+ "learning_rate": 0.00022096690905513446,
2041
+ "loss": 0.4597,
2042
+ "step": 678
2043
+ },
2044
+ {
2045
+ "epoch": 1.94,
2046
+ "learning_rate": 0.00022004845774092882,
2047
+ "loss": 0.4471,
2048
+ "step": 680
2049
+ },
2050
+ {
2051
+ "epoch": 1.94,
2052
+ "learning_rate": 0.0002191295791357832,
2053
+ "loss": 0.4499,
2054
+ "step": 682
2055
+ },
2056
+ {
2057
+ "epoch": 1.95,
2058
+ "learning_rate": 0.00021821029282367298,
2059
+ "loss": 0.4552,
2060
+ "step": 684
2061
+ },
2062
+ {
2063
+ "epoch": 1.95,
2064
+ "learning_rate": 0.0002172906183972629,
2065
+ "loss": 0.4394,
2066
+ "step": 686
2067
+ },
2068
+ {
2069
+ "epoch": 1.96,
2070
+ "learning_rate": 0.0002163705754574897,
2071
+ "loss": 0.4647,
2072
+ "step": 688
2073
+ },
2074
+ {
2075
+ "epoch": 1.97,
2076
+ "learning_rate": 0.00021545018361314404,
2077
+ "loss": 0.4505,
2078
+ "step": 690
2079
+ },
2080
+ {
2081
+ "epoch": 1.97,
2082
+ "learning_rate": 0.00021452946248045284,
2083
+ "loss": 0.4663,
2084
+ "step": 692
2085
+ },
2086
+ {
2087
+ "epoch": 1.98,
2088
+ "learning_rate": 0.00021360843168266125,
2089
+ "loss": 0.4594,
2090
+ "step": 694
2091
+ },
2092
+ {
2093
+ "epoch": 1.98,
2094
+ "learning_rate": 0.00021268711084961397,
2095
+ "loss": 0.4636,
2096
+ "step": 696
2097
+ },
2098
+ {
2099
+ "epoch": 1.99,
2100
+ "learning_rate": 0.00021176551961733746,
2101
+ "loss": 0.4335,
2102
+ "step": 698
2103
+ },
2104
+ {
2105
+ "epoch": 1.99,
2106
+ "learning_rate": 0.00021084367762762106,
2107
+ "loss": 0.4706,
2108
+ "step": 700
2109
+ },
2110
+ {
2111
+ "epoch": 2.0,
2112
+ "learning_rate": 0.0002099216045275985,
2113
+ "loss": 0.4764,
2114
+ "step": 702
2115
+ },
2116
+ {
2117
+ "epoch": 2.01,
2118
+ "learning_rate": 0.0002089993199693291,
2119
+ "loss": 0.4501,
2120
+ "step": 704
2121
+ },
2122
+ {
2123
+ "epoch": 2.01,
2124
+ "learning_rate": 0.0002080768436093791,
2125
+ "loss": 0.4456,
2126
+ "step": 706
2127
+ },
2128
+ {
2129
+ "epoch": 2.02,
2130
+ "learning_rate": 0.00020715419510840234,
2131
+ "loss": 0.4476,
2132
+ "step": 708
2133
+ },
2134
+ {
2135
+ "epoch": 2.02,
2136
+ "learning_rate": 0.0002062313941307219,
2137
+ "loss": 0.4237,
2138
+ "step": 710
2139
+ },
2140
+ {
2141
+ "epoch": 2.03,
2142
+ "learning_rate": 0.00020530846034391012,
2143
+ "loss": 0.4244,
2144
+ "step": 712
2145
+ },
2146
+ {
2147
+ "epoch": 2.03,
2148
+ "learning_rate": 0.00020438541341837025,
2149
+ "loss": 0.4559,
2150
+ "step": 714
2151
+ },
2152
+ {
2153
+ "epoch": 2.04,
2154
+ "learning_rate": 0.00020346227302691657,
2155
+ "loss": 0.4291,
2156
+ "step": 716
2157
+ },
2158
+ {
2159
+ "epoch": 2.05,
2160
+ "learning_rate": 0.00020253905884435564,
2161
+ "loss": 0.4382,
2162
+ "step": 718
2163
+ },
2164
+ {
2165
+ "epoch": 2.05,
2166
+ "learning_rate": 0.00020161579054706654,
2167
+ "loss": 0.4504,
2168
+ "step": 720
2169
+ },
2170
+ {
2171
+ "epoch": 2.06,
2172
+ "learning_rate": 0.00020069248781258184,
2173
+ "loss": 0.4425,
2174
+ "step": 722
2175
+ },
2176
+ {
2177
+ "epoch": 2.06,
2178
+ "learning_rate": 0.00019976917031916784,
2179
+ "loss": 0.4178,
2180
+ "step": 724
2181
+ },
2182
+ {
2183
+ "epoch": 2.07,
2184
+ "learning_rate": 0.00019884585774540575,
2185
+ "loss": 0.4357,
2186
+ "step": 726
2187
+ },
2188
+ {
2189
+ "epoch": 2.07,
2190
+ "learning_rate": 0.00019792256976977154,
2191
+ "loss": 0.4217,
2192
+ "step": 728
2193
+ },
2194
+ {
2195
+ "epoch": 2.08,
2196
+ "learning_rate": 0.00019699932607021727,
2197
+ "loss": 0.4504,
2198
+ "step": 730
2199
+ },
2200
+ {
2201
+ "epoch": 2.09,
2202
+ "learning_rate": 0.00019607614632375102,
2203
+ "loss": 0.4306,
2204
+ "step": 732
2205
+ },
2206
+ {
2207
+ "epoch": 2.09,
2208
+ "learning_rate": 0.00019515305020601813,
2209
+ "loss": 0.4165,
2210
+ "step": 734
2211
+ },
2212
+ {
2213
+ "epoch": 2.1,
2214
+ "learning_rate": 0.00019423005739088137,
2215
+ "loss": 0.4169,
2216
+ "step": 736
2217
+ },
2218
+ {
2219
+ "epoch": 2.1,
2220
+ "learning_rate": 0.00019330718755000196,
2221
+ "loss": 0.4376,
2222
+ "step": 738
2223
+ },
2224
+ {
2225
+ "epoch": 2.11,
2226
+ "learning_rate": 0.0001923844603524201,
2227
+ "loss": 0.431,
2228
+ "step": 740
2229
+ },
2230
+ {
2231
+ "epoch": 2.11,
2232
+ "learning_rate": 0.0001914618954641359,
2233
+ "loss": 0.4529,
2234
+ "step": 742
2235
+ },
2236
+ {
2237
+ "epoch": 2.12,
2238
+ "learning_rate": 0.0001905395125476901,
2239
+ "loss": 0.4637,
2240
+ "step": 744
2241
+ },
2242
+ {
2243
+ "epoch": 2.13,
2244
+ "learning_rate": 0.00018961733126174522,
2245
+ "loss": 0.4561,
2246
+ "step": 746
2247
+ },
2248
+ {
2249
+ "epoch": 2.13,
2250
+ "learning_rate": 0.00018869537126066632,
2251
+ "loss": 0.4135,
2252
+ "step": 748
2253
+ },
2254
+ {
2255
+ "epoch": 2.14,
2256
+ "learning_rate": 0.0001877736521941023,
2257
+ "loss": 0.437,
2258
+ "step": 750
2259
+ },
2260
+ {
2261
+ "epoch": 2.14,
2262
+ "learning_rate": 0.00018685219370656714,
2263
+ "loss": 0.4294,
2264
+ "step": 752
2265
+ },
2266
+ {
2267
+ "epoch": 2.15,
2268
+ "learning_rate": 0.0001859310154370209,
2269
+ "loss": 0.4162,
2270
+ "step": 754
2271
+ },
2272
+ {
2273
+ "epoch": 2.15,
2274
+ "learning_rate": 0.0001850101370184516,
2275
+ "loss": 0.4272,
2276
+ "step": 756
2277
+ },
2278
+ {
2279
+ "epoch": 2.16,
2280
+ "learning_rate": 0.0001840895780774564,
2281
+ "loss": 0.4711,
2282
+ "step": 758
2283
+ },
2284
+ {
2285
+ "epoch": 2.17,
2286
+ "learning_rate": 0.0001831693582338236,
2287
+ "loss": 0.4131,
2288
+ "step": 760
2289
+ },
2290
+ {
2291
+ "epoch": 2.17,
2292
+ "learning_rate": 0.00018224949710011416,
2293
+ "loss": 0.4377,
2294
+ "step": 762
2295
+ },
2296
+ {
2297
+ "epoch": 2.18,
2298
+ "learning_rate": 0.00018133001428124416,
2299
+ "loss": 0.4124,
2300
+ "step": 764
2301
+ },
2302
+ {
2303
+ "epoch": 2.18,
2304
+ "learning_rate": 0.0001804109293740664,
2305
+ "loss": 0.4577,
2306
+ "step": 766
2307
+ },
2308
+ {
2309
+ "epoch": 2.19,
2310
+ "learning_rate": 0.00017949226196695327,
2311
+ "loss": 0.4371,
2312
+ "step": 768
2313
+ },
2314
+ {
2315
+ "epoch": 2.19,
2316
+ "learning_rate": 0.00017857403163937884,
2317
+ "loss": 0.4465,
2318
+ "step": 770
2319
+ },
2320
+ {
2321
+ "epoch": 2.2,
2322
+ "learning_rate": 0.00017765625796150184,
2323
+ "loss": 0.4421,
2324
+ "step": 772
2325
+ },
2326
+ {
2327
+ "epoch": 2.21,
2328
+ "learning_rate": 0.00017673896049374846,
2329
+ "loss": 0.4143,
2330
+ "step": 774
2331
+ },
2332
+ {
2333
+ "epoch": 2.21,
2334
+ "learning_rate": 0.00017582215878639542,
2335
+ "loss": 0.4397,
2336
+ "step": 776
2337
+ },
2338
+ {
2339
+ "epoch": 2.22,
2340
+ "learning_rate": 0.00017490587237915342,
2341
+ "loss": 0.3917,
2342
+ "step": 778
2343
+ },
2344
+ {
2345
+ "epoch": 2.22,
2346
+ "learning_rate": 0.0001739901208007505,
2347
+ "loss": 0.4459,
2348
+ "step": 780
2349
+ },
2350
+ {
2351
+ "epoch": 2.23,
2352
+ "learning_rate": 0.0001730749235685161,
2353
+ "loss": 0.4342,
2354
+ "step": 782
2355
+ },
2356
+ {
2357
+ "epoch": 2.23,
2358
+ "learning_rate": 0.00017216030018796471,
2359
+ "loss": 0.4459,
2360
+ "step": 784
2361
+ },
2362
+ {
2363
+ "epoch": 2.24,
2364
+ "learning_rate": 0.0001712462701523807,
2365
+ "loss": 0.3873,
2366
+ "step": 786
2367
+ },
2368
+ {
2369
+ "epoch": 2.25,
2370
+ "learning_rate": 0.0001703328529424022,
2371
+ "loss": 0.435,
2372
+ "step": 788
2373
+ },
2374
+ {
2375
+ "epoch": 2.25,
2376
+ "learning_rate": 0.00016942006802560643,
2377
+ "loss": 0.4163,
2378
+ "step": 790
2379
+ },
2380
+ {
2381
+ "epoch": 2.26,
2382
+ "learning_rate": 0.00016850793485609453,
2383
+ "loss": 0.4514,
2384
+ "step": 792
2385
+ },
2386
+ {
2387
+ "epoch": 2.26,
2388
+ "learning_rate": 0.00016759647287407708,
2389
+ "loss": 0.4189,
2390
+ "step": 794
2391
+ },
2392
+ {
2393
+ "epoch": 2.27,
2394
+ "learning_rate": 0.00016668570150545962,
2395
+ "loss": 0.4232,
2396
+ "step": 796
2397
+ },
2398
+ {
2399
+ "epoch": 2.27,
2400
+ "learning_rate": 0.00016577564016142875,
2401
+ "loss": 0.3958,
2402
+ "step": 798
2403
+ },
2404
+ {
2405
+ "epoch": 2.28,
2406
+ "learning_rate": 0.00016486630823803837,
2407
+ "loss": 0.4073,
2408
+ "step": 800
2409
+ },
2410
+ {
2411
+ "epoch": 2.28,
2412
+ "learning_rate": 0.0001639577251157964,
2413
+ "loss": 0.4277,
2414
+ "step": 802
2415
+ },
2416
+ {
2417
+ "epoch": 2.29,
2418
+ "learning_rate": 0.0001630499101592514,
2419
+ "loss": 0.415,
2420
+ "step": 804
2421
+ },
2422
+ {
2423
+ "epoch": 2.3,
2424
+ "learning_rate": 0.00016214288271658035,
2425
+ "loss": 0.4244,
2426
+ "step": 806
2427
+ },
2428
+ {
2429
+ "epoch": 2.3,
2430
+ "learning_rate": 0.00016123666211917582,
2431
+ "loss": 0.4209,
2432
+ "step": 808
2433
+ },
2434
+ {
2435
+ "epoch": 2.31,
2436
+ "learning_rate": 0.00016033126768123435,
2437
+ "loss": 0.4036,
2438
+ "step": 810
2439
+ },
2440
+ {
2441
+ "epoch": 2.31,
2442
+ "learning_rate": 0.00015942671869934437,
2443
+ "loss": 0.4185,
2444
+ "step": 812
2445
+ },
2446
+ {
2447
+ "epoch": 2.32,
2448
+ "learning_rate": 0.00015852303445207543,
2449
+ "loss": 0.4152,
2450
+ "step": 814
2451
+ },
2452
+ {
2453
+ "epoch": 2.32,
2454
+ "learning_rate": 0.0001576202341995668,
2455
+ "loss": 0.409,
2456
+ "step": 816
2457
+ },
2458
+ {
2459
+ "epoch": 2.33,
2460
+ "learning_rate": 0.00015671833718311766,
2461
+ "loss": 0.4182,
2462
+ "step": 818
2463
+ },
2464
+ {
2465
+ "epoch": 2.34,
2466
+ "learning_rate": 0.00015581736262477617,
2467
+ "loss": 0.4158,
2468
+ "step": 820
2469
+ },
2470
+ {
2471
+ "epoch": 2.34,
2472
+ "learning_rate": 0.0001549173297269305,
2473
+ "loss": 0.4292,
2474
+ "step": 822
2475
+ },
2476
+ {
2477
+ "epoch": 2.35,
2478
+ "learning_rate": 0.00015401825767189916,
2479
+ "loss": 0.4116,
2480
+ "step": 824
2481
+ },
2482
+ {
2483
+ "epoch": 2.35,
2484
+ "learning_rate": 0.0001531201656215224,
2485
+ "loss": 0.4074,
2486
+ "step": 826
2487
+ },
2488
+ {
2489
+ "epoch": 2.36,
2490
+ "learning_rate": 0.0001522230727167536,
2491
+ "loss": 0.419,
2492
+ "step": 828
2493
+ },
2494
+ {
2495
+ "epoch": 2.36,
2496
+ "learning_rate": 0.0001513269980772516,
2497
+ "loss": 0.4095,
2498
+ "step": 830
2499
+ },
2500
+ {
2501
+ "epoch": 2.37,
2502
+ "learning_rate": 0.00015043196080097285,
2503
+ "loss": 0.4093,
2504
+ "step": 832
2505
+ },
2506
+ {
2507
+ "epoch": 2.38,
2508
+ "learning_rate": 0.00014953797996376483,
2509
+ "loss": 0.4015,
2510
+ "step": 834
2511
+ },
2512
+ {
2513
+ "epoch": 2.38,
2514
+ "learning_rate": 0.00014864507461895894,
2515
+ "loss": 0.4379,
2516
+ "step": 836
2517
+ },
2518
+ {
2519
+ "epoch": 2.39,
2520
+ "learning_rate": 0.00014775326379696494,
2521
+ "loss": 0.4175,
2522
+ "step": 838
2523
+ },
2524
+ {
2525
+ "epoch": 2.39,
2526
+ "learning_rate": 0.00014686256650486498,
2527
+ "loss": 0.4092,
2528
+ "step": 840
2529
+ },
2530
+ {
2531
+ "epoch": 2.4,
2532
+ "learning_rate": 0.00014597300172600877,
2533
+ "loss": 0.4226,
2534
+ "step": 842
2535
+ },
2536
+ {
2537
+ "epoch": 2.4,
2538
+ "learning_rate": 0.00014508458841960877,
2539
+ "loss": 0.4073,
2540
+ "step": 844
2541
+ },
2542
+ {
2543
+ "epoch": 2.41,
2544
+ "learning_rate": 0.00014419734552033626,
2545
+ "loss": 0.4199,
2546
+ "step": 846
2547
+ },
2548
+ {
2549
+ "epoch": 2.42,
2550
+ "learning_rate": 0.0001433112919379177,
2551
+ "loss": 0.4296,
2552
+ "step": 848
2553
+ },
2554
+ {
2555
+ "epoch": 2.42,
2556
+ "learning_rate": 0.00014242644655673177,
2557
+ "loss": 0.417,
2558
+ "step": 850
2559
+ },
2560
+ {
2561
+ "epoch": 2.43,
2562
+ "learning_rate": 0.00014154282823540702,
2563
+ "loss": 0.3994,
2564
+ "step": 852
2565
+ },
2566
+ {
2567
+ "epoch": 2.43,
2568
+ "learning_rate": 0.0001406604558064195,
2569
+ "loss": 0.4364,
2570
+ "step": 854
2571
+ },
2572
+ {
2573
+ "epoch": 2.44,
2574
+ "learning_rate": 0.00013977934807569186,
2575
+ "loss": 0.4115,
2576
+ "step": 856
2577
+ },
2578
+ {
2579
+ "epoch": 2.44,
2580
+ "learning_rate": 0.0001388995238221923,
2581
+ "loss": 0.4242,
2582
+ "step": 858
2583
+ },
2584
+ {
2585
+ "epoch": 2.45,
2586
+ "learning_rate": 0.00013802100179753442,
2587
+ "loss": 0.4149,
2588
+ "step": 860
2589
+ },
2590
+ {
2591
+ "epoch": 2.46,
2592
+ "learning_rate": 0.00013714380072557745,
2593
+ "loss": 0.387,
2594
+ "step": 862
2595
+ },
2596
+ {
2597
+ "epoch": 2.46,
2598
+ "learning_rate": 0.0001362679393020274,
2599
+ "loss": 0.416,
2600
+ "step": 864
2601
+ },
2602
+ {
2603
+ "epoch": 2.47,
2604
+ "learning_rate": 0.00013539343619403836,
2605
+ "loss": 0.4054,
2606
+ "step": 866
2607
+ },
2608
+ {
2609
+ "epoch": 2.47,
2610
+ "learning_rate": 0.0001345203100398149,
2611
+ "loss": 0.4125,
2612
+ "step": 868
2613
+ },
2614
+ {
2615
+ "epoch": 2.48,
2616
+ "learning_rate": 0.0001336485794482145,
2617
+ "loss": 0.3986,
2618
+ "step": 870
2619
+ },
2620
+ {
2621
+ "epoch": 2.48,
2622
+ "learning_rate": 0.00013277826299835138,
2623
+ "loss": 0.3871,
2624
+ "step": 872
2625
+ },
2626
+ {
2627
+ "epoch": 2.49,
2628
+ "learning_rate": 0.00013190937923920006,
2629
+ "loss": 0.4413,
2630
+ "step": 874
2631
+ },
2632
+ {
2633
+ "epoch": 2.5,
2634
+ "learning_rate": 0.00013104194668920046,
2635
+ "loss": 0.3995,
2636
+ "step": 876
2637
+ },
2638
+ {
2639
+ "epoch": 2.5,
2640
+ "learning_rate": 0.00013017598383586285,
2641
+ "loss": 0.3851,
2642
+ "step": 878
2643
+ },
2644
+ {
2645
+ "epoch": 2.51,
2646
+ "learning_rate": 0.00012931150913537411,
2647
+ "loss": 0.4124,
2648
+ "step": 880
2649
+ },
2650
+ {
2651
+ "epoch": 2.51,
2652
+ "learning_rate": 0.00012844854101220414,
2653
+ "loss": 0.3702,
2654
+ "step": 882
2655
+ },
2656
+ {
2657
+ "epoch": 2.52,
2658
+ "learning_rate": 0.0001275870978587133,
2659
+ "loss": 0.3955,
2660
+ "step": 884
2661
+ },
2662
+ {
2663
+ "epoch": 2.52,
2664
+ "learning_rate": 0.00012672719803476066,
2665
+ "loss": 0.4327,
2666
+ "step": 886
2667
+ },
2668
+ {
2669
+ "epoch": 2.53,
2670
+ "learning_rate": 0.0001258688598673121,
2671
+ "loss": 0.4021,
2672
+ "step": 888
2673
+ },
2674
+ {
2675
+ "epoch": 2.54,
2676
+ "learning_rate": 0.00012501210165005028,
2677
+ "loss": 0.3967,
2678
+ "step": 890
2679
+ },
2680
+ {
2681
+ "epoch": 2.54,
2682
+ "learning_rate": 0.00012415694164298442,
2683
+ "loss": 0.4014,
2684
+ "step": 892
2685
+ },
2686
+ {
2687
+ "epoch": 2.55,
2688
+ "learning_rate": 0.00012330339807206133,
2689
+ "loss": 0.3862,
2690
+ "step": 894
2691
+ },
2692
+ {
2693
+ "epoch": 2.55,
2694
+ "learning_rate": 0.00012245148912877672,
2695
+ "loss": 0.4086,
2696
+ "step": 896
2697
+ },
2698
+ {
2699
+ "epoch": 2.56,
2700
+ "learning_rate": 0.00012160123296978782,
2701
+ "loss": 0.4086,
2702
+ "step": 898
2703
+ },
2704
+ {
2705
+ "epoch": 2.56,
2706
+ "learning_rate": 0.00012075264771652603,
2707
+ "loss": 0.4196,
2708
+ "step": 900
2709
+ },
2710
+ {
2711
+ "epoch": 2.57,
2712
+ "learning_rate": 0.00011990575145481107,
2713
+ "loss": 0.3967,
2714
+ "step": 902
2715
+ },
2716
+ {
2717
+ "epoch": 2.58,
2718
+ "learning_rate": 0.00011906056223446514,
2719
+ "loss": 0.4201,
2720
+ "step": 904
2721
+ },
2722
+ {
2723
+ "epoch": 2.58,
2724
+ "learning_rate": 0.00011821709806892863,
2725
+ "loss": 0.3791,
2726
+ "step": 906
2727
+ },
2728
+ {
2729
+ "epoch": 2.59,
2730
+ "learning_rate": 0.00011737537693487582,
2731
+ "loss": 0.3999,
2732
+ "step": 908
2733
+ },
2734
+ {
2735
+ "epoch": 2.59,
2736
+ "learning_rate": 0.00011653541677183199,
2737
+ "loss": 0.3837,
2738
+ "step": 910
2739
+ },
2740
+ {
2741
+ "epoch": 2.6,
2742
+ "learning_rate": 0.00011569723548179114,
2743
+ "loss": 0.3836,
2744
+ "step": 912
2745
+ },
2746
+ {
2747
+ "epoch": 2.6,
2748
+ "learning_rate": 0.00011486085092883409,
2749
+ "loss": 0.3927,
2750
+ "step": 914
2751
+ },
2752
+ {
2753
+ "epoch": 2.61,
2754
+ "learning_rate": 0.00011402628093874805,
2755
+ "loss": 0.4033,
2756
+ "step": 916
2757
+ },
2758
+ {
2759
+ "epoch": 2.62,
2760
+ "learning_rate": 0.00011319354329864653,
2761
+ "loss": 0.4194,
2762
+ "step": 918
2763
+ },
2764
+ {
2765
+ "epoch": 2.62,
2766
+ "learning_rate": 0.0001123626557565907,
2767
+ "loss": 0.42,
2768
+ "step": 920
2769
+ },
2770
+ {
2771
+ "epoch": 2.63,
2772
+ "learning_rate": 0.00011153363602121034,
2773
+ "loss": 0.3982,
2774
+ "step": 922
2775
+ },
2776
+ {
2777
+ "epoch": 2.63,
2778
+ "learning_rate": 0.00011070650176132702,
2779
+ "loss": 0.3783,
2780
+ "step": 924
2781
+ },
2782
+ {
2783
+ "epoch": 2.64,
2784
+ "learning_rate": 0.0001098812706055772,
2785
+ "loss": 0.3788,
2786
+ "step": 926
2787
+ },
2788
+ {
2789
+ "epoch": 2.64,
2790
+ "learning_rate": 0.00010905796014203696,
2791
+ "loss": 0.37,
2792
+ "step": 928
2793
+ },
2794
+ {
2795
+ "epoch": 2.65,
2796
+ "learning_rate": 0.00010823658791784657,
2797
+ "loss": 0.3754,
2798
+ "step": 930
2799
+ },
2800
+ {
2801
+ "epoch": 2.66,
2802
+ "learning_rate": 0.00010741717143883687,
2803
+ "loss": 0.3842,
2804
+ "step": 932
2805
+ },
2806
+ {
2807
+ "epoch": 2.66,
2808
+ "learning_rate": 0.00010659972816915598,
2809
+ "loss": 0.386,
2810
+ "step": 934
2811
+ },
2812
+ {
2813
+ "epoch": 2.67,
2814
+ "learning_rate": 0.00010578427553089747,
2815
+ "loss": 0.3629,
2816
+ "step": 936
2817
+ },
2818
+ {
2819
+ "epoch": 2.67,
2820
+ "learning_rate": 0.0001049708309037285,
2821
+ "loss": 0.434,
2822
+ "step": 938
2823
+ },
2824
+ {
2825
+ "epoch": 2.68,
2826
+ "learning_rate": 0.00010415941162451974,
2827
+ "loss": 0.3866,
2828
+ "step": 940
2829
+ },
2830
+ {
2831
+ "epoch": 2.68,
2832
+ "learning_rate": 0.00010335003498697582,
2833
+ "loss": 0.4052,
2834
+ "step": 942
2835
+ },
2836
+ {
2837
+ "epoch": 2.69,
2838
+ "learning_rate": 0.0001025427182412668,
2839
+ "loss": 0.3788,
2840
+ "step": 944
2841
+ },
2842
+ {
2843
+ "epoch": 2.7,
2844
+ "learning_rate": 0.00010173747859366037,
2845
+ "loss": 0.3999,
2846
+ "step": 946
2847
+ },
2848
+ {
2849
+ "epoch": 2.7,
2850
+ "learning_rate": 0.00010093433320615525,
2851
+ "loss": 0.3692,
2852
+ "step": 948
2853
+ },
2854
+ {
2855
+ "epoch": 2.71,
2856
+ "learning_rate": 0.00010013329919611526,
2857
+ "loss": 0.4155,
2858
+ "step": 950
2859
+ },
2860
+ {
2861
+ "epoch": 2.71,
2862
+ "learning_rate": 9.933439363590487e-05,
2863
+ "loss": 0.4225,
2864
+ "step": 952
2865
+ },
2866
+ {
2867
+ "epoch": 2.72,
2868
+ "learning_rate": 9.853763355252501e-05,
2869
+ "loss": 0.403,
2870
+ "step": 954
2871
+ },
2872
+ {
2873
+ "epoch": 2.72,
2874
+ "learning_rate": 9.774303592725018e-05,
2875
+ "loss": 0.3906,
2876
+ "step": 956
2877
+ },
2878
+ {
2879
+ "epoch": 2.73,
2880
+ "learning_rate": 9.695061769526664e-05,
2881
+ "loss": 0.3817,
2882
+ "step": 958
2883
+ },
2884
+ {
2885
+ "epoch": 2.74,
2886
+ "learning_rate": 9.616039574531133e-05,
2887
+ "loss": 0.3836,
2888
+ "step": 960
2889
+ },
2890
+ {
2891
+ "epoch": 2.74,
2892
+ "learning_rate": 9.537238691931234e-05,
2893
+ "loss": 0.377,
2894
+ "step": 962
2895
+ },
2896
+ {
2897
+ "epoch": 2.75,
2898
+ "learning_rate": 9.458660801202939e-05,
2899
+ "loss": 0.3961,
2900
+ "step": 964
2901
+ },
2902
+ {
2903
+ "epoch": 2.75,
2904
+ "learning_rate": 9.38030757706962e-05,
2905
+ "loss": 0.3856,
2906
+ "step": 966
2907
+ },
2908
+ {
2909
+ "epoch": 2.76,
2910
+ "learning_rate": 9.302180689466349e-05,
2911
+ "loss": 0.417,
2912
+ "step": 968
2913
+ },
2914
+ {
2915
+ "epoch": 2.76,
2916
+ "learning_rate": 9.224281803504322e-05,
2917
+ "loss": 0.4236,
2918
+ "step": 970
2919
+ },
2920
+ {
2921
+ "epoch": 2.77,
2922
+ "learning_rate": 9.146612579435343e-05,
2923
+ "loss": 0.3682,
2924
+ "step": 972
2925
+ },
2926
+ {
2927
+ "epoch": 2.77,
2928
+ "learning_rate": 9.069174672616456e-05,
2929
+ "loss": 0.4029,
2930
+ "step": 974
2931
+ },
2932
+ {
2933
+ "epoch": 2.78,
2934
+ "learning_rate": 8.991969733474652e-05,
2935
+ "loss": 0.3887,
2936
+ "step": 976
2937
+ },
2938
+ {
2939
+ "epoch": 2.79,
2940
+ "learning_rate": 8.914999407471735e-05,
2941
+ "loss": 0.4017,
2942
+ "step": 978
2943
+ },
2944
+ {
2945
+ "epoch": 2.79,
2946
+ "learning_rate": 8.838265335069186e-05,
2947
+ "loss": 0.3597,
2948
+ "step": 980
2949
+ },
2950
+ {
2951
+ "epoch": 2.8,
2952
+ "learning_rate": 8.761769151693253e-05,
2953
+ "loss": 0.3692,
2954
+ "step": 982
2955
+ },
2956
+ {
2957
+ "epoch": 2.8,
2958
+ "learning_rate": 8.685512487700062e-05,
2959
+ "loss": 0.4036,
2960
+ "step": 984
2961
+ },
2962
+ {
2963
+ "epoch": 2.81,
2964
+ "learning_rate": 8.609496968340904e-05,
2965
+ "loss": 0.3759,
2966
+ "step": 986
2967
+ },
2968
+ {
2969
+ "epoch": 2.81,
2970
+ "learning_rate": 8.533724213727572e-05,
2971
+ "loss": 0.4027,
2972
+ "step": 988
2973
+ },
2974
+ {
2975
+ "epoch": 2.82,
2976
+ "learning_rate": 8.45819583879783e-05,
2977
+ "loss": 0.3934,
2978
+ "step": 990
2979
+ },
2980
+ {
2981
+ "epoch": 2.83,
2982
+ "learning_rate": 8.382913453280991e-05,
2983
+ "loss": 0.362,
2984
+ "step": 992
2985
+ },
2986
+ {
2987
+ "epoch": 2.83,
2988
+ "learning_rate": 8.307878661663632e-05,
2989
+ "loss": 0.4174,
2990
+ "step": 994
2991
+ },
2992
+ {
2993
+ "epoch": 2.84,
2994
+ "learning_rate": 8.233093063155392e-05,
2995
+ "loss": 0.3895,
2996
+ "step": 996
2997
+ },
2998
+ {
2999
+ "epoch": 2.84,
3000
+ "learning_rate": 8.158558251654861e-05,
3001
+ "loss": 0.3769,
3002
+ "step": 998
3003
+ },
3004
+ {
3005
+ "epoch": 2.85,
3006
+ "learning_rate": 8.084275815715637e-05,
3007
+ "loss": 0.4198,
3008
+ "step": 1000
3009
+ },
3010
+ {
3011
+ "epoch": 2.85,
3012
+ "learning_rate": 8.010247338512453e-05,
3013
+ "loss": 0.3912,
3014
+ "step": 1002
3015
+ },
3016
+ {
3017
+ "epoch": 2.86,
3018
+ "learning_rate": 7.936474397807463e-05,
3019
+ "loss": 0.3714,
3020
+ "step": 1004
3021
+ },
3022
+ {
3023
+ "epoch": 2.87,
3024
+ "learning_rate": 7.862958565916579e-05,
3025
+ "loss": 0.3935,
3026
+ "step": 1006
3027
+ },
3028
+ {
3029
+ "epoch": 2.87,
3030
+ "learning_rate": 7.789701409675976e-05,
3031
+ "loss": 0.3822,
3032
+ "step": 1008
3033
+ },
3034
+ {
3035
+ "epoch": 2.88,
3036
+ "learning_rate": 7.716704490408697e-05,
3037
+ "loss": 0.3716,
3038
+ "step": 1010
3039
+ },
3040
+ {
3041
+ "epoch": 2.88,
3042
+ "learning_rate": 7.643969363891397e-05,
3043
+ "loss": 0.3969,
3044
+ "step": 1012
3045
+ },
3046
+ {
3047
+ "epoch": 2.89,
3048
+ "learning_rate": 7.571497580321145e-05,
3049
+ "loss": 0.4048,
3050
+ "step": 1014
3051
+ },
3052
+ {
3053
+ "epoch": 2.89,
3054
+ "learning_rate": 7.499290684282414e-05,
3055
+ "loss": 0.4254,
3056
+ "step": 1016
3057
+ },
3058
+ {
3059
+ "epoch": 2.9,
3060
+ "learning_rate": 7.42735021471414e-05,
3061
+ "loss": 0.3768,
3062
+ "step": 1018
3063
+ },
3064
+ {
3065
+ "epoch": 2.91,
3066
+ "learning_rate": 7.355677704876976e-05,
3067
+ "loss": 0.3968,
3068
+ "step": 1020
3069
+ },
3070
+ {
3071
+ "epoch": 2.91,
3072
+ "learning_rate": 7.284274682320533e-05,
3073
+ "loss": 0.3742,
3074
+ "step": 1022
3075
+ },
3076
+ {
3077
+ "epoch": 2.92,
3078
+ "learning_rate": 7.213142668850876e-05,
3079
+ "loss": 0.3829,
3080
+ "step": 1024
3081
+ },
3082
+ {
3083
+ "epoch": 2.92,
3084
+ "learning_rate": 7.142283180498079e-05,
3085
+ "loss": 0.3943,
3086
+ "step": 1026
3087
+ },
3088
+ {
3089
+ "epoch": 2.93,
3090
+ "learning_rate": 7.071697727483924e-05,
3091
+ "loss": 0.3884,
3092
+ "step": 1028
3093
+ },
3094
+ {
3095
+ "epoch": 2.93,
3096
+ "learning_rate": 7.001387814189685e-05,
3097
+ "loss": 0.3574,
3098
+ "step": 1030
3099
+ },
3100
+ {
3101
+ "epoch": 2.94,
3102
+ "learning_rate": 6.931354939124097e-05,
3103
+ "loss": 0.3909,
3104
+ "step": 1032
3105
+ },
3106
+ {
3107
+ "epoch": 2.95,
3108
+ "learning_rate": 6.861600594891391e-05,
3109
+ "loss": 0.3953,
3110
+ "step": 1034
3111
+ },
3112
+ {
3113
+ "epoch": 2.95,
3114
+ "learning_rate": 6.792126268159516e-05,
3115
+ "loss": 0.3741,
3116
+ "step": 1036
3117
+ },
3118
+ {
3119
+ "epoch": 2.96,
3120
+ "learning_rate": 6.722933439628416e-05,
3121
+ "loss": 0.3807,
3122
+ "step": 1038
3123
+ },
3124
+ {
3125
+ "epoch": 2.96,
3126
+ "learning_rate": 6.654023583998495e-05,
3127
+ "loss": 0.3761,
3128
+ "step": 1040
3129
+ },
3130
+ {
3131
+ "epoch": 2.97,
3132
+ "learning_rate": 6.585398169939179e-05,
3133
+ "loss": 0.385,
3134
+ "step": 1042
3135
+ },
3136
+ {
3137
+ "epoch": 2.97,
3138
+ "learning_rate": 6.517058660057629e-05,
3139
+ "loss": 0.3625,
3140
+ "step": 1044
3141
+ },
3142
+ {
3143
+ "epoch": 2.98,
3144
+ "learning_rate": 6.449006510867544e-05,
3145
+ "loss": 0.3673,
3146
+ "step": 1046
3147
+ },
3148
+ {
3149
+ "epoch": 2.99,
3150
+ "learning_rate": 6.381243172758133e-05,
3151
+ "loss": 0.3793,
3152
+ "step": 1048
3153
+ },
3154
+ {
3155
+ "epoch": 2.99,
3156
+ "learning_rate": 6.313770089963196e-05,
3157
+ "loss": 0.4076,
3158
+ "step": 1050
3159
+ },
3160
+ {
3161
+ "epoch": 3.0,
3162
+ "learning_rate": 6.246588700530364e-05,
3163
+ "loss": 0.4015,
3164
+ "step": 1052
3165
+ },
3166
+ {
3167
+ "epoch": 3.0,
3168
+ "learning_rate": 6.179700436290428e-05,
3169
+ "loss": 0.387,
3170
+ "step": 1054
3171
+ },
3172
+ {
3173
+ "epoch": 3.01,
3174
+ "learning_rate": 6.113106722826822e-05,
3175
+ "loss": 0.3787,
3176
+ "step": 1056
3177
+ },
3178
+ {
3179
+ "epoch": 3.01,
3180
+ "learning_rate": 6.0468089794452463e-05,
3181
+ "loss": 0.3771,
3182
+ "step": 1058
3183
+ },
3184
+ {
3185
+ "epoch": 3.02,
3186
+ "learning_rate": 5.9808086191434185e-05,
3187
+ "loss": 0.3622,
3188
+ "step": 1060
3189
+ },
3190
+ {
3191
+ "epoch": 3.03,
3192
+ "learning_rate": 5.9151070485809704e-05,
3193
+ "loss": 0.3603,
3194
+ "step": 1062
3195
+ },
3196
+ {
3197
+ "epoch": 3.03,
3198
+ "learning_rate": 5.849705668049443e-05,
3199
+ "loss": 0.3642,
3200
+ "step": 1064
3201
+ },
3202
+ {
3203
+ "epoch": 3.04,
3204
+ "learning_rate": 5.7846058714424546e-05,
3205
+ "loss": 0.3696,
3206
+ "step": 1066
3207
+ },
3208
+ {
3209
+ "epoch": 3.04,
3210
+ "learning_rate": 5.7198090462259926e-05,
3211
+ "loss": 0.3658,
3212
+ "step": 1068
3213
+ },
3214
+ {
3215
+ "epoch": 3.05,
3216
+ "learning_rate": 5.6553165734088575e-05,
3217
+ "loss": 0.3847,
3218
+ "step": 1070
3219
+ },
3220
+ {
3221
+ "epoch": 3.05,
3222
+ "learning_rate": 5.591129827513197e-05,
3223
+ "loss": 0.3739,
3224
+ "step": 1072
3225
+ },
3226
+ {
3227
+ "epoch": 3.06,
3228
+ "learning_rate": 5.52725017654524e-05,
3229
+ "loss": 0.3892,
3230
+ "step": 1074
3231
+ },
3232
+ {
3233
+ "epoch": 3.07,
3234
+ "learning_rate": 5.463678981966114e-05,
3235
+ "loss": 0.3983,
3236
+ "step": 1076
3237
+ },
3238
+ {
3239
+ "epoch": 3.07,
3240
+ "learning_rate": 5.4004175986628705e-05,
3241
+ "loss": 0.3799,
3242
+ "step": 1078
3243
+ },
3244
+ {
3245
+ "epoch": 3.08,
3246
+ "learning_rate": 5.33746737491956e-05,
3247
+ "loss": 0.3927,
3248
+ "step": 1080
3249
+ },
3250
+ {
3251
+ "epoch": 3.08,
3252
+ "learning_rate": 5.274829652388524e-05,
3253
+ "loss": 0.3849,
3254
+ "step": 1082
3255
+ },
3256
+ {
3257
+ "epoch": 3.09,
3258
+ "learning_rate": 5.2125057660617905e-05,
3259
+ "loss": 0.3722,
3260
+ "step": 1084
3261
+ },
3262
+ {
3263
+ "epoch": 3.09,
3264
+ "learning_rate": 5.15049704424263e-05,
3265
+ "loss": 0.3493,
3266
+ "step": 1086
3267
+ },
3268
+ {
3269
+ "epoch": 3.1,
3270
+ "learning_rate": 5.088804808517253e-05,
3271
+ "loss": 0.3846,
3272
+ "step": 1088
3273
+ },
3274
+ {
3275
+ "epoch": 3.11,
3276
+ "learning_rate": 5.027430373726607e-05,
3277
+ "loss": 0.3681,
3278
+ "step": 1090
3279
+ },
3280
+ {
3281
+ "epoch": 3.11,
3282
+ "learning_rate": 4.9663750479383894e-05,
3283
+ "loss": 0.3685,
3284
+ "step": 1092
3285
+ },
3286
+ {
3287
+ "epoch": 3.12,
3288
+ "learning_rate": 4.9056401324191516e-05,
3289
+ "loss": 0.3818,
3290
+ "step": 1094
3291
+ },
3292
+ {
3293
+ "epoch": 3.12,
3294
+ "learning_rate": 4.845226921606587e-05,
3295
+ "loss": 0.3884,
3296
+ "step": 1096
3297
+ },
3298
+ {
3299
+ "epoch": 3.13,
3300
+ "learning_rate": 4.7851367030819026e-05,
3301
+ "loss": 0.3732,
3302
+ "step": 1098
3303
+ },
3304
+ {
3305
+ "epoch": 3.13,
3306
+ "learning_rate": 4.725370757542418e-05,
3307
+ "loss": 0.4064,
3308
+ "step": 1100
3309
+ },
3310
+ {
3311
+ "epoch": 3.14,
3312
+ "learning_rate": 4.6659303587742356e-05,
3313
+ "loss": 0.36,
3314
+ "step": 1102
3315
+ },
3316
+ {
3317
+ "epoch": 3.15,
3318
+ "learning_rate": 4.6068167736251335e-05,
3319
+ "loss": 0.3637,
3320
+ "step": 1104
3321
+ },
3322
+ {
3323
+ "epoch": 3.15,
3324
+ "learning_rate": 4.548031261977519e-05,
3325
+ "loss": 0.3608,
3326
+ "step": 1106
3327
+ },
3328
+ {
3329
+ "epoch": 3.16,
3330
+ "learning_rate": 4.4895750767216085e-05,
3331
+ "loss": 0.3946,
3332
+ "step": 1108
3333
+ },
3334
+ {
3335
+ "epoch": 3.16,
3336
+ "learning_rate": 4.431449463728707e-05,
3337
+ "loss": 0.3578,
3338
+ "step": 1110
3339
+ },
3340
+ {
3341
+ "epoch": 3.17,
3342
+ "learning_rate": 4.373655661824685e-05,
3343
+ "loss": 0.3779,
3344
+ "step": 1112
3345
+ },
3346
+ {
3347
+ "epoch": 3.17,
3348
+ "learning_rate": 4.316194902763531e-05,
3349
+ "loss": 0.3648,
3350
+ "step": 1114
3351
+ },
3352
+ {
3353
+ "epoch": 3.18,
3354
+ "learning_rate": 4.259068411201128e-05,
3355
+ "loss": 0.3845,
3356
+ "step": 1116
3357
+ },
3358
+ {
3359
+ "epoch": 3.19,
3360
+ "learning_rate": 4.202277404669144e-05,
3361
+ "loss": 0.373,
3362
+ "step": 1118
3363
+ },
3364
+ {
3365
+ "epoch": 3.19,
3366
+ "learning_rate": 4.1458230935490974e-05,
3367
+ "loss": 0.367,
3368
+ "step": 1120
3369
+ },
3370
+ {
3371
+ "epoch": 3.2,
3372
+ "learning_rate": 4.089706681046541e-05,
3373
+ "loss": 0.3856,
3374
+ "step": 1122
3375
+ },
3376
+ {
3377
+ "epoch": 3.2,
3378
+ "learning_rate": 4.033929363165425e-05,
3379
+ "loss": 0.3748,
3380
+ "step": 1124
3381
+ },
3382
+ {
3383
+ "epoch": 3.21,
3384
+ "learning_rate": 3.978492328682597e-05,
3385
+ "loss": 0.359,
3386
+ "step": 1126
3387
+ },
3388
+ {
3389
+ "epoch": 3.21,
3390
+ "learning_rate": 3.923396759122497e-05,
3391
+ "loss": 0.3932,
3392
+ "step": 1128
3393
+ },
3394
+ {
3395
+ "epoch": 3.22,
3396
+ "learning_rate": 3.868643828731942e-05,
3397
+ "loss": 0.3848,
3398
+ "step": 1130
3399
+ },
3400
+ {
3401
+ "epoch": 3.23,
3402
+ "learning_rate": 3.8142347044551044e-05,
3403
+ "loss": 0.3626,
3404
+ "step": 1132
3405
+ },
3406
+ {
3407
+ "epoch": 3.23,
3408
+ "learning_rate": 3.760170545908659e-05,
3409
+ "loss": 0.3665,
3410
+ "step": 1134
3411
+ },
3412
+ {
3413
+ "epoch": 3.24,
3414
+ "learning_rate": 3.706452505357063e-05,
3415
+ "loss": 0.3594,
3416
+ "step": 1136
3417
+ },
3418
+ {
3419
+ "epoch": 3.24,
3420
+ "learning_rate": 3.653081727687979e-05,
3421
+ "loss": 0.3892,
3422
+ "step": 1138
3423
+ },
3424
+ {
3425
+ "epoch": 3.25,
3426
+ "learning_rate": 3.6000593503878965e-05,
3427
+ "loss": 0.3906,
3428
+ "step": 1140
3429
+ },
3430
+ {
3431
+ "epoch": 3.25,
3432
+ "learning_rate": 3.547386503517869e-05,
3433
+ "loss": 0.3774,
3434
+ "step": 1142
3435
+ },
3436
+ {
3437
+ "epoch": 3.26,
3438
+ "learning_rate": 3.495064309689462e-05,
3439
+ "loss": 0.3849,
3440
+ "step": 1144
3441
+ },
3442
+ {
3443
+ "epoch": 3.26,
3444
+ "learning_rate": 3.44309388404078e-05,
3445
+ "loss": 0.3646,
3446
+ "step": 1146
3447
+ },
3448
+ {
3449
+ "epoch": 3.27,
3450
+ "learning_rate": 3.391476334212742e-05,
3451
+ "loss": 0.3791,
3452
+ "step": 1148
3453
+ },
3454
+ {
3455
+ "epoch": 3.28,
3456
+ "learning_rate": 3.340212760325441e-05,
3457
+ "loss": 0.3775,
3458
+ "step": 1150
3459
+ },
3460
+ {
3461
+ "epoch": 3.28,
3462
+ "learning_rate": 3.2893042549547305e-05,
3463
+ "loss": 0.3772,
3464
+ "step": 1152
3465
+ },
3466
+ {
3467
+ "epoch": 3.29,
3468
+ "learning_rate": 3.2387519031089185e-05,
3469
+ "loss": 0.3676,
3470
+ "step": 1154
3471
+ },
3472
+ {
3473
+ "epoch": 3.29,
3474
+ "learning_rate": 3.188556782205634e-05,
3475
+ "loss": 0.3717,
3476
+ "step": 1156
3477
+ },
3478
+ {
3479
+ "epoch": 3.3,
3480
+ "learning_rate": 3.1387199620488816e-05,
3481
+ "loss": 0.3529,
3482
+ "step": 1158
3483
+ },
3484
+ {
3485
+ "epoch": 3.3,
3486
+ "learning_rate": 3.0892425048062355e-05,
3487
+ "loss": 0.387,
3488
+ "step": 1160
3489
+ },
3490
+ {
3491
+ "epoch": 3.31,
3492
+ "learning_rate": 3.0401254649862075e-05,
3493
+ "loss": 0.3555,
3494
+ "step": 1162
3495
+ },
3496
+ {
3497
+ "epoch": 3.32,
3498
+ "learning_rate": 2.991369889415756e-05,
3499
+ "loss": 0.3747,
3500
+ "step": 1164
3501
+ },
3502
+ {
3503
+ "epoch": 3.32,
3504
+ "learning_rate": 2.9429768172179907e-05,
3505
+ "loss": 0.3653,
3506
+ "step": 1166
3507
+ },
3508
+ {
3509
+ "epoch": 3.33,
3510
+ "learning_rate": 2.89494727979001e-05,
3511
+ "loss": 0.3814,
3512
+ "step": 1168
3513
+ },
3514
+ {
3515
+ "epoch": 3.33,
3516
+ "learning_rate": 2.847282300780949e-05,
3517
+ "loss": 0.3917,
3518
+ "step": 1170
3519
+ },
3520
+ {
3521
+ "epoch": 3.34,
3522
+ "learning_rate": 2.7999828960701258e-05,
3523
+ "loss": 0.3891,
3524
+ "step": 1172
3525
+ },
3526
+ {
3527
+ "epoch": 3.34,
3528
+ "learning_rate": 2.7530500737454135e-05,
3529
+ "loss": 0.366,
3530
+ "step": 1174
3531
+ },
3532
+ {
3533
+ "epoch": 3.35,
3534
+ "learning_rate": 2.7064848340817418e-05,
3535
+ "loss": 0.3816,
3536
+ "step": 1176
3537
+ },
3538
+ {
3539
+ "epoch": 3.36,
3540
+ "learning_rate": 2.6602881695198002e-05,
3541
+ "loss": 0.3734,
3542
+ "step": 1178
3543
+ },
3544
+ {
3545
+ "epoch": 3.36,
3546
+ "learning_rate": 2.6144610646448553e-05,
3547
+ "loss": 0.3593,
3548
+ "step": 1180
3549
+ },
3550
+ {
3551
+ "epoch": 3.37,
3552
+ "learning_rate": 2.5690044961657877e-05,
3553
+ "loss": 0.3732,
3554
+ "step": 1182
3555
+ },
3556
+ {
3557
+ "epoch": 3.37,
3558
+ "learning_rate": 2.5239194328942684e-05,
3559
+ "loss": 0.3699,
3560
+ "step": 1184
3561
+ },
3562
+ {
3563
+ "epoch": 3.38,
3564
+ "learning_rate": 2.4792068357241148e-05,
3565
+ "loss": 0.3829,
3566
+ "step": 1186
3567
+ },
3568
+ {
3569
+ "epoch": 3.38,
3570
+ "learning_rate": 2.434867657610813e-05,
3571
+ "loss": 0.3397,
3572
+ "step": 1188
3573
+ },
3574
+ {
3575
+ "epoch": 3.39,
3576
+ "learning_rate": 2.3909028435511928e-05,
3577
+ "loss": 0.3451,
3578
+ "step": 1190
3579
+ },
3580
+ {
3581
+ "epoch": 3.4,
3582
+ "learning_rate": 2.3473133305633012e-05,
3583
+ "loss": 0.3498,
3584
+ "step": 1192
3585
+ },
3586
+ {
3587
+ "epoch": 3.4,
3588
+ "learning_rate": 2.3041000476664242e-05,
3589
+ "loss": 0.3585,
3590
+ "step": 1194
3591
+ },
3592
+ {
3593
+ "epoch": 3.41,
3594
+ "learning_rate": 2.2612639158613046e-05,
3595
+ "loss": 0.3637,
3596
+ "step": 1196
3597
+ },
3598
+ {
3599
+ "epoch": 3.41,
3600
+ "learning_rate": 2.218805848110481e-05,
3601
+ "loss": 0.3702,
3602
+ "step": 1198
3603
+ },
3604
+ {
3605
+ "epoch": 3.42,
3606
+ "learning_rate": 2.176726749318856e-05,
3607
+ "loss": 0.3695,
3608
+ "step": 1200
3609
+ },
3610
+ {
3611
+ "epoch": 3.42,
3612
+ "learning_rate": 2.1350275163143983e-05,
3613
+ "loss": 0.3509,
3614
+ "step": 1202
3615
+ },
3616
+ {
3617
+ "epoch": 3.43,
3618
+ "learning_rate": 2.0937090378290412e-05,
3619
+ "loss": 0.3992,
3620
+ "step": 1204
3621
+ },
3622
+ {
3623
+ "epoch": 3.44,
3624
+ "learning_rate": 2.0527721944797216e-05,
3625
+ "loss": 0.3609,
3626
+ "step": 1206
3627
+ },
3628
+ {
3629
+ "epoch": 3.44,
3630
+ "learning_rate": 2.0122178587496253e-05,
3631
+ "loss": 0.3842,
3632
+ "step": 1208
3633
+ },
3634
+ {
3635
+ "epoch": 3.45,
3636
+ "learning_rate": 1.9720468949695815e-05,
3637
+ "loss": 0.3618,
3638
+ "step": 1210
3639
+ },
3640
+ {
3641
+ "epoch": 3.45,
3642
+ "learning_rate": 1.932260159299666e-05,
3643
+ "loss": 0.3936,
3644
+ "step": 1212
3645
+ },
3646
+ {
3647
+ "epoch": 3.46,
3648
+ "learning_rate": 1.8928584997109188e-05,
3649
+ "loss": 0.3775,
3650
+ "step": 1214
3651
+ },
3652
+ {
3653
+ "epoch": 3.46,
3654
+ "learning_rate": 1.8538427559672923e-05,
3655
+ "loss": 0.3782,
3656
+ "step": 1216
3657
+ },
3658
+ {
3659
+ "epoch": 3.47,
3660
+ "learning_rate": 1.815213759607759e-05,
3661
+ "loss": 0.377,
3662
+ "step": 1218
3663
+ },
3664
+ {
3665
+ "epoch": 3.48,
3666
+ "learning_rate": 1.776972333928566e-05,
3667
+ "loss": 0.3679,
3668
+ "step": 1220
3669
+ },
3670
+ {
3671
+ "epoch": 3.48,
3672
+ "learning_rate": 1.7391192939657184e-05,
3673
+ "loss": 0.3905,
3674
+ "step": 1222
3675
+ },
3676
+ {
3677
+ "epoch": 3.49,
3678
+ "learning_rate": 1.701655446477577e-05,
3679
+ "loss": 0.3598,
3680
+ "step": 1224
3681
+ },
3682
+ {
3683
+ "epoch": 3.49,
3684
+ "learning_rate": 1.6645815899276852e-05,
3685
+ "loss": 0.3906,
3686
+ "step": 1226
3687
+ },
3688
+ {
3689
+ "epoch": 3.5,
3690
+ "learning_rate": 1.6278985144677493e-05,
3691
+ "loss": 0.3544,
3692
+ "step": 1228
3693
+ },
3694
+ {
3695
+ "epoch": 3.5,
3696
+ "learning_rate": 1.5916070019207874e-05,
3697
+ "loss": 0.3591,
3698
+ "step": 1230
3699
+ },
3700
+ {
3701
+ "epoch": 3.51,
3702
+ "learning_rate": 1.5557078257644763e-05,
3703
+ "loss": 0.3525,
3704
+ "step": 1232
3705
+ },
3706
+ {
3707
+ "epoch": 3.52,
3708
+ "learning_rate": 1.5202017511146605e-05,
3709
+ "loss": 0.3536,
3710
+ "step": 1234
3711
+ },
3712
+ {
3713
+ "epoch": 3.52,
3714
+ "learning_rate": 1.4850895347090588e-05,
3715
+ "loss": 0.3589,
3716
+ "step": 1236
3717
+ },
3718
+ {
3719
+ "epoch": 3.53,
3720
+ "learning_rate": 1.4503719248911074e-05,
3721
+ "loss": 0.3666,
3722
+ "step": 1238
3723
+ },
3724
+ {
3725
+ "epoch": 3.53,
3726
+ "learning_rate": 1.4160496615940455e-05,
3727
+ "loss": 0.3742,
3728
+ "step": 1240
3729
+ },
3730
+ {
3731
+ "epoch": 3.54,
3732
+ "learning_rate": 1.3821234763251145e-05,
3733
+ "loss": 0.3774,
3734
+ "step": 1242
3735
+ },
3736
+ {
3737
+ "epoch": 3.54,
3738
+ "learning_rate": 1.348594092149995e-05,
3739
+ "loss": 0.38,
3740
+ "step": 1244
3741
+ },
3742
+ {
3743
+ "epoch": 3.55,
3744
+ "learning_rate": 1.3154622236773706e-05,
3745
+ "loss": 0.3646,
3746
+ "step": 1246
3747
+ },
3748
+ {
3749
+ "epoch": 3.56,
3750
+ "learning_rate": 1.2827285770437148e-05,
3751
+ "loss": 0.3869,
3752
+ "step": 1248
3753
+ },
3754
+ {
3755
+ "epoch": 3.56,
3756
+ "learning_rate": 1.2503938498982304e-05,
3757
+ "loss": 0.3581,
3758
+ "step": 1250
3759
+ },
3760
+ {
3761
+ "epoch": 3.57,
3762
+ "learning_rate": 1.218458731387997e-05,
3763
+ "loss": 0.3802,
3764
+ "step": 1252
3765
+ },
3766
+ {
3767
+ "epoch": 3.57,
3768
+ "learning_rate": 1.1869239021432554e-05,
3769
+ "loss": 0.3986,
3770
+ "step": 1254
3771
+ },
3772
+ {
3773
+ "epoch": 3.58,
3774
+ "learning_rate": 1.1557900342629403e-05,
3775
+ "loss": 0.3718,
3776
+ "step": 1256
3777
+ },
3778
+ {
3779
+ "epoch": 3.58,
3780
+ "learning_rate": 1.1250577913003101e-05,
3781
+ "loss": 0.3724,
3782
+ "step": 1258
3783
+ },
3784
+ {
3785
+ "epoch": 3.59,
3786
+ "learning_rate": 1.0947278282488404e-05,
3787
+ "loss": 0.3783,
3788
+ "step": 1260
3789
+ },
3790
+ {
3791
+ "epoch": 3.6,
3792
+ "learning_rate": 1.0648007915282554e-05,
3793
+ "loss": 0.3928,
3794
+ "step": 1262
3795
+ },
3796
+ {
3797
+ "epoch": 3.6,
3798
+ "learning_rate": 1.0352773189707377e-05,
3799
+ "loss": 0.3598,
3800
+ "step": 1264
3801
+ },
3802
+ {
3803
+ "epoch": 3.61,
3804
+ "learning_rate": 1.00615803980735e-05,
3805
+ "loss": 0.3629,
3806
+ "step": 1266
3807
+ },
3808
+ {
3809
+ "epoch": 3.61,
3810
+ "learning_rate": 9.77443574654615e-06,
3811
+ "loss": 0.3759,
3812
+ "step": 1268
3813
+ },
3814
+ {
3815
+ "epoch": 3.62,
3816
+ "learning_rate": 9.49134535501297e-06,
3817
+ "loss": 0.3815,
3818
+ "step": 1270
3819
+ },
3820
+ {
3821
+ "epoch": 3.62,
3822
+ "learning_rate": 9.212315256953497e-06,
3823
+ "loss": 0.3566,
3824
+ "step": 1272
3825
+ },
3826
+ {
3827
+ "epoch": 3.63,
3828
+ "learning_rate": 8.937351399310557e-06,
3829
+ "loss": 0.3885,
3830
+ "step": 1274
3831
+ },
3832
+ {
3833
+ "epoch": 3.64,
3834
+ "learning_rate": 8.666459642363633e-06,
3835
+ "loss": 0.3698,
3836
+ "step": 1276
3837
+ },
3838
+ {
3839
+ "epoch": 3.64,
3840
+ "learning_rate": 8.399645759603924e-06,
3841
+ "loss": 0.3671,
3842
+ "step": 1278
3843
+ },
3844
+ {
3845
+ "epoch": 3.65,
3846
+ "learning_rate": 8.136915437611193e-06,
3847
+ "loss": 0.3572,
3848
+ "step": 1280
3849
+ },
3850
+ {
3851
+ "epoch": 3.65,
3852
+ "learning_rate": 7.878274275932662e-06,
3853
+ "loss": 0.3812,
3854
+ "step": 1282
3855
+ },
3856
+ {
3857
+ "epoch": 3.66,
3858
+ "learning_rate": 7.623727786963697e-06,
3859
+ "loss": 0.3933,
3860
+ "step": 1284
3861
+ },
3862
+ {
3863
+ "epoch": 3.66,
3864
+ "learning_rate": 7.373281395830289e-06,
3865
+ "loss": 0.3667,
3866
+ "step": 1286
3867
+ },
3868
+ {
3869
+ "epoch": 3.67,
3870
+ "learning_rate": 7.126940440273311e-06,
3871
+ "loss": 0.3808,
3872
+ "step": 1288
3873
+ },
3874
+ {
3875
+ "epoch": 3.68,
3876
+ "learning_rate": 6.884710170535025e-06,
3877
+ "loss": 0.3645,
3878
+ "step": 1290
3879
+ },
3880
+ {
3881
+ "epoch": 3.68,
3882
+ "learning_rate": 6.646595749246909e-06,
3883
+ "loss": 0.3694,
3884
+ "step": 1292
3885
+ },
3886
+ {
3887
+ "epoch": 3.69,
3888
+ "learning_rate": 6.412602251319744e-06,
3889
+ "loss": 0.3808,
3890
+ "step": 1294
3891
+ },
3892
+ {
3893
+ "epoch": 3.69,
3894
+ "learning_rate": 6.182734663835521e-06,
3895
+ "loss": 0.3807,
3896
+ "step": 1296
3897
+ },
3898
+ {
3899
+ "epoch": 3.7,
3900
+ "learning_rate": 5.956997885941018e-06,
3901
+ "loss": 0.3421,
3902
+ "step": 1298
3903
+ },
3904
+ {
3905
+ "epoch": 3.7,
3906
+ "learning_rate": 5.735396728743481e-06,
3907
+ "loss": 0.3903,
3908
+ "step": 1300
3909
+ },
3910
+ {
3911
+ "epoch": 3.71,
3912
+ "learning_rate": 5.517935915207995e-06,
3913
+ "loss": 0.3798,
3914
+ "step": 1302
3915
+ },
3916
+ {
3917
+ "epoch": 3.72,
3918
+ "learning_rate": 5.3046200800569655e-06,
3919
+ "loss": 0.3535,
3920
+ "step": 1304
3921
+ },
3922
+ {
3923
+ "epoch": 3.72,
3924
+ "learning_rate": 5.095453769671199e-06,
3925
+ "loss": 0.3873,
3926
+ "step": 1306
3927
+ },
3928
+ {
3929
+ "epoch": 3.73,
3930
+ "learning_rate": 4.890441441993087e-06,
3931
+ "loss": 0.352,
3932
+ "step": 1308
3933
+ },
3934
+ {
3935
+ "epoch": 3.73,
3936
+ "learning_rate": 4.689587466431511e-06,
3937
+ "loss": 0.3501,
3938
+ "step": 1310
3939
+ },
3940
+ {
3941
+ "epoch": 3.74,
3942
+ "learning_rate": 4.492896123768864e-06,
3943
+ "loss": 0.3694,
3944
+ "step": 1312
3945
+ },
3946
+ {
3947
+ "epoch": 3.74,
3948
+ "learning_rate": 4.300371606069686e-06,
3949
+ "loss": 0.3873,
3950
+ "step": 1314
3951
+ },
3952
+ {
3953
+ "epoch": 3.75,
3954
+ "learning_rate": 4.112018016591313e-06,
3955
+ "loss": 0.3556,
3956
+ "step": 1316
3957
+ },
3958
+ {
3959
+ "epoch": 3.75,
3960
+ "learning_rate": 3.92783936969654e-06,
3961
+ "loss": 0.3664,
3962
+ "step": 1318
3963
+ },
3964
+ {
3965
+ "epoch": 3.76,
3966
+ "learning_rate": 3.747839590767965e-06,
3967
+ "loss": 0.3754,
3968
+ "step": 1320
3969
+ },
3970
+ {
3971
+ "epoch": 3.77,
3972
+ "learning_rate": 3.5720225161243624e-06,
3973
+ "loss": 0.3835,
3974
+ "step": 1322
3975
+ },
3976
+ {
3977
+ "epoch": 3.77,
3978
+ "learning_rate": 3.4003918929389033e-06,
3979
+ "loss": 0.3751,
3980
+ "step": 1324
3981
+ },
3982
+ {
3983
+ "epoch": 3.78,
3984
+ "learning_rate": 3.232951379159288e-06,
3985
+ "loss": 0.3577,
3986
+ "step": 1326
3987
+ },
3988
+ {
3989
+ "epoch": 3.78,
3990
+ "learning_rate": 3.0697045434298744e-06,
3991
+ "loss": 0.3515,
3992
+ "step": 1328
3993
+ },
3994
+ {
3995
+ "epoch": 3.79,
3996
+ "learning_rate": 2.91065486501545e-06,
3997
+ "loss": 0.3856,
3998
+ "step": 1330
3999
+ },
4000
+ {
4001
+ "epoch": 3.79,
4002
+ "learning_rate": 2.7558057337272234e-06,
4003
+ "loss": 0.3807,
4004
+ "step": 1332
4005
+ },
4006
+ {
4007
+ "epoch": 3.8,
4008
+ "learning_rate": 2.605160449850508e-06,
4009
+ "loss": 0.369,
4010
+ "step": 1334
4011
+ },
4012
+ {
4013
+ "epoch": 3.81,
4014
+ "learning_rate": 2.458722224074439e-06,
4015
+ "loss": 0.3743,
4016
+ "step": 1336
4017
+ },
4018
+ {
4019
+ "epoch": 3.81,
4020
+ "learning_rate": 2.316494177423456e-06,
4021
+ "loss": 0.359,
4022
+ "step": 1338
4023
+ },
4024
+ {
4025
+ "epoch": 3.82,
4026
+ "learning_rate": 2.178479341190909e-06,
4027
+ "loss": 0.3892,
4028
+ "step": 1340
4029
+ },
4030
+ {
4031
+ "epoch": 3.82,
4032
+ "learning_rate": 2.0446806568742873e-06,
4033
+ "loss": 0.3819,
4034
+ "step": 1342
4035
+ },
4036
+ {
4037
+ "epoch": 3.83,
4038
+ "learning_rate": 1.9151009761126937e-06,
4039
+ "loss": 0.4031,
4040
+ "step": 1344
4041
+ },
4042
+ {
4043
+ "epoch": 3.83,
4044
+ "learning_rate": 1.7897430606259812e-06,
4045
+ "loss": 0.3721,
4046
+ "step": 1346
4047
+ },
4048
+ {
4049
+ "epoch": 3.84,
4050
+ "learning_rate": 1.6686095821558888e-06,
4051
+ "loss": 0.3928,
4052
+ "step": 1348
4053
+ },
4054
+ {
4055
+ "epoch": 3.85,
4056
+ "learning_rate": 1.5517031224091095e-06,
4057
+ "loss": 0.3874,
4058
+ "step": 1350
4059
+ },
4060
+ {
4061
+ "epoch": 3.85,
4062
+ "learning_rate": 1.4390261730023337e-06,
4063
+ "loss": 0.3787,
4064
+ "step": 1352
4065
+ },
4066
+ {
4067
+ "epoch": 3.86,
4068
+ "learning_rate": 1.3305811354090037e-06,
4069
+ "loss": 0.3714,
4070
+ "step": 1354
4071
+ },
4072
+ {
4073
+ "epoch": 3.86,
4074
+ "learning_rate": 1.2263703209082878e-06,
4075
+ "loss": 0.3527,
4076
+ "step": 1356
4077
+ },
4078
+ {
4079
+ "epoch": 3.87,
4080
+ "learning_rate": 1.1263959505356747e-06,
4081
+ "loss": 0.3698,
4082
+ "step": 1358
4083
+ },
4084
+ {
4085
+ "epoch": 3.87,
4086
+ "learning_rate": 1.0306601550357676e-06,
4087
+ "loss": 0.3914,
4088
+ "step": 1360
4089
+ },
4090
+ {
4091
+ "epoch": 3.88,
4092
+ "learning_rate": 9.391649748167863e-07,
4093
+ "loss": 0.3822,
4094
+ "step": 1362
4095
+ },
4096
+ {
4097
+ "epoch": 3.89,
4098
+ "learning_rate": 8.519123599071366e-07,
4099
+ "loss": 0.3616,
4100
+ "step": 1364
4101
+ },
4102
+ {
4103
+ "epoch": 3.89,
4104
+ "learning_rate": 7.689041699137534e-07,
4105
+ "loss": 0.3453,
4106
+ "step": 1366
4107
+ },
4108
+ {
4109
+ "epoch": 3.9,
4110
+ "learning_rate": 6.901421739825775e-07,
4111
+ "loss": 0.3632,
4112
+ "step": 1368
4113
+ },
4114
+ {
4115
+ "epoch": 3.9,
4116
+ "learning_rate": 6.15628050760808e-07,
4117
+ "loss": 0.3649,
4118
+ "step": 1370
4119
+ },
4120
+ {
4121
+ "epoch": 3.91,
4122
+ "learning_rate": 5.453633883611309e-07,
4123
+ "loss": 0.3617,
4124
+ "step": 1372
4125
+ },
4126
+ {
4127
+ "epoch": 3.91,
4128
+ "learning_rate": 4.793496843278345e-07,
4129
+ "loss": 0.3749,
4130
+ "step": 1374
4131
+ },
4132
+ {
4133
+ "epoch": 3.92,
4134
+ "learning_rate": 4.1758834560492457e-07,
4135
+ "loss": 0.3972,
4136
+ "step": 1376
4137
+ },
4138
+ {
4139
+ "epoch": 3.93,
4140
+ "learning_rate": 3.6008068850617025e-07,
4141
+ "loss": 0.3565,
4142
+ "step": 1378
4143
+ },
4144
+ {
4145
+ "epoch": 3.93,
4146
+ "learning_rate": 3.06827938686971e-07,
4147
+ "loss": 0.3714,
4148
+ "step": 1380
4149
+ },
4150
+ {
4151
+ "epoch": 3.94,
4152
+ "learning_rate": 2.5783123111828846e-07,
4153
+ "loss": 0.3537,
4154
+ "step": 1382
4155
+ },
4156
+ {
4157
+ "epoch": 3.94,
4158
+ "learning_rate": 2.13091610062488e-07,
4159
+ "loss": 0.3692,
4160
+ "step": 1384
4161
+ },
4162
+ {
4163
+ "epoch": 3.95,
4164
+ "learning_rate": 1.726100290509791e-07,
4165
+ "loss": 0.3682,
4166
+ "step": 1386
4167
+ },
4168
+ {
4169
+ "epoch": 3.95,
4170
+ "learning_rate": 1.363873508639646e-07,
4171
+ "loss": 0.3834,
4172
+ "step": 1388
4173
+ },
4174
+ {
4175
+ "epoch": 3.96,
4176
+ "learning_rate": 1.0442434751212205e-07,
4177
+ "loss": 0.3683,
4178
+ "step": 1390
4179
+ },
4180
+ {
4181
+ "epoch": 3.97,
4182
+ "learning_rate": 7.672170021995052e-08,
4183
+ "loss": 0.356,
4184
+ "step": 1392
4185
+ },
4186
+ {
4187
+ "epoch": 3.97,
4188
+ "learning_rate": 5.327999941147077e-08,
4189
+ "loss": 0.3902,
4190
+ "step": 1394
4191
+ },
4192
+ {
4193
+ "epoch": 3.98,
4194
+ "learning_rate": 3.40997446974578e-08,
4195
+ "loss": 0.3715,
4196
+ "step": 1396
4197
+ },
4198
+ {
4199
+ "epoch": 3.98,
4200
+ "learning_rate": 1.918134486496026e-08,
4201
+ "loss": 0.3553,
4202
+ "step": 1398
4203
+ },
4204
+ {
4205
+ "epoch": 3.99,
4206
+ "learning_rate": 8.525117868463105e-09,
4207
+ "loss": 0.3679,
4208
+ "step": 1400
4209
+ },
4210
+ {
4211
+ "epoch": 3.99,
4212
+ "learning_rate": 2.1312908231152506e-09,
4213
+ "loss": 0.3738,
4214
+ "step": 1402
4215
+ },
4216
+ {
4217
+ "epoch": 4.0,
4218
+ "learning_rate": 0.0,
4219
+ "loss": 0.3884,
4220
+ "step": 1404
4221
+ },
4222
+ {
4223
+ "epoch": 4.0,
4224
+ "step": 1404,
4225
+ "total_flos": 2.6724511228991898e+17,
4226
+ "train_loss": 0.5554745374751566,
4227
+ "train_runtime": 66017.7666,
4228
+ "train_samples_per_second": 10.888,
4229
+ "train_steps_per_second": 0.021
4230
+ }
4231
+ ],
4232
+ "max_steps": 1404,
4233
+ "num_train_epochs": 4,
4234
+ "total_flos": 2.6724511228991898e+17,
4235
+ "trial_name": null,
4236
+ "trial_params": null
4237
+ }