End of training
Browse files- trainer_state.json +1406 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1406 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 5300,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.47,
|
| 13 |
+
"learning_rate": 8.679245283018868e-07,
|
| 14 |
+
"loss": 130.658,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.47,
|
| 19 |
+
"eval_runtime": 2.1932,
|
| 20 |
+
"eval_samples_per_second": 327.381,
|
| 21 |
+
"eval_steps_per_second": 10.487,
|
| 22 |
+
"step": 50
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.94,
|
| 26 |
+
"learning_rate": 2.716981132075472e-06,
|
| 27 |
+
"loss": 119.614,
|
| 28 |
+
"step": 100
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 0.94,
|
| 32 |
+
"eval_runtime": 1.9516,
|
| 33 |
+
"eval_samples_per_second": 367.905,
|
| 34 |
+
"eval_steps_per_second": 11.785,
|
| 35 |
+
"step": 100
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.42,
|
| 39 |
+
"learning_rate": 4.603773584905661e-06,
|
| 40 |
+
"loss": 84.4188,
|
| 41 |
+
"step": 150
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 1.42,
|
| 45 |
+
"eval_runtime": 1.9342,
|
| 46 |
+
"eval_samples_per_second": 371.215,
|
| 47 |
+
"eval_steps_per_second": 11.891,
|
| 48 |
+
"step": 150
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 1.89,
|
| 52 |
+
"learning_rate": 6.490566037735849e-06,
|
| 53 |
+
"loss": 53.1678,
|
| 54 |
+
"step": 200
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 1.89,
|
| 58 |
+
"eval_runtime": 1.9027,
|
| 59 |
+
"eval_samples_per_second": 377.349,
|
| 60 |
+
"eval_steps_per_second": 12.088,
|
| 61 |
+
"step": 200
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 2.36,
|
| 65 |
+
"learning_rate": 8.377358490566039e-06,
|
| 66 |
+
"loss": 33.2104,
|
| 67 |
+
"step": 250
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 2.36,
|
| 71 |
+
"eval_runtime": 1.9458,
|
| 72 |
+
"eval_samples_per_second": 368.991,
|
| 73 |
+
"eval_steps_per_second": 11.82,
|
| 74 |
+
"step": 250
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.83,
|
| 78 |
+
"learning_rate": 1.0264150943396227e-05,
|
| 79 |
+
"loss": 28.033,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 2.83,
|
| 84 |
+
"eval_runtime": 1.9188,
|
| 85 |
+
"eval_samples_per_second": 374.183,
|
| 86 |
+
"eval_steps_per_second": 11.986,
|
| 87 |
+
"step": 300
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 3.3,
|
| 91 |
+
"learning_rate": 1.2150943396226416e-05,
|
| 92 |
+
"loss": 22.0028,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 3.3,
|
| 97 |
+
"eval_runtime": 2.939,
|
| 98 |
+
"eval_samples_per_second": 244.303,
|
| 99 |
+
"eval_steps_per_second": 7.826,
|
| 100 |
+
"step": 350
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 3.77,
|
| 104 |
+
"learning_rate": 1.4037735849056604e-05,
|
| 105 |
+
"loss": 17.8244,
|
| 106 |
+
"step": 400
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 3.77,
|
| 110 |
+
"eval_runtime": 2.0678,
|
| 111 |
+
"eval_samples_per_second": 347.226,
|
| 112 |
+
"eval_steps_per_second": 11.123,
|
| 113 |
+
"step": 400
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 4.25,
|
| 117 |
+
"learning_rate": 1.5924528301886794e-05,
|
| 118 |
+
"loss": 14.5603,
|
| 119 |
+
"step": 450
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 4.25,
|
| 123 |
+
"eval_runtime": 2.0344,
|
| 124 |
+
"eval_samples_per_second": 352.926,
|
| 125 |
+
"eval_steps_per_second": 11.305,
|
| 126 |
+
"step": 450
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 4.72,
|
| 130 |
+
"learning_rate": 1.7811320754716983e-05,
|
| 131 |
+
"loss": 11.8745,
|
| 132 |
+
"step": 500
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 4.72,
|
| 136 |
+
"eval_runtime": 2.0323,
|
| 137 |
+
"eval_samples_per_second": 353.287,
|
| 138 |
+
"eval_steps_per_second": 11.317,
|
| 139 |
+
"step": 500
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 5.19,
|
| 143 |
+
"learning_rate": 1.969811320754717e-05,
|
| 144 |
+
"loss": 9.6856,
|
| 145 |
+
"step": 550
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 5.19,
|
| 149 |
+
"eval_runtime": 2.0201,
|
| 150 |
+
"eval_samples_per_second": 355.425,
|
| 151 |
+
"eval_steps_per_second": 11.385,
|
| 152 |
+
"step": 550
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 5.66,
|
| 156 |
+
"learning_rate": 1.999617436193858e-05,
|
| 157 |
+
"loss": 7.8251,
|
| 158 |
+
"step": 600
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 5.66,
|
| 162 |
+
"eval_runtime": 2.0209,
|
| 163 |
+
"eval_samples_per_second": 355.286,
|
| 164 |
+
"eval_steps_per_second": 11.381,
|
| 165 |
+
"step": 600
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 6.13,
|
| 169 |
+
"learning_rate": 1.9981648322537017e-05,
|
| 170 |
+
"loss": 6.5604,
|
| 171 |
+
"step": 650
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 6.13,
|
| 175 |
+
"eval_runtime": 1.9676,
|
| 176 |
+
"eval_samples_per_second": 364.903,
|
| 177 |
+
"eval_steps_per_second": 11.689,
|
| 178 |
+
"step": 650
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 6.6,
|
| 182 |
+
"learning_rate": 1.995629882570864e-05,
|
| 183 |
+
"loss": 5.9968,
|
| 184 |
+
"step": 700
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 6.6,
|
| 188 |
+
"eval_runtime": 1.933,
|
| 189 |
+
"eval_samples_per_second": 371.443,
|
| 190 |
+
"eval_steps_per_second": 11.899,
|
| 191 |
+
"step": 700
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 7.08,
|
| 195 |
+
"learning_rate": 1.992015335881735e-05,
|
| 196 |
+
"loss": 5.3714,
|
| 197 |
+
"step": 750
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 7.08,
|
| 201 |
+
"eval_runtime": 1.928,
|
| 202 |
+
"eval_samples_per_second": 372.4,
|
| 203 |
+
"eval_steps_per_second": 11.929,
|
| 204 |
+
"step": 750
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 7.55,
|
| 208 |
+
"learning_rate": 1.9873251115682577e-05,
|
| 209 |
+
"loss": 5.0835,
|
| 210 |
+
"step": 800
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 7.55,
|
| 214 |
+
"eval_runtime": 1.9194,
|
| 215 |
+
"eval_samples_per_second": 374.085,
|
| 216 |
+
"eval_steps_per_second": 11.983,
|
| 217 |
+
"step": 800
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 8.02,
|
| 221 |
+
"learning_rate": 1.9815642954080055e-05,
|
| 222 |
+
"loss": 4.8417,
|
| 223 |
+
"step": 850
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 8.02,
|
| 227 |
+
"eval_runtime": 1.9205,
|
| 228 |
+
"eval_samples_per_second": 373.866,
|
| 229 |
+
"eval_steps_per_second": 11.976,
|
| 230 |
+
"step": 850
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 8.49,
|
| 234 |
+
"learning_rate": 1.97473913405949e-05,
|
| 235 |
+
"loss": 4.651,
|
| 236 |
+
"step": 900
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 8.49,
|
| 240 |
+
"eval_runtime": 1.9167,
|
| 241 |
+
"eval_samples_per_second": 374.597,
|
| 242 |
+
"eval_steps_per_second": 12.0,
|
| 243 |
+
"step": 900
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 8.96,
|
| 247 |
+
"learning_rate": 1.966857028288687e-05,
|
| 248 |
+
"loss": 4.5806,
|
| 249 |
+
"step": 950
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 8.96,
|
| 253 |
+
"eval_runtime": 1.9211,
|
| 254 |
+
"eval_samples_per_second": 373.737,
|
| 255 |
+
"eval_steps_per_second": 11.972,
|
| 256 |
+
"step": 950
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 9.43,
|
| 260 |
+
"learning_rate": 1.9579265249441216e-05,
|
| 261 |
+
"loss": 4.5062,
|
| 262 |
+
"step": 1000
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 9.43,
|
| 266 |
+
"eval_runtime": 1.9281,
|
| 267 |
+
"eval_samples_per_second": 372.389,
|
| 268 |
+
"eval_steps_per_second": 11.929,
|
| 269 |
+
"step": 1000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 9.91,
|
| 273 |
+
"learning_rate": 1.9479573076892152e-05,
|
| 274 |
+
"loss": 4.4466,
|
| 275 |
+
"step": 1050
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 9.91,
|
| 279 |
+
"eval_runtime": 1.9203,
|
| 280 |
+
"eval_samples_per_second": 373.895,
|
| 281 |
+
"eval_steps_per_second": 11.977,
|
| 282 |
+
"step": 1050
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 10.38,
|
| 286 |
+
"learning_rate": 1.9369601865019452e-05,
|
| 287 |
+
"loss": 4.4225,
|
| 288 |
+
"step": 1100
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 10.38,
|
| 292 |
+
"eval_runtime": 1.9182,
|
| 293 |
+
"eval_samples_per_second": 374.311,
|
| 294 |
+
"eval_steps_per_second": 11.99,
|
| 295 |
+
"step": 1100
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 10.85,
|
| 299 |
+
"learning_rate": 1.9249470859531976e-05,
|
| 300 |
+
"loss": 4.4022,
|
| 301 |
+
"step": 1150
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 10.85,
|
| 305 |
+
"eval_runtime": 1.9262,
|
| 306 |
+
"eval_samples_per_second": 372.754,
|
| 307 |
+
"eval_steps_per_second": 11.941,
|
| 308 |
+
"step": 1150
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 11.32,
|
| 312 |
+
"learning_rate": 1.9119310322765315e-05,
|
| 313 |
+
"loss": 4.3985,
|
| 314 |
+
"step": 1200
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 11.32,
|
| 318 |
+
"eval_runtime": 1.94,
|
| 319 |
+
"eval_samples_per_second": 370.112,
|
| 320 |
+
"eval_steps_per_second": 11.856,
|
| 321 |
+
"step": 1200
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 11.79,
|
| 325 |
+
"learning_rate": 1.8979261392433685e-05,
|
| 326 |
+
"loss": 4.3822,
|
| 327 |
+
"step": 1250
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 11.79,
|
| 331 |
+
"eval_runtime": 1.9393,
|
| 332 |
+
"eval_samples_per_second": 370.229,
|
| 333 |
+
"eval_steps_per_second": 11.86,
|
| 334 |
+
"step": 1250
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 12.26,
|
| 338 |
+
"learning_rate": 1.8829475928589272e-05,
|
| 339 |
+
"loss": 4.3754,
|
| 340 |
+
"step": 1300
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 12.26,
|
| 344 |
+
"eval_runtime": 1.9864,
|
| 345 |
+
"eval_samples_per_second": 361.455,
|
| 346 |
+
"eval_steps_per_second": 11.579,
|
| 347 |
+
"step": 1300
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 12.74,
|
| 351 |
+
"learning_rate": 1.8670116348954945e-05,
|
| 352 |
+
"loss": 4.3684,
|
| 353 |
+
"step": 1350
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 12.74,
|
| 357 |
+
"eval_runtime": 1.9783,
|
| 358 |
+
"eval_samples_per_second": 362.937,
|
| 359 |
+
"eval_steps_per_second": 11.626,
|
| 360 |
+
"step": 1350
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 13.21,
|
| 364 |
+
"learning_rate": 1.850135545280894e-05,
|
| 365 |
+
"loss": 4.3725,
|
| 366 |
+
"step": 1400
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 13.21,
|
| 370 |
+
"eval_runtime": 1.9796,
|
| 371 |
+
"eval_samples_per_second": 362.707,
|
| 372 |
+
"eval_steps_per_second": 11.619,
|
| 373 |
+
"step": 1400
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 13.68,
|
| 377 |
+
"learning_rate": 1.832337623361242e-05,
|
| 378 |
+
"loss": 4.3738,
|
| 379 |
+
"step": 1450
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 13.68,
|
| 383 |
+
"eval_runtime": 1.9672,
|
| 384 |
+
"eval_samples_per_second": 364.987,
|
| 385 |
+
"eval_steps_per_second": 11.692,
|
| 386 |
+
"step": 1450
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 14.15,
|
| 390 |
+
"learning_rate": 1.8136371680583176e-05,
|
| 391 |
+
"loss": 4.3524,
|
| 392 |
+
"step": 1500
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 14.15,
|
| 396 |
+
"eval_runtime": 1.917,
|
| 397 |
+
"eval_samples_per_second": 374.54,
|
| 398 |
+
"eval_steps_per_second": 11.998,
|
| 399 |
+
"step": 1500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 14.62,
|
| 403 |
+
"learning_rate": 1.7940544569430468e-05,
|
| 404 |
+
"loss": 4.3487,
|
| 405 |
+
"step": 1550
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 14.62,
|
| 409 |
+
"eval_runtime": 1.9212,
|
| 410 |
+
"eval_samples_per_second": 373.718,
|
| 411 |
+
"eval_steps_per_second": 11.971,
|
| 412 |
+
"step": 1550
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 15.09,
|
| 416 |
+
"learning_rate": 1.7736107242478143e-05,
|
| 417 |
+
"loss": 4.347,
|
| 418 |
+
"step": 1600
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 15.09,
|
| 422 |
+
"eval_runtime": 1.9221,
|
| 423 |
+
"eval_samples_per_second": 373.543,
|
| 424 |
+
"eval_steps_per_second": 11.966,
|
| 425 |
+
"step": 1600
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 15.57,
|
| 429 |
+
"learning_rate": 1.7523281378414246e-05,
|
| 430 |
+
"loss": 4.3417,
|
| 431 |
+
"step": 1650
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 15.57,
|
| 435 |
+
"eval_runtime": 2.1716,
|
| 436 |
+
"eval_samples_per_second": 330.631,
|
| 437 |
+
"eval_steps_per_second": 10.591,
|
| 438 |
+
"step": 1650
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 16.04,
|
| 442 |
+
"learning_rate": 1.730229775191693e-05,
|
| 443 |
+
"loss": 4.3224,
|
| 444 |
+
"step": 1700
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 16.04,
|
| 448 |
+
"eval_runtime": 1.9519,
|
| 449 |
+
"eval_samples_per_second": 367.842,
|
| 450 |
+
"eval_steps_per_second": 11.783,
|
| 451 |
+
"step": 1700
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 16.51,
|
| 455 |
+
"learning_rate": 1.7073395983417227e-05,
|
| 456 |
+
"loss": 4.3169,
|
| 457 |
+
"step": 1750
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 16.51,
|
| 461 |
+
"eval_runtime": 1.9446,
|
| 462 |
+
"eval_samples_per_second": 369.221,
|
| 463 |
+
"eval_steps_per_second": 11.827,
|
| 464 |
+
"step": 1750
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 16.98,
|
| 468 |
+
"learning_rate": 1.6836824279270053e-05,
|
| 469 |
+
"loss": 4.3111,
|
| 470 |
+
"step": 1800
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 16.98,
|
| 474 |
+
"eval_runtime": 1.9452,
|
| 475 |
+
"eval_samples_per_second": 369.117,
|
| 476 |
+
"eval_steps_per_second": 11.824,
|
| 477 |
+
"step": 1800
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 17.45,
|
| 481 |
+
"learning_rate": 1.6592839162615223e-05,
|
| 482 |
+
"loss": 4.3045,
|
| 483 |
+
"step": 1850
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 17.45,
|
| 487 |
+
"eval_runtime": 1.9364,
|
| 488 |
+
"eval_samples_per_second": 370.798,
|
| 489 |
+
"eval_steps_per_second": 11.878,
|
| 490 |
+
"step": 1850
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 17.92,
|
| 494 |
+
"learning_rate": 1.6341705195220257e-05,
|
| 495 |
+
"loss": 4.2946,
|
| 496 |
+
"step": 1900
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 17.92,
|
| 500 |
+
"eval_runtime": 1.9407,
|
| 501 |
+
"eval_samples_per_second": 369.973,
|
| 502 |
+
"eval_steps_per_second": 11.852,
|
| 503 |
+
"step": 1900
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 18.4,
|
| 507 |
+
"learning_rate": 1.6083694690606592e-05,
|
| 508 |
+
"loss": 4.2868,
|
| 509 |
+
"step": 1950
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 18.4,
|
| 513 |
+
"eval_runtime": 1.9401,
|
| 514 |
+
"eval_samples_per_second": 370.079,
|
| 515 |
+
"eval_steps_per_second": 11.855,
|
| 516 |
+
"step": 1950
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 18.87,
|
| 520 |
+
"learning_rate": 1.581908741877034e-05,
|
| 521 |
+
"loss": 4.2791,
|
| 522 |
+
"step": 2000
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 18.87,
|
| 526 |
+
"eval_runtime": 1.9388,
|
| 527 |
+
"eval_samples_per_second": 370.327,
|
| 528 |
+
"eval_steps_per_second": 11.863,
|
| 529 |
+
"step": 2000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 19.34,
|
| 533 |
+
"learning_rate": 1.5548170302817683e-05,
|
| 534 |
+
"loss": 4.2596,
|
| 535 |
+
"step": 2050
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 19.34,
|
| 539 |
+
"eval_runtime": 1.9504,
|
| 540 |
+
"eval_samples_per_second": 368.122,
|
| 541 |
+
"eval_steps_per_second": 11.792,
|
| 542 |
+
"step": 2050
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 19.81,
|
| 546 |
+
"learning_rate": 1.5271237107843925e-05,
|
| 547 |
+
"loss": 4.253,
|
| 548 |
+
"step": 2100
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 19.81,
|
| 552 |
+
"eval_runtime": 1.9966,
|
| 553 |
+
"eval_samples_per_second": 359.617,
|
| 554 |
+
"eval_steps_per_second": 11.52,
|
| 555 |
+
"step": 2100
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 20.28,
|
| 559 |
+
"learning_rate": 1.4988588122393497e-05,
|
| 560 |
+
"loss": 4.2556,
|
| 561 |
+
"step": 2150
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 20.28,
|
| 565 |
+
"eval_runtime": 2.0261,
|
| 566 |
+
"eval_samples_per_second": 354.367,
|
| 567 |
+
"eval_steps_per_second": 11.352,
|
| 568 |
+
"step": 2150
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 20.75,
|
| 572 |
+
"learning_rate": 1.47005298328464e-05,
|
| 573 |
+
"loss": 4.2289,
|
| 574 |
+
"step": 2200
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 20.75,
|
| 578 |
+
"eval_runtime": 2.0081,
|
| 579 |
+
"eval_samples_per_second": 357.56,
|
| 580 |
+
"eval_steps_per_second": 11.454,
|
| 581 |
+
"step": 2200
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 21.23,
|
| 585 |
+
"learning_rate": 1.4407374591084064e-05,
|
| 586 |
+
"loss": 4.211,
|
| 587 |
+
"step": 2250
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 21.23,
|
| 591 |
+
"eval_runtime": 1.925,
|
| 592 |
+
"eval_samples_per_second": 372.984,
|
| 593 |
+
"eval_steps_per_second": 11.948,
|
| 594 |
+
"step": 2250
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 21.7,
|
| 598 |
+
"learning_rate": 1.4109440275795071e-05,
|
| 599 |
+
"loss": 4.2125,
|
| 600 |
+
"step": 2300
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 21.7,
|
| 604 |
+
"eval_runtime": 1.9426,
|
| 605 |
+
"eval_samples_per_second": 369.601,
|
| 606 |
+
"eval_steps_per_second": 11.84,
|
| 607 |
+
"step": 2300
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 22.17,
|
| 611 |
+
"learning_rate": 1.3807049947787954e-05,
|
| 612 |
+
"loss": 4.1935,
|
| 613 |
+
"step": 2350
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 22.17,
|
| 617 |
+
"eval_runtime": 1.944,
|
| 618 |
+
"eval_samples_per_second": 369.342,
|
| 619 |
+
"eval_steps_per_second": 11.831,
|
| 620 |
+
"step": 2350
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 22.64,
|
| 624 |
+
"learning_rate": 1.3500531499684819e-05,
|
| 625 |
+
"loss": 4.1803,
|
| 626 |
+
"step": 2400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 22.64,
|
| 630 |
+
"eval_runtime": 1.9484,
|
| 631 |
+
"eval_samples_per_second": 368.517,
|
| 632 |
+
"eval_steps_per_second": 11.805,
|
| 633 |
+
"step": 2400
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 23.11,
|
| 637 |
+
"learning_rate": 1.3190217300375694e-05,
|
| 638 |
+
"loss": 4.173,
|
| 639 |
+
"step": 2450
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 23.11,
|
| 643 |
+
"eval_runtime": 1.9375,
|
| 644 |
+
"eval_samples_per_second": 370.574,
|
| 645 |
+
"eval_steps_per_second": 11.871,
|
| 646 |
+
"step": 2450
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 23.58,
|
| 650 |
+
"learning_rate": 1.2876443834619066e-05,
|
| 651 |
+
"loss": 4.1564,
|
| 652 |
+
"step": 2500
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 23.58,
|
| 656 |
+
"eval_runtime": 1.9431,
|
| 657 |
+
"eval_samples_per_second": 369.514,
|
| 658 |
+
"eval_steps_per_second": 11.837,
|
| 659 |
+
"step": 2500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 24.06,
|
| 663 |
+
"learning_rate": 1.2559551338179468e-05,
|
| 664 |
+
"loss": 4.1486,
|
| 665 |
+
"step": 2550
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 24.06,
|
| 669 |
+
"eval_runtime": 1.9508,
|
| 670 |
+
"eval_samples_per_second": 368.046,
|
| 671 |
+
"eval_steps_per_second": 11.79,
|
| 672 |
+
"step": 2550
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 24.53,
|
| 676 |
+
"learning_rate": 1.2239883428897687e-05,
|
| 677 |
+
"loss": 4.1223,
|
| 678 |
+
"step": 2600
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 24.53,
|
| 682 |
+
"eval_runtime": 1.9403,
|
| 683 |
+
"eval_samples_per_second": 370.043,
|
| 684 |
+
"eval_steps_per_second": 11.854,
|
| 685 |
+
"step": 2600
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 25.0,
|
| 689 |
+
"learning_rate": 1.1917786734093682e-05,
|
| 690 |
+
"loss": 4.1107,
|
| 691 |
+
"step": 2650
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 25.0,
|
| 695 |
+
"eval_runtime": 1.9435,
|
| 696 |
+
"eval_samples_per_second": 369.437,
|
| 697 |
+
"eval_steps_per_second": 11.834,
|
| 698 |
+
"step": 2650
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 25.47,
|
| 702 |
+
"learning_rate": 1.1593610514706217e-05,
|
| 703 |
+
"loss": 4.0779,
|
| 704 |
+
"step": 2700
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 25.47,
|
| 708 |
+
"eval_runtime": 1.9405,
|
| 709 |
+
"eval_samples_per_second": 370.004,
|
| 710 |
+
"eval_steps_per_second": 11.853,
|
| 711 |
+
"step": 2700
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 25.94,
|
| 715 |
+
"learning_rate": 1.1267706286576759e-05,
|
| 716 |
+
"loss": 4.0284,
|
| 717 |
+
"step": 2750
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 25.94,
|
| 721 |
+
"eval_runtime": 1.944,
|
| 722 |
+
"eval_samples_per_second": 369.342,
|
| 723 |
+
"eval_steps_per_second": 11.831,
|
| 724 |
+
"step": 2750
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 26.42,
|
| 728 |
+
"learning_rate": 1.094042743928831e-05,
|
| 729 |
+
"loss": 3.9635,
|
| 730 |
+
"step": 2800
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 26.42,
|
| 734 |
+
"eval_runtime": 3.7643,
|
| 735 |
+
"eval_samples_per_second": 190.741,
|
| 736 |
+
"eval_steps_per_second": 6.11,
|
| 737 |
+
"step": 2800
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 26.89,
|
| 741 |
+
"learning_rate": 1.0612128852972474e-05,
|
| 742 |
+
"loss": 3.9045,
|
| 743 |
+
"step": 2850
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 26.89,
|
| 747 |
+
"eval_runtime": 1.9311,
|
| 748 |
+
"eval_samples_per_second": 371.804,
|
| 749 |
+
"eval_steps_per_second": 11.91,
|
| 750 |
+
"step": 2850
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 27.36,
|
| 754 |
+
"learning_rate": 1.0283166513500267e-05,
|
| 755 |
+
"loss": 3.8146,
|
| 756 |
+
"step": 2900
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 27.36,
|
| 760 |
+
"eval_runtime": 1.9224,
|
| 761 |
+
"eval_samples_per_second": 373.485,
|
| 762 |
+
"eval_steps_per_second": 11.964,
|
| 763 |
+
"step": 2900
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 27.83,
|
| 767 |
+
"learning_rate": 9.953897126473933e-06,
|
| 768 |
+
"loss": 3.7245,
|
| 769 |
+
"step": 2950
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 27.83,
|
| 773 |
+
"eval_runtime": 1.9864,
|
| 774 |
+
"eval_samples_per_second": 361.449,
|
| 775 |
+
"eval_steps_per_second": 11.578,
|
| 776 |
+
"step": 2950
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 28.3,
|
| 780 |
+
"learning_rate": 9.624677730438344e-06,
|
| 781 |
+
"loss": 3.6334,
|
| 782 |
+
"step": 3000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 28.3,
|
| 786 |
+
"eval_runtime": 1.9535,
|
| 787 |
+
"eval_samples_per_second": 367.553,
|
| 788 |
+
"eval_steps_per_second": 11.774,
|
| 789 |
+
"step": 3000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 28.77,
|
| 793 |
+
"learning_rate": 9.295865309731342e-06,
|
| 794 |
+
"loss": 3.5344,
|
| 795 |
+
"step": 3050
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 28.77,
|
| 799 |
+
"eval_runtime": 1.9514,
|
| 800 |
+
"eval_samples_per_second": 367.939,
|
| 801 |
+
"eval_steps_per_second": 11.786,
|
| 802 |
+
"step": 3050
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 29.25,
|
| 806 |
+
"learning_rate": 8.96781640739291e-06,
|
| 807 |
+
"loss": 3.4395,
|
| 808 |
+
"step": 3100
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 29.25,
|
| 812 |
+
"eval_runtime": 1.9254,
|
| 813 |
+
"eval_samples_per_second": 372.918,
|
| 814 |
+
"eval_steps_per_second": 11.946,
|
| 815 |
+
"step": 3100
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 29.72,
|
| 819 |
+
"learning_rate": 8.64088673855282e-06,
|
| 820 |
+
"loss": 3.3242,
|
| 821 |
+
"step": 3150
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 29.72,
|
| 825 |
+
"eval_runtime": 1.9225,
|
| 826 |
+
"eval_samples_per_second": 373.464,
|
| 827 |
+
"eval_steps_per_second": 11.963,
|
| 828 |
+
"step": 3150
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 30.19,
|
| 832 |
+
"learning_rate": 8.315430804716022e-06,
|
| 833 |
+
"loss": 3.205,
|
| 834 |
+
"step": 3200
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 30.19,
|
| 838 |
+
"eval_runtime": 1.9123,
|
| 839 |
+
"eval_samples_per_second": 375.463,
|
| 840 |
+
"eval_steps_per_second": 12.027,
|
| 841 |
+
"step": 3200
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 30.66,
|
| 845 |
+
"learning_rate": 7.991801509364023e-06,
|
| 846 |
+
"loss": 3.1021,
|
| 847 |
+
"step": 3250
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 30.66,
|
| 851 |
+
"eval_runtime": 1.9257,
|
| 852 |
+
"eval_samples_per_second": 372.855,
|
| 853 |
+
"eval_steps_per_second": 11.944,
|
| 854 |
+
"step": 3250
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 31.13,
|
| 858 |
+
"learning_rate": 7.670349775289047e-06,
|
| 859 |
+
"loss": 2.9841,
|
| 860 |
+
"step": 3300
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 31.13,
|
| 864 |
+
"eval_runtime": 1.9263,
|
| 865 |
+
"eval_samples_per_second": 372.732,
|
| 866 |
+
"eval_steps_per_second": 11.94,
|
| 867 |
+
"step": 3300
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 31.6,
|
| 871 |
+
"learning_rate": 7.3514241640759175e-06,
|
| 872 |
+
"loss": 2.8855,
|
| 873 |
+
"step": 3350
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 31.6,
|
| 877 |
+
"eval_runtime": 1.9276,
|
| 878 |
+
"eval_samples_per_second": 372.486,
|
| 879 |
+
"eval_steps_per_second": 11.932,
|
| 880 |
+
"step": 3350
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 32.08,
|
| 884 |
+
"learning_rate": 7.035370498144325e-06,
|
| 885 |
+
"loss": 2.8144,
|
| 886 |
+
"step": 3400
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 32.08,
|
| 890 |
+
"eval_runtime": 1.9176,
|
| 891 |
+
"eval_samples_per_second": 374.418,
|
| 892 |
+
"eval_steps_per_second": 11.994,
|
| 893 |
+
"step": 3400
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 32.55,
|
| 897 |
+
"learning_rate": 6.722531485761199e-06,
|
| 898 |
+
"loss": 2.701,
|
| 899 |
+
"step": 3450
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 32.55,
|
| 903 |
+
"eval_runtime": 1.9236,
|
| 904 |
+
"eval_samples_per_second": 373.25,
|
| 905 |
+
"eval_steps_per_second": 11.956,
|
| 906 |
+
"step": 3450
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 33.02,
|
| 910 |
+
"learning_rate": 6.413246349429934e-06,
|
| 911 |
+
"loss": 2.615,
|
| 912 |
+
"step": 3500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 33.02,
|
| 916 |
+
"eval_runtime": 1.9223,
|
| 917 |
+
"eval_samples_per_second": 373.516,
|
| 918 |
+
"eval_steps_per_second": 11.965,
|
| 919 |
+
"step": 3500
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 33.49,
|
| 923 |
+
"learning_rate": 6.107850458059322e-06,
|
| 924 |
+
"loss": 2.5553,
|
| 925 |
+
"step": 3550
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 33.49,
|
| 929 |
+
"eval_runtime": 1.9299,
|
| 930 |
+
"eval_samples_per_second": 372.031,
|
| 931 |
+
"eval_steps_per_second": 11.917,
|
| 932 |
+
"step": 3550
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 33.96,
|
| 936 |
+
"learning_rate": 5.8066749633110675e-06,
|
| 937 |
+
"loss": 2.4601,
|
| 938 |
+
"step": 3600
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 33.96,
|
| 942 |
+
"eval_runtime": 1.9178,
|
| 943 |
+
"eval_samples_per_second": 374.392,
|
| 944 |
+
"eval_steps_per_second": 11.993,
|
| 945 |
+
"step": 3600
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 34.43,
|
| 949 |
+
"learning_rate": 5.510046440520228e-06,
|
| 950 |
+
"loss": 2.4033,
|
| 951 |
+
"step": 3650
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 34.43,
|
| 955 |
+
"eval_runtime": 1.9204,
|
| 956 |
+
"eval_samples_per_second": 373.879,
|
| 957 |
+
"eval_steps_per_second": 11.977,
|
| 958 |
+
"step": 3650
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 34.91,
|
| 962 |
+
"learning_rate": 5.218286534577938e-06,
|
| 963 |
+
"loss": 2.3601,
|
| 964 |
+
"step": 3700
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 34.91,
|
| 968 |
+
"eval_runtime": 1.913,
|
| 969 |
+
"eval_samples_per_second": 375.322,
|
| 970 |
+
"eval_steps_per_second": 12.023,
|
| 971 |
+
"step": 3700
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 35.38,
|
| 975 |
+
"learning_rate": 4.93171161116037e-06,
|
| 976 |
+
"loss": 2.3119,
|
| 977 |
+
"step": 3750
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 35.38,
|
| 981 |
+
"eval_runtime": 1.9156,
|
| 982 |
+
"eval_samples_per_second": 374.813,
|
| 983 |
+
"eval_steps_per_second": 12.007,
|
| 984 |
+
"step": 3750
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 35.85,
|
| 988 |
+
"learning_rate": 4.656198158017416e-06,
|
| 989 |
+
"loss": 2.2631,
|
| 990 |
+
"step": 3800
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 35.85,
|
| 994 |
+
"eval_runtime": 1.9287,
|
| 995 |
+
"eval_samples_per_second": 372.271,
|
| 996 |
+
"eval_steps_per_second": 11.925,
|
| 997 |
+
"step": 3800
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 36.32,
|
| 1001 |
+
"learning_rate": 4.380800513736635e-06,
|
| 1002 |
+
"loss": 2.2201,
|
| 1003 |
+
"step": 3850
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 36.32,
|
| 1007 |
+
"eval_runtime": 1.9261,
|
| 1008 |
+
"eval_samples_per_second": 372.768,
|
| 1009 |
+
"eval_steps_per_second": 11.941,
|
| 1010 |
+
"step": 3850
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 36.79,
|
| 1014 |
+
"learning_rate": 4.111495967954926e-06,
|
| 1015 |
+
"loss": 2.1818,
|
| 1016 |
+
"step": 3900
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 36.79,
|
| 1020 |
+
"eval_runtime": 1.9179,
|
| 1021 |
+
"eval_samples_per_second": 374.378,
|
| 1022 |
+
"eval_steps_per_second": 11.993,
|
| 1023 |
+
"step": 3900
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 37.26,
|
| 1027 |
+
"learning_rate": 3.848576537200217e-06,
|
| 1028 |
+
"loss": 2.1282,
|
| 1029 |
+
"step": 3950
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 37.26,
|
| 1033 |
+
"eval_runtime": 1.921,
|
| 1034 |
+
"eval_samples_per_second": 373.772,
|
| 1035 |
+
"eval_steps_per_second": 11.973,
|
| 1036 |
+
"step": 3950
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 37.74,
|
| 1040 |
+
"learning_rate": 3.5923273143923885e-06,
|
| 1041 |
+
"loss": 2.1174,
|
| 1042 |
+
"step": 4000
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 37.74,
|
| 1046 |
+
"eval_runtime": 1.9202,
|
| 1047 |
+
"eval_samples_per_second": 373.92,
|
| 1048 |
+
"eval_steps_per_second": 11.978,
|
| 1049 |
+
"step": 4000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 38.21,
|
| 1053 |
+
"learning_rate": 3.343026159706837e-06,
|
| 1054 |
+
"loss": 2.091,
|
| 1055 |
+
"step": 4050
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 38.21,
|
| 1059 |
+
"eval_runtime": 1.9212,
|
| 1060 |
+
"eval_samples_per_second": 373.727,
|
| 1061 |
+
"eval_steps_per_second": 11.972,
|
| 1062 |
+
"step": 4050
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 38.68,
|
| 1066 |
+
"learning_rate": 3.1009433992807925e-06,
|
| 1067 |
+
"loss": 2.0562,
|
| 1068 |
+
"step": 4100
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 38.68,
|
| 1072 |
+
"eval_runtime": 1.9201,
|
| 1073 |
+
"eval_samples_per_second": 373.932,
|
| 1074 |
+
"eval_steps_per_second": 11.978,
|
| 1075 |
+
"step": 4100
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 39.15,
|
| 1079 |
+
"learning_rate": 2.8709585976496825e-06,
|
| 1080 |
+
"loss": 2.0276,
|
| 1081 |
+
"step": 4150
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 39.15,
|
| 1085 |
+
"eval_runtime": 1.9126,
|
| 1086 |
+
"eval_samples_per_second": 375.405,
|
| 1087 |
+
"eval_steps_per_second": 12.025,
|
| 1088 |
+
"step": 4150
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 39.62,
|
| 1092 |
+
"learning_rate": 2.6439348660701634e-06,
|
| 1093 |
+
"loss": 2.0138,
|
| 1094 |
+
"step": 4200
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 39.62,
|
| 1098 |
+
"eval_runtime": 1.918,
|
| 1099 |
+
"eval_samples_per_second": 374.351,
|
| 1100 |
+
"eval_steps_per_second": 11.992,
|
| 1101 |
+
"step": 4200
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 40.09,
|
| 1105 |
+
"learning_rate": 2.424887578383799e-06,
|
| 1106 |
+
"loss": 1.9901,
|
| 1107 |
+
"step": 4250
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 40.09,
|
| 1111 |
+
"eval_runtime": 2.0806,
|
| 1112 |
+
"eval_samples_per_second": 345.101,
|
| 1113 |
+
"eval_steps_per_second": 11.055,
|
| 1114 |
+
"step": 4250
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 40.57,
|
| 1118 |
+
"learning_rate": 2.21405425538036e-06,
|
| 1119 |
+
"loss": 1.9743,
|
| 1120 |
+
"step": 4300
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 40.57,
|
| 1124 |
+
"eval_runtime": 2.0154,
|
| 1125 |
+
"eval_samples_per_second": 356.253,
|
| 1126 |
+
"eval_steps_per_second": 11.412,
|
| 1127 |
+
"step": 4300
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 41.04,
|
| 1131 |
+
"learning_rate": 2.011663511154628e-06,
|
| 1132 |
+
"loss": 1.9583,
|
| 1133 |
+
"step": 4350
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 41.04,
|
| 1137 |
+
"eval_runtime": 2.7415,
|
| 1138 |
+
"eval_samples_per_second": 261.903,
|
| 1139 |
+
"eval_steps_per_second": 8.39,
|
| 1140 |
+
"step": 4350
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 41.51,
|
| 1144 |
+
"learning_rate": 1.817934805211976e-06,
|
| 1145 |
+
"loss": 1.9442,
|
| 1146 |
+
"step": 4400
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 41.51,
|
| 1150 |
+
"eval_runtime": 1.9318,
|
| 1151 |
+
"eval_samples_per_second": 371.67,
|
| 1152 |
+
"eval_steps_per_second": 11.906,
|
| 1153 |
+
"step": 4400
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 41.98,
|
| 1157 |
+
"learning_rate": 1.6330782045006088e-06,
|
| 1158 |
+
"loss": 1.9376,
|
| 1159 |
+
"step": 4450
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 41.98,
|
| 1163 |
+
"eval_runtime": 1.9289,
|
| 1164 |
+
"eval_samples_per_second": 372.225,
|
| 1165 |
+
"eval_steps_per_second": 11.924,
|
| 1166 |
+
"step": 4450
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 42.45,
|
| 1170 |
+
"learning_rate": 1.457294155628457e-06,
|
| 1171 |
+
"loss": 1.919,
|
| 1172 |
+
"step": 4500
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 42.45,
|
| 1176 |
+
"eval_runtime": 1.9343,
|
| 1177 |
+
"eval_samples_per_second": 371.203,
|
| 1178 |
+
"eval_steps_per_second": 11.891,
|
| 1179 |
+
"step": 4500
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 42.92,
|
| 1183 |
+
"learning_rate": 1.2907732675117878e-06,
|
| 1184 |
+
"loss": 1.9033,
|
| 1185 |
+
"step": 4550
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 42.92,
|
| 1189 |
+
"eval_runtime": 1.9245,
|
| 1190 |
+
"eval_samples_per_second": 373.093,
|
| 1191 |
+
"eval_steps_per_second": 11.951,
|
| 1192 |
+
"step": 4550
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 43.4,
|
| 1196 |
+
"learning_rate": 1.1336961046911443e-06,
|
| 1197 |
+
"loss": 1.8907,
|
| 1198 |
+
"step": 4600
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 43.4,
|
| 1202 |
+
"eval_runtime": 1.9188,
|
| 1203 |
+
"eval_samples_per_second": 374.186,
|
| 1204 |
+
"eval_steps_per_second": 11.986,
|
| 1205 |
+
"step": 4600
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 43.87,
|
| 1209 |
+
"learning_rate": 9.862329915387669e-07,
|
| 1210 |
+
"loss": 1.8839,
|
| 1211 |
+
"step": 4650
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 43.87,
|
| 1215 |
+
"eval_runtime": 1.9284,
|
| 1216 |
+
"eval_samples_per_second": 372.325,
|
| 1217 |
+
"eval_steps_per_second": 11.927,
|
| 1218 |
+
"step": 4650
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 44.34,
|
| 1222 |
+
"learning_rate": 8.485438275698154e-07,
|
| 1223 |
+
"loss": 1.8705,
|
| 1224 |
+
"step": 4700
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 44.34,
|
| 1228 |
+
"eval_runtime": 1.9257,
|
| 1229 |
+
"eval_samples_per_second": 372.846,
|
| 1230 |
+
"eval_steps_per_second": 11.944,
|
| 1231 |
+
"step": 4700
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 44.81,
|
| 1235 |
+
"learning_rate": 7.207779140576066e-07,
|
| 1236 |
+
"loss": 1.8764,
|
| 1237 |
+
"step": 4750
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 44.81,
|
| 1241 |
+
"eval_runtime": 1.9191,
|
| 1242 |
+
"eval_samples_per_second": 374.139,
|
| 1243 |
+
"eval_steps_per_second": 11.985,
|
| 1244 |
+
"step": 4750
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 45.28,
|
| 1248 |
+
"learning_rate": 6.030737921409169e-07,
|
| 1249 |
+
"loss": 1.8671,
|
| 1250 |
+
"step": 4800
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 45.28,
|
| 1254 |
+
"eval_runtime": 1.9182,
|
| 1255 |
+
"eval_samples_per_second": 374.313,
|
| 1256 |
+
"eval_steps_per_second": 11.991,
|
| 1257 |
+
"step": 4800
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 45.75,
|
| 1261 |
+
"learning_rate": 4.955590925988896e-07,
|
| 1262 |
+
"loss": 1.8624,
|
| 1263 |
+
"step": 4850
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 45.75,
|
| 1267 |
+
"eval_runtime": 1.9234,
|
| 1268 |
+
"eval_samples_per_second": 373.294,
|
| 1269 |
+
"eval_steps_per_second": 11.958,
|
| 1270 |
+
"step": 4850
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 46.23,
|
| 1274 |
+
"learning_rate": 3.983503974564229e-07,
|
| 1275 |
+
"loss": 1.8639,
|
| 1276 |
+
"step": 4900
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 46.23,
|
| 1280 |
+
"eval_runtime": 1.9237,
|
| 1281 |
+
"eval_samples_per_second": 373.238,
|
| 1282 |
+
"eval_steps_per_second": 11.956,
|
| 1283 |
+
"step": 4900
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 46.7,
|
| 1287 |
+
"learning_rate": 3.115531135701155e-07,
|
| 1288 |
+
"loss": 1.8532,
|
| 1289 |
+
"step": 4950
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 46.7,
|
| 1293 |
+
"eval_runtime": 1.9206,
|
| 1294 |
+
"eval_samples_per_second": 373.851,
|
| 1295 |
+
"eval_steps_per_second": 11.976,
|
| 1296 |
+
"step": 4950
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 47.17,
|
| 1300 |
+
"learning_rate": 2.3526135833186527e-07,
|
| 1301 |
+
"loss": 1.849,
|
| 1302 |
+
"step": 5000
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 47.17,
|
| 1306 |
+
"eval_runtime": 1.9156,
|
| 1307 |
+
"eval_samples_per_second": 374.809,
|
| 1308 |
+
"eval_steps_per_second": 12.006,
|
| 1309 |
+
"step": 5000
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 47.64,
|
| 1313 |
+
"learning_rate": 1.6955785761400444e-07,
|
| 1314 |
+
"loss": 1.844,
|
| 1315 |
+
"step": 5050
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 47.64,
|
| 1319 |
+
"eval_runtime": 1.9207,
|
| 1320 |
+
"eval_samples_per_second": 373.812,
|
| 1321 |
+
"eval_steps_per_second": 11.974,
|
| 1322 |
+
"step": 5050
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 48.11,
|
| 1326 |
+
"learning_rate": 1.145138560667003e-07,
|
| 1327 |
+
"loss": 1.8538,
|
| 1328 |
+
"step": 5100
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 48.11,
|
| 1332 |
+
"eval_runtime": 1.9259,
|
| 1333 |
+
"eval_samples_per_second": 372.809,
|
| 1334 |
+
"eval_steps_per_second": 11.942,
|
| 1335 |
+
"step": 5100
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 48.58,
|
| 1339 |
+
"learning_rate": 7.018903986483083e-08,
|
| 1340 |
+
"loss": 1.8575,
|
| 1341 |
+
"step": 5150
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 48.58,
|
| 1345 |
+
"eval_runtime": 1.9229,
|
| 1346 |
+
"eval_samples_per_second": 373.387,
|
| 1347 |
+
"eval_steps_per_second": 11.961,
|
| 1348 |
+
"step": 5150
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 49.06,
|
| 1352 |
+
"learning_rate": 3.663147198813666e-08,
|
| 1353 |
+
"loss": 1.8503,
|
| 1354 |
+
"step": 5200
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 49.06,
|
| 1358 |
+
"eval_runtime": 1.9176,
|
| 1359 |
+
"eval_samples_per_second": 374.432,
|
| 1360 |
+
"eval_steps_per_second": 11.994,
|
| 1361 |
+
"step": 5200
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 49.53,
|
| 1365 |
+
"learning_rate": 1.3877540104818566e-08,
|
| 1366 |
+
"loss": 1.8424,
|
| 1367 |
+
"step": 5250
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 49.53,
|
| 1371 |
+
"eval_runtime": 1.9158,
|
| 1372 |
+
"eval_samples_per_second": 374.782,
|
| 1373 |
+
"eval_steps_per_second": 12.006,
|
| 1374 |
+
"step": 5250
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 50.0,
|
| 1378 |
+
"learning_rate": 1.951917115091684e-09,
|
| 1379 |
+
"loss": 1.8519,
|
| 1380 |
+
"step": 5300
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 50.0,
|
| 1384 |
+
"eval_runtime": 1.9163,
|
| 1385 |
+
"eval_samples_per_second": 374.689,
|
| 1386 |
+
"eval_steps_per_second": 12.003,
|
| 1387 |
+
"step": 5300
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 50.0,
|
| 1391 |
+
"step": 5300,
|
| 1392 |
+
"total_flos": 3.2498053856362496e+16,
|
| 1393 |
+
"train_loss": 7.9858926362811395,
|
| 1394 |
+
"train_runtime": 3481.491,
|
| 1395 |
+
"train_samples_per_second": 48.6,
|
| 1396 |
+
"train_steps_per_second": 1.522
|
| 1397 |
+
}
|
| 1398 |
+
],
|
| 1399 |
+
"logging_steps": 50,
|
| 1400 |
+
"max_steps": 5300,
|
| 1401 |
+
"num_train_epochs": 50,
|
| 1402 |
+
"save_steps": 50,
|
| 1403 |
+
"total_flos": 3.2498053856362496e+16,
|
| 1404 |
+
"trial_name": null,
|
| 1405 |
+
"trial_params": null
|
| 1406 |
+
}
|