Upload ./trainer_state.json with huggingface_hub
#7
by
sminpark
- opened
- trainer_state.json +1936 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1936 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5135612261274274,
|
| 5 |
+
"global_step": 16000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 0.00029951853635050554,
|
| 13 |
+
"loss": 1.7298,
|
| 14 |
+
"step": 50
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.0,
|
| 18 |
+
"learning_rate": 0.00029903707270101105,
|
| 19 |
+
"loss": 1.7254,
|
| 20 |
+
"step": 100
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.0,
|
| 24 |
+
"learning_rate": 0.00029855560905151657,
|
| 25 |
+
"loss": 1.7693,
|
| 26 |
+
"step": 150
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 0.00029807414540202213,
|
| 31 |
+
"loss": 1.7563,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 0.00029759268175252765,
|
| 37 |
+
"loss": 1.7625,
|
| 38 |
+
"step": 250
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.01,
|
| 42 |
+
"learning_rate": 0.0002971112181030332,
|
| 43 |
+
"loss": 1.8455,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.01,
|
| 48 |
+
"learning_rate": 0.00029662975445353873,
|
| 49 |
+
"loss": 1.5512,
|
| 50 |
+
"step": 350
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.01,
|
| 54 |
+
"learning_rate": 0.00029614829080404424,
|
| 55 |
+
"loss": 1.7412,
|
| 56 |
+
"step": 400
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.01,
|
| 60 |
+
"learning_rate": 0.0002956668271545498,
|
| 61 |
+
"loss": 1.8086,
|
| 62 |
+
"step": 450
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.02,
|
| 66 |
+
"learning_rate": 0.0002951853635050553,
|
| 67 |
+
"loss": 1.6451,
|
| 68 |
+
"step": 500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.02,
|
| 72 |
+
"learning_rate": 0.0002947038998555609,
|
| 73 |
+
"loss": 1.616,
|
| 74 |
+
"step": 550
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.02,
|
| 78 |
+
"learning_rate": 0.0002942224362060664,
|
| 79 |
+
"loss": 1.6528,
|
| 80 |
+
"step": 600
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.02,
|
| 84 |
+
"learning_rate": 0.0002937409725565719,
|
| 85 |
+
"loss": 1.69,
|
| 86 |
+
"step": 650
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.02,
|
| 90 |
+
"learning_rate": 0.0002932595089070775,
|
| 91 |
+
"loss": 1.6358,
|
| 92 |
+
"step": 700
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.02,
|
| 96 |
+
"learning_rate": 0.00029277804525758305,
|
| 97 |
+
"loss": 1.6699,
|
| 98 |
+
"step": 750
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.03,
|
| 102 |
+
"learning_rate": 0.00029229658160808856,
|
| 103 |
+
"loss": 1.7081,
|
| 104 |
+
"step": 800
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.03,
|
| 108 |
+
"learning_rate": 0.00029181511795859413,
|
| 109 |
+
"loss": 1.7176,
|
| 110 |
+
"step": 850
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.03,
|
| 114 |
+
"learning_rate": 0.00029133365430909964,
|
| 115 |
+
"loss": 1.5933,
|
| 116 |
+
"step": 900
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.03,
|
| 120 |
+
"learning_rate": 0.00029085219065960516,
|
| 121 |
+
"loss": 1.6155,
|
| 122 |
+
"step": 950
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.03,
|
| 126 |
+
"learning_rate": 0.0002903707270101107,
|
| 127 |
+
"loss": 1.7633,
|
| 128 |
+
"step": 1000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.03,
|
| 132 |
+
"learning_rate": 0.0002898892633606163,
|
| 133 |
+
"loss": 1.7233,
|
| 134 |
+
"step": 1050
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.04,
|
| 138 |
+
"learning_rate": 0.0002894077997111218,
|
| 139 |
+
"loss": 1.6779,
|
| 140 |
+
"step": 1100
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.04,
|
| 144 |
+
"learning_rate": 0.0002889263360616273,
|
| 145 |
+
"loss": 1.7086,
|
| 146 |
+
"step": 1150
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.04,
|
| 150 |
+
"learning_rate": 0.00028844487241213283,
|
| 151 |
+
"loss": 1.5796,
|
| 152 |
+
"step": 1200
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.04,
|
| 156 |
+
"learning_rate": 0.0002879634087626384,
|
| 157 |
+
"loss": 1.6772,
|
| 158 |
+
"step": 1250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.04,
|
| 162 |
+
"learning_rate": 0.00028748194511314396,
|
| 163 |
+
"loss": 1.6648,
|
| 164 |
+
"step": 1300
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.04,
|
| 168 |
+
"learning_rate": 0.0002870004814636495,
|
| 169 |
+
"loss": 1.6431,
|
| 170 |
+
"step": 1350
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.04,
|
| 174 |
+
"learning_rate": 0.000286519017814155,
|
| 175 |
+
"loss": 1.6461,
|
| 176 |
+
"step": 1400
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.05,
|
| 180 |
+
"learning_rate": 0.0002860471834376504,
|
| 181 |
+
"loss": 1.5384,
|
| 182 |
+
"step": 1450
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.05,
|
| 186 |
+
"learning_rate": 0.000285565719788156,
|
| 187 |
+
"loss": 1.5367,
|
| 188 |
+
"step": 1500
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.05,
|
| 192 |
+
"learning_rate": 0.0002850842561386615,
|
| 193 |
+
"loss": 1.7496,
|
| 194 |
+
"step": 1550
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.05,
|
| 198 |
+
"learning_rate": 0.000284602792489167,
|
| 199 |
+
"loss": 1.6623,
|
| 200 |
+
"step": 1600
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.05,
|
| 204 |
+
"learning_rate": 0.0002841213288396726,
|
| 205 |
+
"loss": 1.7164,
|
| 206 |
+
"step": 1650
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.05,
|
| 210 |
+
"learning_rate": 0.0002836398651901781,
|
| 211 |
+
"loss": 1.7689,
|
| 212 |
+
"step": 1700
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.06,
|
| 216 |
+
"learning_rate": 0.00028315840154068367,
|
| 217 |
+
"loss": 1.6634,
|
| 218 |
+
"step": 1750
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.06,
|
| 222 |
+
"learning_rate": 0.0002826769378911892,
|
| 223 |
+
"loss": 1.631,
|
| 224 |
+
"step": 1800
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.06,
|
| 228 |
+
"learning_rate": 0.0002821954742416947,
|
| 229 |
+
"loss": 1.7521,
|
| 230 |
+
"step": 1850
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.06,
|
| 234 |
+
"learning_rate": 0.00028171401059220026,
|
| 235 |
+
"loss": 1.7121,
|
| 236 |
+
"step": 1900
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.06,
|
| 240 |
+
"learning_rate": 0.00028123254694270583,
|
| 241 |
+
"loss": 1.6616,
|
| 242 |
+
"step": 1950
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.06,
|
| 246 |
+
"learning_rate": 0.00028075108329321134,
|
| 247 |
+
"loss": 1.7649,
|
| 248 |
+
"step": 2000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.07,
|
| 252 |
+
"learning_rate": 0.0002802696196437169,
|
| 253 |
+
"loss": 1.7271,
|
| 254 |
+
"step": 2050
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.07,
|
| 258 |
+
"learning_rate": 0.0002797881559942224,
|
| 259 |
+
"loss": 1.6985,
|
| 260 |
+
"step": 2100
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.07,
|
| 264 |
+
"learning_rate": 0.00027930669234472793,
|
| 265 |
+
"loss": 1.7036,
|
| 266 |
+
"step": 2150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.07,
|
| 270 |
+
"learning_rate": 0.0002788252286952335,
|
| 271 |
+
"loss": 1.7253,
|
| 272 |
+
"step": 2200
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.07,
|
| 276 |
+
"learning_rate": 0.000278343765045739,
|
| 277 |
+
"loss": 1.6715,
|
| 278 |
+
"step": 2250
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.07,
|
| 282 |
+
"learning_rate": 0.0002778623013962446,
|
| 283 |
+
"loss": 1.6576,
|
| 284 |
+
"step": 2300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.08,
|
| 288 |
+
"learning_rate": 0.0002773808377467501,
|
| 289 |
+
"loss": 1.7675,
|
| 290 |
+
"step": 2350
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.08,
|
| 294 |
+
"learning_rate": 0.0002768993740972556,
|
| 295 |
+
"loss": 1.7118,
|
| 296 |
+
"step": 2400
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.08,
|
| 300 |
+
"learning_rate": 0.0002764179104477612,
|
| 301 |
+
"loss": 1.6123,
|
| 302 |
+
"step": 2450
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.08,
|
| 306 |
+
"learning_rate": 0.00027593644679826674,
|
| 307 |
+
"loss": 1.5685,
|
| 308 |
+
"step": 2500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.08,
|
| 312 |
+
"learning_rate": 0.00027545498314877225,
|
| 313 |
+
"loss": 1.756,
|
| 314 |
+
"step": 2550
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.08,
|
| 318 |
+
"learning_rate": 0.00027497351949927777,
|
| 319 |
+
"loss": 1.7509,
|
| 320 |
+
"step": 2600
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.09,
|
| 324 |
+
"learning_rate": 0.00027449205584978334,
|
| 325 |
+
"loss": 1.691,
|
| 326 |
+
"step": 2650
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.09,
|
| 330 |
+
"learning_rate": 0.00027401059220028885,
|
| 331 |
+
"loss": 1.6535,
|
| 332 |
+
"step": 2700
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.09,
|
| 336 |
+
"learning_rate": 0.0002735291285507944,
|
| 337 |
+
"loss": 1.636,
|
| 338 |
+
"step": 2750
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.09,
|
| 342 |
+
"learning_rate": 0.00027304766490129993,
|
| 343 |
+
"loss": 1.7285,
|
| 344 |
+
"step": 2800
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.09,
|
| 348 |
+
"learning_rate": 0.00027256620125180544,
|
| 349 |
+
"loss": 1.735,
|
| 350 |
+
"step": 2850
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.09,
|
| 354 |
+
"learning_rate": 0.000272084737602311,
|
| 355 |
+
"loss": 1.6377,
|
| 356 |
+
"step": 2900
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.09,
|
| 360 |
+
"learning_rate": 0.0002716032739528165,
|
| 361 |
+
"loss": 1.5383,
|
| 362 |
+
"step": 2950
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.1,
|
| 366 |
+
"learning_rate": 0.0002711218103033221,
|
| 367 |
+
"loss": 1.6402,
|
| 368 |
+
"step": 3000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.1,
|
| 372 |
+
"learning_rate": 0.0002706403466538276,
|
| 373 |
+
"loss": 1.6001,
|
| 374 |
+
"step": 3050
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.1,
|
| 378 |
+
"learning_rate": 0.0002701588830043331,
|
| 379 |
+
"loss": 1.5789,
|
| 380 |
+
"step": 3100
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.1,
|
| 384 |
+
"learning_rate": 0.0002696774193548387,
|
| 385 |
+
"loss": 1.6518,
|
| 386 |
+
"step": 3150
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.1,
|
| 390 |
+
"learning_rate": 0.00026919595570534425,
|
| 391 |
+
"loss": 1.6663,
|
| 392 |
+
"step": 3200
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.1,
|
| 396 |
+
"learning_rate": 0.00026871449205584976,
|
| 397 |
+
"loss": 1.6436,
|
| 398 |
+
"step": 3250
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.11,
|
| 402 |
+
"learning_rate": 0.0002682330284063553,
|
| 403 |
+
"loss": 1.612,
|
| 404 |
+
"step": 3300
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.11,
|
| 408 |
+
"learning_rate": 0.00026775156475686084,
|
| 409 |
+
"loss": 1.601,
|
| 410 |
+
"step": 3350
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.11,
|
| 414 |
+
"learning_rate": 0.00026727010110736636,
|
| 415 |
+
"loss": 1.5697,
|
| 416 |
+
"step": 3400
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.11,
|
| 420 |
+
"learning_rate": 0.0002667886374578719,
|
| 421 |
+
"loss": 1.628,
|
| 422 |
+
"step": 3450
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.11,
|
| 426 |
+
"learning_rate": 0.00026630717380837744,
|
| 427 |
+
"loss": 1.673,
|
| 428 |
+
"step": 3500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.11,
|
| 432 |
+
"learning_rate": 0.00026582571015888295,
|
| 433 |
+
"loss": 1.6835,
|
| 434 |
+
"step": 3550
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.12,
|
| 438 |
+
"learning_rate": 0.0002653442465093885,
|
| 439 |
+
"loss": 1.7061,
|
| 440 |
+
"step": 3600
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.12,
|
| 444 |
+
"learning_rate": 0.00026486278285989403,
|
| 445 |
+
"loss": 1.5947,
|
| 446 |
+
"step": 3650
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.12,
|
| 450 |
+
"learning_rate": 0.0002643813192103996,
|
| 451 |
+
"loss": 1.7096,
|
| 452 |
+
"step": 3700
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.12,
|
| 456 |
+
"learning_rate": 0.00026389985556090516,
|
| 457 |
+
"loss": 1.6038,
|
| 458 |
+
"step": 3750
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.12,
|
| 462 |
+
"learning_rate": 0.00026342802118440055,
|
| 463 |
+
"loss": 1.7822,
|
| 464 |
+
"step": 3800
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.12,
|
| 468 |
+
"learning_rate": 0.00026294655753490606,
|
| 469 |
+
"loss": 1.639,
|
| 470 |
+
"step": 3850
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.13,
|
| 474 |
+
"learning_rate": 0.0002624650938854116,
|
| 475 |
+
"loss": 1.6256,
|
| 476 |
+
"step": 3900
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.13,
|
| 480 |
+
"learning_rate": 0.0002619836302359172,
|
| 481 |
+
"loss": 1.5954,
|
| 482 |
+
"step": 3950
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.13,
|
| 486 |
+
"learning_rate": 0.0002615021665864227,
|
| 487 |
+
"loss": 1.6369,
|
| 488 |
+
"step": 4000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.13,
|
| 492 |
+
"learning_rate": 0.0002610207029369282,
|
| 493 |
+
"loss": 1.7314,
|
| 494 |
+
"step": 4050
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.13,
|
| 498 |
+
"learning_rate": 0.0002605392392874338,
|
| 499 |
+
"loss": 1.6665,
|
| 500 |
+
"step": 4100
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.13,
|
| 504 |
+
"learning_rate": 0.0002600577756379393,
|
| 505 |
+
"loss": 1.6713,
|
| 506 |
+
"step": 4150
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.13,
|
| 510 |
+
"learning_rate": 0.00025957631198844487,
|
| 511 |
+
"loss": 1.569,
|
| 512 |
+
"step": 4200
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.14,
|
| 516 |
+
"learning_rate": 0.0002590948483389504,
|
| 517 |
+
"loss": 1.7352,
|
| 518 |
+
"step": 4250
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.14,
|
| 522 |
+
"learning_rate": 0.0002586133846894559,
|
| 523 |
+
"loss": 1.6418,
|
| 524 |
+
"step": 4300
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.14,
|
| 528 |
+
"learning_rate": 0.00025813192103996146,
|
| 529 |
+
"loss": 1.6525,
|
| 530 |
+
"step": 4350
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.14,
|
| 534 |
+
"learning_rate": 0.00025765045739046703,
|
| 535 |
+
"loss": 1.6305,
|
| 536 |
+
"step": 4400
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.14,
|
| 540 |
+
"learning_rate": 0.00025716899374097254,
|
| 541 |
+
"loss": 1.6631,
|
| 542 |
+
"step": 4450
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.14,
|
| 546 |
+
"learning_rate": 0.00025668753009147805,
|
| 547 |
+
"loss": 1.7426,
|
| 548 |
+
"step": 4500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.15,
|
| 552 |
+
"learning_rate": 0.00025620606644198357,
|
| 553 |
+
"loss": 1.6466,
|
| 554 |
+
"step": 4550
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.15,
|
| 558 |
+
"learning_rate": 0.00025572460279248913,
|
| 559 |
+
"loss": 1.6907,
|
| 560 |
+
"step": 4600
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.15,
|
| 564 |
+
"learning_rate": 0.0002552431391429947,
|
| 565 |
+
"loss": 1.7573,
|
| 566 |
+
"step": 4650
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.15,
|
| 570 |
+
"learning_rate": 0.0002547616754935002,
|
| 571 |
+
"loss": 1.7466,
|
| 572 |
+
"step": 4700
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.15,
|
| 576 |
+
"learning_rate": 0.00025428021184400573,
|
| 577 |
+
"loss": 1.5971,
|
| 578 |
+
"step": 4750
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.15,
|
| 582 |
+
"learning_rate": 0.0002537987481945113,
|
| 583 |
+
"loss": 1.8605,
|
| 584 |
+
"step": 4800
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.16,
|
| 588 |
+
"learning_rate": 0.0002533172845450168,
|
| 589 |
+
"loss": 1.7469,
|
| 590 |
+
"step": 4850
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.16,
|
| 594 |
+
"learning_rate": 0.0002528358208955224,
|
| 595 |
+
"loss": 1.5872,
|
| 596 |
+
"step": 4900
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.16,
|
| 600 |
+
"learning_rate": 0.00025235435724602794,
|
| 601 |
+
"loss": 1.7249,
|
| 602 |
+
"step": 4950
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.16,
|
| 606 |
+
"learning_rate": 0.00025187289359653346,
|
| 607 |
+
"loss": 1.5394,
|
| 608 |
+
"step": 5000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.16,
|
| 612 |
+
"learning_rate": 0.00025139142994703897,
|
| 613 |
+
"loss": 1.6919,
|
| 614 |
+
"step": 5050
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.16,
|
| 618 |
+
"learning_rate": 0.00025090996629754454,
|
| 619 |
+
"loss": 1.6369,
|
| 620 |
+
"step": 5100
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.17,
|
| 624 |
+
"learning_rate": 0.00025042850264805005,
|
| 625 |
+
"loss": 1.5953,
|
| 626 |
+
"step": 5150
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.17,
|
| 630 |
+
"learning_rate": 0.0002499470389985556,
|
| 631 |
+
"loss": 1.6929,
|
| 632 |
+
"step": 5200
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.17,
|
| 636 |
+
"learning_rate": 0.00024946557534906113,
|
| 637 |
+
"loss": 1.6558,
|
| 638 |
+
"step": 5250
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.17,
|
| 642 |
+
"learning_rate": 0.00024898411169956664,
|
| 643 |
+
"loss": 1.6672,
|
| 644 |
+
"step": 5300
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.17,
|
| 648 |
+
"learning_rate": 0.0002485026480500722,
|
| 649 |
+
"loss": 1.7025,
|
| 650 |
+
"step": 5350
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.17,
|
| 654 |
+
"learning_rate": 0.0002480211844005777,
|
| 655 |
+
"loss": 1.6714,
|
| 656 |
+
"step": 5400
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.17,
|
| 660 |
+
"learning_rate": 0.0002475397207510833,
|
| 661 |
+
"loss": 1.6522,
|
| 662 |
+
"step": 5450
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.18,
|
| 666 |
+
"learning_rate": 0.0002470582571015888,
|
| 667 |
+
"loss": 1.5868,
|
| 668 |
+
"step": 5500
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.18,
|
| 672 |
+
"learning_rate": 0.0002465767934520943,
|
| 673 |
+
"loss": 1.5421,
|
| 674 |
+
"step": 5550
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.18,
|
| 678 |
+
"learning_rate": 0.0002460953298025999,
|
| 679 |
+
"loss": 1.8258,
|
| 680 |
+
"step": 5600
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.18,
|
| 684 |
+
"learning_rate": 0.00024561386615310545,
|
| 685 |
+
"loss": 1.7433,
|
| 686 |
+
"step": 5650
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.18,
|
| 690 |
+
"learning_rate": 0.00024513240250361096,
|
| 691 |
+
"loss": 1.6886,
|
| 692 |
+
"step": 5700
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.18,
|
| 696 |
+
"learning_rate": 0.0002446509388541165,
|
| 697 |
+
"loss": 1.6047,
|
| 698 |
+
"step": 5750
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.19,
|
| 702 |
+
"learning_rate": 0.00024416947520462204,
|
| 703 |
+
"loss": 1.6474,
|
| 704 |
+
"step": 5800
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.19,
|
| 708 |
+
"learning_rate": 0.00024368801155512756,
|
| 709 |
+
"loss": 1.6839,
|
| 710 |
+
"step": 5850
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.19,
|
| 714 |
+
"learning_rate": 0.0002432065479056331,
|
| 715 |
+
"loss": 1.6664,
|
| 716 |
+
"step": 5900
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.19,
|
| 720 |
+
"learning_rate": 0.00024272508425613864,
|
| 721 |
+
"loss": 1.545,
|
| 722 |
+
"step": 5950
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.19,
|
| 726 |
+
"learning_rate": 0.00024224362060664418,
|
| 727 |
+
"loss": 1.7395,
|
| 728 |
+
"step": 6000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.19,
|
| 732 |
+
"learning_rate": 0.00024176215695714972,
|
| 733 |
+
"loss": 1.6451,
|
| 734 |
+
"step": 6050
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.2,
|
| 738 |
+
"learning_rate": 0.00024128069330765523,
|
| 739 |
+
"loss": 1.6876,
|
| 740 |
+
"step": 6100
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.2,
|
| 744 |
+
"learning_rate": 0.0002407992296581608,
|
| 745 |
+
"loss": 1.6001,
|
| 746 |
+
"step": 6150
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.2,
|
| 750 |
+
"learning_rate": 0.00024031776600866634,
|
| 751 |
+
"loss": 1.6769,
|
| 752 |
+
"step": 6200
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.2,
|
| 756 |
+
"learning_rate": 0.00023983630235917185,
|
| 757 |
+
"loss": 1.7025,
|
| 758 |
+
"step": 6250
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.2,
|
| 762 |
+
"learning_rate": 0.00023936446798266726,
|
| 763 |
+
"loss": 1.6464,
|
| 764 |
+
"step": 6300
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.2,
|
| 768 |
+
"learning_rate": 0.00023888300433317283,
|
| 769 |
+
"loss": 1.5731,
|
| 770 |
+
"step": 6350
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.21,
|
| 774 |
+
"learning_rate": 0.00023840154068367837,
|
| 775 |
+
"loss": 1.7313,
|
| 776 |
+
"step": 6400
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.21,
|
| 780 |
+
"learning_rate": 0.00023792007703418388,
|
| 781 |
+
"loss": 1.6279,
|
| 782 |
+
"step": 6450
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.21,
|
| 786 |
+
"learning_rate": 0.00023743861338468945,
|
| 787 |
+
"loss": 1.6481,
|
| 788 |
+
"step": 6500
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.21,
|
| 792 |
+
"learning_rate": 0.000236957149735195,
|
| 793 |
+
"loss": 1.5608,
|
| 794 |
+
"step": 6550
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.21,
|
| 798 |
+
"learning_rate": 0.0002364756860857005,
|
| 799 |
+
"loss": 1.6423,
|
| 800 |
+
"step": 6600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.21,
|
| 804 |
+
"learning_rate": 0.00023599422243620604,
|
| 805 |
+
"loss": 1.7948,
|
| 806 |
+
"step": 6650
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.22,
|
| 810 |
+
"learning_rate": 0.0002355127587867116,
|
| 811 |
+
"loss": 1.8034,
|
| 812 |
+
"step": 6700
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.22,
|
| 816 |
+
"learning_rate": 0.00023503129513721712,
|
| 817 |
+
"loss": 1.6419,
|
| 818 |
+
"step": 6750
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.22,
|
| 822 |
+
"learning_rate": 0.00023454983148772266,
|
| 823 |
+
"loss": 1.7487,
|
| 824 |
+
"step": 6800
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.22,
|
| 828 |
+
"learning_rate": 0.00023406836783822818,
|
| 829 |
+
"loss": 1.6582,
|
| 830 |
+
"step": 6850
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.22,
|
| 834 |
+
"learning_rate": 0.00023358690418873372,
|
| 835 |
+
"loss": 1.6091,
|
| 836 |
+
"step": 6900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.22,
|
| 840 |
+
"learning_rate": 0.00023310544053923928,
|
| 841 |
+
"loss": 1.612,
|
| 842 |
+
"step": 6950
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.22,
|
| 846 |
+
"learning_rate": 0.0002326239768897448,
|
| 847 |
+
"loss": 1.6118,
|
| 848 |
+
"step": 7000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.23,
|
| 852 |
+
"learning_rate": 0.00023214251324025034,
|
| 853 |
+
"loss": 1.7555,
|
| 854 |
+
"step": 7050
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.23,
|
| 858 |
+
"learning_rate": 0.00023166104959075588,
|
| 859 |
+
"loss": 1.6902,
|
| 860 |
+
"step": 7100
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.23,
|
| 864 |
+
"learning_rate": 0.00023117958594126142,
|
| 865 |
+
"loss": 1.7443,
|
| 866 |
+
"step": 7150
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.23,
|
| 870 |
+
"learning_rate": 0.00023069812229176696,
|
| 871 |
+
"loss": 1.6269,
|
| 872 |
+
"step": 7200
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.23,
|
| 876 |
+
"learning_rate": 0.00023022628791526237,
|
| 877 |
+
"loss": 1.7843,
|
| 878 |
+
"step": 7250
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.23,
|
| 882 |
+
"learning_rate": 0.00022974482426576793,
|
| 883 |
+
"loss": 1.7436,
|
| 884 |
+
"step": 7300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.24,
|
| 888 |
+
"learning_rate": 0.00022926336061627345,
|
| 889 |
+
"loss": 1.6917,
|
| 890 |
+
"step": 7350
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.24,
|
| 894 |
+
"learning_rate": 0.00022878189696677899,
|
| 895 |
+
"loss": 1.6136,
|
| 896 |
+
"step": 7400
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.24,
|
| 900 |
+
"learning_rate": 0.00022830043331728453,
|
| 901 |
+
"loss": 1.6501,
|
| 902 |
+
"step": 7450
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.24,
|
| 906 |
+
"learning_rate": 0.00022781896966779004,
|
| 907 |
+
"loss": 1.5681,
|
| 908 |
+
"step": 7500
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.24,
|
| 912 |
+
"learning_rate": 0.0002273375060182956,
|
| 913 |
+
"loss": 1.7366,
|
| 914 |
+
"step": 7550
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.24,
|
| 918 |
+
"learning_rate": 0.00022685604236880115,
|
| 919 |
+
"loss": 1.6177,
|
| 920 |
+
"step": 7600
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.25,
|
| 924 |
+
"learning_rate": 0.00022637457871930666,
|
| 925 |
+
"loss": 1.7367,
|
| 926 |
+
"step": 7650
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.25,
|
| 930 |
+
"learning_rate": 0.00022589311506981223,
|
| 931 |
+
"loss": 1.5551,
|
| 932 |
+
"step": 7700
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.25,
|
| 936 |
+
"learning_rate": 0.00022541165142031777,
|
| 937 |
+
"loss": 1.6179,
|
| 938 |
+
"step": 7750
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.25,
|
| 942 |
+
"learning_rate": 0.00022493018777082328,
|
| 943 |
+
"loss": 1.65,
|
| 944 |
+
"step": 7800
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.25,
|
| 948 |
+
"learning_rate": 0.00022444872412132882,
|
| 949 |
+
"loss": 1.7241,
|
| 950 |
+
"step": 7850
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.25,
|
| 954 |
+
"learning_rate": 0.00022396726047183433,
|
| 955 |
+
"loss": 1.6435,
|
| 956 |
+
"step": 7900
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.26,
|
| 960 |
+
"learning_rate": 0.0002234857968223399,
|
| 961 |
+
"loss": 1.7284,
|
| 962 |
+
"step": 7950
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.26,
|
| 966 |
+
"learning_rate": 0.00022300433317284544,
|
| 967 |
+
"loss": 1.6536,
|
| 968 |
+
"step": 8000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.26,
|
| 972 |
+
"learning_rate": 0.00022252286952335095,
|
| 973 |
+
"loss": 1.632,
|
| 974 |
+
"step": 8050
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.26,
|
| 978 |
+
"learning_rate": 0.0002220414058738565,
|
| 979 |
+
"loss": 1.5747,
|
| 980 |
+
"step": 8100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.26,
|
| 984 |
+
"learning_rate": 0.00022155994222436206,
|
| 985 |
+
"loss": 1.775,
|
| 986 |
+
"step": 8150
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.26,
|
| 990 |
+
"learning_rate": 0.00022107847857486757,
|
| 991 |
+
"loss": 1.7017,
|
| 992 |
+
"step": 8200
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.26,
|
| 996 |
+
"learning_rate": 0.00022059701492537311,
|
| 997 |
+
"loss": 1.7259,
|
| 998 |
+
"step": 8250
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.27,
|
| 1002 |
+
"learning_rate": 0.00022011555127587865,
|
| 1003 |
+
"loss": 1.682,
|
| 1004 |
+
"step": 8300
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.27,
|
| 1008 |
+
"learning_rate": 0.00021963408762638417,
|
| 1009 |
+
"loss": 1.6752,
|
| 1010 |
+
"step": 8350
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.27,
|
| 1014 |
+
"learning_rate": 0.00021915262397688973,
|
| 1015 |
+
"loss": 1.7571,
|
| 1016 |
+
"step": 8400
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.27,
|
| 1020 |
+
"learning_rate": 0.00021867116032739527,
|
| 1021 |
+
"loss": 1.7478,
|
| 1022 |
+
"step": 8450
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.27,
|
| 1026 |
+
"learning_rate": 0.0002181896966779008,
|
| 1027 |
+
"loss": 1.672,
|
| 1028 |
+
"step": 8500
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.27,
|
| 1032 |
+
"learning_rate": 0.00021770823302840636,
|
| 1033 |
+
"loss": 1.7433,
|
| 1034 |
+
"step": 8550
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.28,
|
| 1038 |
+
"learning_rate": 0.00021722676937891187,
|
| 1039 |
+
"loss": 1.6447,
|
| 1040 |
+
"step": 8600
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.28,
|
| 1044 |
+
"learning_rate": 0.0002167453057294174,
|
| 1045 |
+
"loss": 1.6748,
|
| 1046 |
+
"step": 8650
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.28,
|
| 1050 |
+
"learning_rate": 0.00021626384207992295,
|
| 1051 |
+
"loss": 1.6528,
|
| 1052 |
+
"step": 8700
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.28,
|
| 1056 |
+
"learning_rate": 0.00021578237843042846,
|
| 1057 |
+
"loss": 1.6351,
|
| 1058 |
+
"step": 8750
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.28,
|
| 1062 |
+
"learning_rate": 0.00021530091478093403,
|
| 1063 |
+
"loss": 1.6457,
|
| 1064 |
+
"step": 8800
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.28,
|
| 1068 |
+
"learning_rate": 0.00021481945113143957,
|
| 1069 |
+
"loss": 1.7337,
|
| 1070 |
+
"step": 8850
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.29,
|
| 1074 |
+
"learning_rate": 0.00021433798748194508,
|
| 1075 |
+
"loss": 1.6852,
|
| 1076 |
+
"step": 8900
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.29,
|
| 1080 |
+
"learning_rate": 0.00021385652383245062,
|
| 1081 |
+
"loss": 1.7768,
|
| 1082 |
+
"step": 8950
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.29,
|
| 1086 |
+
"learning_rate": 0.0002133750601829562,
|
| 1087 |
+
"loss": 1.7785,
|
| 1088 |
+
"step": 9000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.29,
|
| 1092 |
+
"learning_rate": 0.0002128935965334617,
|
| 1093 |
+
"loss": 1.6819,
|
| 1094 |
+
"step": 9050
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.29,
|
| 1098 |
+
"learning_rate": 0.00021241213288396724,
|
| 1099 |
+
"loss": 1.6261,
|
| 1100 |
+
"step": 9100
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.29,
|
| 1104 |
+
"learning_rate": 0.00021193066923447278,
|
| 1105 |
+
"loss": 1.6154,
|
| 1106 |
+
"step": 9150
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.3,
|
| 1110 |
+
"learning_rate": 0.00021144920558497832,
|
| 1111 |
+
"loss": 1.6968,
|
| 1112 |
+
"step": 9200
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.3,
|
| 1116 |
+
"learning_rate": 0.00021096774193548386,
|
| 1117 |
+
"loss": 1.6039,
|
| 1118 |
+
"step": 9250
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.3,
|
| 1122 |
+
"learning_rate": 0.00021048627828598938,
|
| 1123 |
+
"loss": 1.6625,
|
| 1124 |
+
"step": 9300
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.3,
|
| 1128 |
+
"learning_rate": 0.00021000481463649492,
|
| 1129 |
+
"loss": 1.7781,
|
| 1130 |
+
"step": 9350
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.3,
|
| 1134 |
+
"learning_rate": 0.00020952335098700048,
|
| 1135 |
+
"loss": 1.6309,
|
| 1136 |
+
"step": 9400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.3,
|
| 1140 |
+
"learning_rate": 0.000209041887337506,
|
| 1141 |
+
"loss": 1.6182,
|
| 1142 |
+
"step": 9450
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.3,
|
| 1146 |
+
"learning_rate": 0.00020856042368801154,
|
| 1147 |
+
"loss": 1.6403,
|
| 1148 |
+
"step": 9500
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.31,
|
| 1152 |
+
"learning_rate": 0.00020807896003851708,
|
| 1153 |
+
"loss": 1.6455,
|
| 1154 |
+
"step": 9550
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.31,
|
| 1158 |
+
"learning_rate": 0.0002075974963890226,
|
| 1159 |
+
"loss": 1.5998,
|
| 1160 |
+
"step": 9600
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.31,
|
| 1164 |
+
"learning_rate": 0.00020711603273952816,
|
| 1165 |
+
"loss": 1.6707,
|
| 1166 |
+
"step": 9650
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.31,
|
| 1170 |
+
"learning_rate": 0.0002066345690900337,
|
| 1171 |
+
"loss": 1.6421,
|
| 1172 |
+
"step": 9700
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.31,
|
| 1176 |
+
"learning_rate": 0.0002061531054405392,
|
| 1177 |
+
"loss": 1.5839,
|
| 1178 |
+
"step": 9750
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.31,
|
| 1182 |
+
"learning_rate": 0.00020567164179104475,
|
| 1183 |
+
"loss": 1.6987,
|
| 1184 |
+
"step": 9800
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.32,
|
| 1188 |
+
"learning_rate": 0.00020519017814155032,
|
| 1189 |
+
"loss": 1.7145,
|
| 1190 |
+
"step": 9850
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.32,
|
| 1194 |
+
"learning_rate": 0.00020470871449205583,
|
| 1195 |
+
"loss": 1.5953,
|
| 1196 |
+
"step": 9900
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.32,
|
| 1200 |
+
"learning_rate": 0.00020422725084256137,
|
| 1201 |
+
"loss": 1.624,
|
| 1202 |
+
"step": 9950
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.32,
|
| 1206 |
+
"learning_rate": 0.00020374578719306688,
|
| 1207 |
+
"loss": 1.628,
|
| 1208 |
+
"step": 10000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.32,
|
| 1212 |
+
"learning_rate": 0.00020326432354357245,
|
| 1213 |
+
"loss": 1.7215,
|
| 1214 |
+
"step": 10050
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.32,
|
| 1218 |
+
"learning_rate": 0.000202782859894078,
|
| 1219 |
+
"loss": 1.7618,
|
| 1220 |
+
"step": 10100
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.33,
|
| 1224 |
+
"learning_rate": 0.0002023013962445835,
|
| 1225 |
+
"loss": 1.6385,
|
| 1226 |
+
"step": 10150
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.33,
|
| 1230 |
+
"learning_rate": 0.00020181993259508904,
|
| 1231 |
+
"loss": 1.6758,
|
| 1232 |
+
"step": 10200
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.33,
|
| 1236 |
+
"learning_rate": 0.0002013384689455946,
|
| 1237 |
+
"loss": 1.6555,
|
| 1238 |
+
"step": 10250
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.33,
|
| 1242 |
+
"learning_rate": 0.00020085700529610012,
|
| 1243 |
+
"loss": 1.7235,
|
| 1244 |
+
"step": 10300
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.33,
|
| 1248 |
+
"learning_rate": 0.00020037554164660566,
|
| 1249 |
+
"loss": 1.7544,
|
| 1250 |
+
"step": 10350
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.33,
|
| 1254 |
+
"learning_rate": 0.0001998940779971112,
|
| 1255 |
+
"loss": 1.7105,
|
| 1256 |
+
"step": 10400
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.34,
|
| 1260 |
+
"learning_rate": 0.00019941261434761672,
|
| 1261 |
+
"loss": 1.6541,
|
| 1262 |
+
"step": 10450
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.34,
|
| 1266 |
+
"learning_rate": 0.00019893115069812229,
|
| 1267 |
+
"loss": 1.5443,
|
| 1268 |
+
"step": 10500
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.34,
|
| 1272 |
+
"learning_rate": 0.00019844968704862783,
|
| 1273 |
+
"loss": 1.6808,
|
| 1274 |
+
"step": 10550
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.34,
|
| 1278 |
+
"learning_rate": 0.00019796822339913334,
|
| 1279 |
+
"loss": 1.787,
|
| 1280 |
+
"step": 10600
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.34,
|
| 1284 |
+
"learning_rate": 0.00019748675974963888,
|
| 1285 |
+
"loss": 1.6119,
|
| 1286 |
+
"step": 10650
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.34,
|
| 1290 |
+
"learning_rate": 0.0001970052961001444,
|
| 1291 |
+
"loss": 1.6405,
|
| 1292 |
+
"step": 10700
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.35,
|
| 1296 |
+
"learning_rate": 0.00019652383245064996,
|
| 1297 |
+
"loss": 1.6875,
|
| 1298 |
+
"step": 10750
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.35,
|
| 1302 |
+
"learning_rate": 0.0001960423688011555,
|
| 1303 |
+
"loss": 1.7438,
|
| 1304 |
+
"step": 10800
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.35,
|
| 1308 |
+
"learning_rate": 0.000195560905151661,
|
| 1309 |
+
"loss": 1.7505,
|
| 1310 |
+
"step": 10850
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.35,
|
| 1314 |
+
"learning_rate": 0.00019507944150216658,
|
| 1315 |
+
"loss": 1.6726,
|
| 1316 |
+
"step": 10900
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 0.35,
|
| 1320 |
+
"learning_rate": 0.00019459797785267212,
|
| 1321 |
+
"loss": 1.5748,
|
| 1322 |
+
"step": 10950
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.35,
|
| 1326 |
+
"learning_rate": 0.00019411651420317763,
|
| 1327 |
+
"loss": 1.7293,
|
| 1328 |
+
"step": 11000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.35,
|
| 1332 |
+
"learning_rate": 0.00019363505055368317,
|
| 1333 |
+
"loss": 1.5851,
|
| 1334 |
+
"step": 11050
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.36,
|
| 1338 |
+
"learning_rate": 0.00019315358690418874,
|
| 1339 |
+
"loss": 1.6417,
|
| 1340 |
+
"step": 11100
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.36,
|
| 1344 |
+
"learning_rate": 0.00019267212325469425,
|
| 1345 |
+
"loss": 1.6152,
|
| 1346 |
+
"step": 11150
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.36,
|
| 1350 |
+
"learning_rate": 0.0001921906596051998,
|
| 1351 |
+
"loss": 1.7118,
|
| 1352 |
+
"step": 11200
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.36,
|
| 1356 |
+
"learning_rate": 0.00019170919595570533,
|
| 1357 |
+
"loss": 1.6621,
|
| 1358 |
+
"step": 11250
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.36,
|
| 1362 |
+
"learning_rate": 0.00019123736157920077,
|
| 1363 |
+
"loss": 1.6811,
|
| 1364 |
+
"step": 11300
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.36,
|
| 1368 |
+
"learning_rate": 0.00019075589792970628,
|
| 1369 |
+
"loss": 1.6491,
|
| 1370 |
+
"step": 11350
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.37,
|
| 1374 |
+
"learning_rate": 0.00019027443428021182,
|
| 1375 |
+
"loss": 1.6998,
|
| 1376 |
+
"step": 11400
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.37,
|
| 1380 |
+
"learning_rate": 0.0001897929706307174,
|
| 1381 |
+
"loss": 1.6629,
|
| 1382 |
+
"step": 11450
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.37,
|
| 1386 |
+
"learning_rate": 0.0001893115069812229,
|
| 1387 |
+
"loss": 1.6979,
|
| 1388 |
+
"step": 11500
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.37,
|
| 1392 |
+
"learning_rate": 0.00018883004333172844,
|
| 1393 |
+
"loss": 1.5112,
|
| 1394 |
+
"step": 11550
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.37,
|
| 1398 |
+
"learning_rate": 0.00018834857968223398,
|
| 1399 |
+
"loss": 1.6527,
|
| 1400 |
+
"step": 11600
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 0.37,
|
| 1404 |
+
"learning_rate": 0.0001878671160327395,
|
| 1405 |
+
"loss": 1.7075,
|
| 1406 |
+
"step": 11650
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.38,
|
| 1410 |
+
"learning_rate": 0.00018738565238324506,
|
| 1411 |
+
"loss": 1.7035,
|
| 1412 |
+
"step": 11700
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.38,
|
| 1416 |
+
"learning_rate": 0.00018690418873375058,
|
| 1417 |
+
"loss": 1.5756,
|
| 1418 |
+
"step": 11750
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.38,
|
| 1422 |
+
"learning_rate": 0.00018642272508425612,
|
| 1423 |
+
"loss": 1.5457,
|
| 1424 |
+
"step": 11800
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.38,
|
| 1428 |
+
"learning_rate": 0.00018594126143476166,
|
| 1429 |
+
"loss": 1.7155,
|
| 1430 |
+
"step": 11850
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.38,
|
| 1434 |
+
"learning_rate": 0.00018545979778526717,
|
| 1435 |
+
"loss": 1.666,
|
| 1436 |
+
"step": 11900
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.38,
|
| 1440 |
+
"learning_rate": 0.00018497833413577274,
|
| 1441 |
+
"loss": 1.6213,
|
| 1442 |
+
"step": 11950
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.39,
|
| 1446 |
+
"learning_rate": 0.00018449687048627828,
|
| 1447 |
+
"loss": 1.6945,
|
| 1448 |
+
"step": 12000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.39,
|
| 1452 |
+
"learning_rate": 0.0001840154068367838,
|
| 1453 |
+
"loss": 1.5999,
|
| 1454 |
+
"step": 12050
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.39,
|
| 1458 |
+
"learning_rate": 0.00018353394318728936,
|
| 1459 |
+
"loss": 1.6824,
|
| 1460 |
+
"step": 12100
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.39,
|
| 1464 |
+
"learning_rate": 0.0001830524795377949,
|
| 1465 |
+
"loss": 1.738,
|
| 1466 |
+
"step": 12150
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.39,
|
| 1470 |
+
"learning_rate": 0.0001825710158883004,
|
| 1471 |
+
"loss": 1.6992,
|
| 1472 |
+
"step": 12200
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.39,
|
| 1476 |
+
"learning_rate": 0.00018208955223880595,
|
| 1477 |
+
"loss": 1.6485,
|
| 1478 |
+
"step": 12250
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.39,
|
| 1482 |
+
"learning_rate": 0.00018160808858931152,
|
| 1483 |
+
"loss": 1.5837,
|
| 1484 |
+
"step": 12300
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 0.4,
|
| 1488 |
+
"learning_rate": 0.00018112662493981703,
|
| 1489 |
+
"loss": 1.7708,
|
| 1490 |
+
"step": 12350
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 0.4,
|
| 1494 |
+
"learning_rate": 0.00018064516129032257,
|
| 1495 |
+
"loss": 1.7078,
|
| 1496 |
+
"step": 12400
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 0.4,
|
| 1500 |
+
"learning_rate": 0.00018016369764082808,
|
| 1501 |
+
"loss": 1.7086,
|
| 1502 |
+
"step": 12450
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.4,
|
| 1506 |
+
"learning_rate": 0.00017968223399133363,
|
| 1507 |
+
"loss": 1.7972,
|
| 1508 |
+
"step": 12500
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.4,
|
| 1512 |
+
"learning_rate": 0.0001792007703418392,
|
| 1513 |
+
"loss": 1.6357,
|
| 1514 |
+
"step": 12550
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.4,
|
| 1518 |
+
"learning_rate": 0.0001787193066923447,
|
| 1519 |
+
"loss": 1.688,
|
| 1520 |
+
"step": 12600
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 0.41,
|
| 1524 |
+
"learning_rate": 0.00017823784304285025,
|
| 1525 |
+
"loss": 1.6416,
|
| 1526 |
+
"step": 12650
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 0.41,
|
| 1530 |
+
"learning_rate": 0.00017775637939335579,
|
| 1531 |
+
"loss": 1.7331,
|
| 1532 |
+
"step": 12700
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.41,
|
| 1536 |
+
"learning_rate": 0.0001772749157438613,
|
| 1537 |
+
"loss": 1.6895,
|
| 1538 |
+
"step": 12750
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.41,
|
| 1542 |
+
"learning_rate": 0.00017679345209436687,
|
| 1543 |
+
"loss": 1.6194,
|
| 1544 |
+
"step": 12800
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.41,
|
| 1548 |
+
"learning_rate": 0.0001763119884448724,
|
| 1549 |
+
"loss": 1.7116,
|
| 1550 |
+
"step": 12850
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.41,
|
| 1554 |
+
"learning_rate": 0.00017583052479537792,
|
| 1555 |
+
"loss": 1.692,
|
| 1556 |
+
"step": 12900
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.42,
|
| 1560 |
+
"learning_rate": 0.00017534906114588349,
|
| 1561 |
+
"loss": 1.5511,
|
| 1562 |
+
"step": 12950
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.42,
|
| 1566 |
+
"learning_rate": 0.00017486759749638903,
|
| 1567 |
+
"loss": 1.7335,
|
| 1568 |
+
"step": 13000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 0.42,
|
| 1572 |
+
"learning_rate": 0.00017438613384689454,
|
| 1573 |
+
"loss": 1.5282,
|
| 1574 |
+
"step": 13050
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.42,
|
| 1578 |
+
"learning_rate": 0.00017390467019740008,
|
| 1579 |
+
"loss": 1.5907,
|
| 1580 |
+
"step": 13100
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.42,
|
| 1584 |
+
"learning_rate": 0.0001734232065479056,
|
| 1585 |
+
"loss": 1.7614,
|
| 1586 |
+
"step": 13150
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.42,
|
| 1590 |
+
"learning_rate": 0.00017294174289841116,
|
| 1591 |
+
"loss": 1.7124,
|
| 1592 |
+
"step": 13200
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.43,
|
| 1596 |
+
"learning_rate": 0.0001724602792489167,
|
| 1597 |
+
"loss": 1.6434,
|
| 1598 |
+
"step": 13250
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.43,
|
| 1602 |
+
"learning_rate": 0.0001719788155994222,
|
| 1603 |
+
"loss": 1.6011,
|
| 1604 |
+
"step": 13300
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.43,
|
| 1608 |
+
"learning_rate": 0.00017149735194992775,
|
| 1609 |
+
"loss": 1.7427,
|
| 1610 |
+
"step": 13350
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.43,
|
| 1614 |
+
"learning_rate": 0.00017101588830043332,
|
| 1615 |
+
"loss": 1.7259,
|
| 1616 |
+
"step": 13400
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.43,
|
| 1620 |
+
"learning_rate": 0.00017053442465093883,
|
| 1621 |
+
"loss": 1.7255,
|
| 1622 |
+
"step": 13450
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.43,
|
| 1626 |
+
"learning_rate": 0.00017006259027443424,
|
| 1627 |
+
"loss": 1.6759,
|
| 1628 |
+
"step": 13500
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.43,
|
| 1632 |
+
"learning_rate": 0.0001695811266249398,
|
| 1633 |
+
"loss": 1.6778,
|
| 1634 |
+
"step": 13550
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.44,
|
| 1638 |
+
"learning_rate": 0.00016909966297544535,
|
| 1639 |
+
"loss": 1.657,
|
| 1640 |
+
"step": 13600
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.44,
|
| 1644 |
+
"learning_rate": 0.00016861819932595086,
|
| 1645 |
+
"loss": 1.6411,
|
| 1646 |
+
"step": 13650
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.44,
|
| 1650 |
+
"learning_rate": 0.0001681367356764564,
|
| 1651 |
+
"loss": 1.612,
|
| 1652 |
+
"step": 13700
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 0.44,
|
| 1656 |
+
"learning_rate": 0.00016765527202696197,
|
| 1657 |
+
"loss": 1.6802,
|
| 1658 |
+
"step": 13750
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.44,
|
| 1662 |
+
"learning_rate": 0.00016717380837746748,
|
| 1663 |
+
"loss": 1.5544,
|
| 1664 |
+
"step": 13800
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.44,
|
| 1668 |
+
"learning_rate": 0.00016669234472797302,
|
| 1669 |
+
"loss": 1.6872,
|
| 1670 |
+
"step": 13850
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.45,
|
| 1674 |
+
"learning_rate": 0.00016621088107847856,
|
| 1675 |
+
"loss": 1.6855,
|
| 1676 |
+
"step": 13900
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.45,
|
| 1680 |
+
"learning_rate": 0.00016572941742898408,
|
| 1681 |
+
"loss": 1.6061,
|
| 1682 |
+
"step": 13950
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.45,
|
| 1686 |
+
"learning_rate": 0.00016524795377948964,
|
| 1687 |
+
"loss": 1.6325,
|
| 1688 |
+
"step": 14000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 0.45,
|
| 1692 |
+
"learning_rate": 0.00016476649012999516,
|
| 1693 |
+
"loss": 1.7166,
|
| 1694 |
+
"step": 14050
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.45,
|
| 1698 |
+
"learning_rate": 0.0001642850264805007,
|
| 1699 |
+
"loss": 1.771,
|
| 1700 |
+
"step": 14100
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 0.45,
|
| 1704 |
+
"learning_rate": 0.00016380356283100626,
|
| 1705 |
+
"loss": 1.6768,
|
| 1706 |
+
"step": 14150
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.46,
|
| 1710 |
+
"learning_rate": 0.00016332209918151178,
|
| 1711 |
+
"loss": 1.6166,
|
| 1712 |
+
"step": 14200
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.46,
|
| 1716 |
+
"learning_rate": 0.00016284063553201732,
|
| 1717 |
+
"loss": 1.6353,
|
| 1718 |
+
"step": 14250
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.46,
|
| 1722 |
+
"learning_rate": 0.00016235917188252286,
|
| 1723 |
+
"loss": 1.603,
|
| 1724 |
+
"step": 14300
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.46,
|
| 1728 |
+
"learning_rate": 0.00016187770823302837,
|
| 1729 |
+
"loss": 1.566,
|
| 1730 |
+
"step": 14350
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.46,
|
| 1734 |
+
"learning_rate": 0.00016139624458353394,
|
| 1735 |
+
"loss": 1.7014,
|
| 1736 |
+
"step": 14400
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 0.46,
|
| 1740 |
+
"learning_rate": 0.00016091478093403948,
|
| 1741 |
+
"loss": 1.605,
|
| 1742 |
+
"step": 14450
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 0.47,
|
| 1746 |
+
"learning_rate": 0.000160433317284545,
|
| 1747 |
+
"loss": 1.5957,
|
| 1748 |
+
"step": 14500
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.47,
|
| 1752 |
+
"learning_rate": 0.00015995185363505053,
|
| 1753 |
+
"loss": 1.6576,
|
| 1754 |
+
"step": 14550
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.47,
|
| 1758 |
+
"learning_rate": 0.0001594703899855561,
|
| 1759 |
+
"loss": 1.7874,
|
| 1760 |
+
"step": 14600
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.47,
|
| 1764 |
+
"learning_rate": 0.0001589889263360616,
|
| 1765 |
+
"loss": 1.6839,
|
| 1766 |
+
"step": 14650
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.47,
|
| 1770 |
+
"learning_rate": 0.00015850746268656715,
|
| 1771 |
+
"loss": 1.6772,
|
| 1772 |
+
"step": 14700
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 0.47,
|
| 1776 |
+
"learning_rate": 0.0001580259990370727,
|
| 1777 |
+
"loss": 1.6047,
|
| 1778 |
+
"step": 14750
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 0.48,
|
| 1782 |
+
"learning_rate": 0.0001575445353875782,
|
| 1783 |
+
"loss": 1.62,
|
| 1784 |
+
"step": 14800
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 0.48,
|
| 1788 |
+
"learning_rate": 0.00015706307173808377,
|
| 1789 |
+
"loss": 1.6616,
|
| 1790 |
+
"step": 14850
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 0.48,
|
| 1794 |
+
"learning_rate": 0.00015658160808858929,
|
| 1795 |
+
"loss": 1.6361,
|
| 1796 |
+
"step": 14900
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.48,
|
| 1800 |
+
"learning_rate": 0.00015610014443909483,
|
| 1801 |
+
"loss": 1.6233,
|
| 1802 |
+
"step": 14950
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.48,
|
| 1806 |
+
"learning_rate": 0.0001556186807896004,
|
| 1807 |
+
"loss": 1.5949,
|
| 1808 |
+
"step": 15000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.48,
|
| 1812 |
+
"learning_rate": 0.0001551372171401059,
|
| 1813 |
+
"loss": 1.7567,
|
| 1814 |
+
"step": 15050
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.48,
|
| 1818 |
+
"learning_rate": 0.00015465575349061145,
|
| 1819 |
+
"loss": 1.6849,
|
| 1820 |
+
"step": 15100
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.49,
|
| 1824 |
+
"learning_rate": 0.000154174289841117,
|
| 1825 |
+
"loss": 1.6694,
|
| 1826 |
+
"step": 15150
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 0.49,
|
| 1830 |
+
"learning_rate": 0.0001536928261916225,
|
| 1831 |
+
"loss": 1.6714,
|
| 1832 |
+
"step": 15200
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.49,
|
| 1836 |
+
"learning_rate": 0.00015321136254212807,
|
| 1837 |
+
"loss": 1.5966,
|
| 1838 |
+
"step": 15250
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.49,
|
| 1842 |
+
"learning_rate": 0.0001527298988926336,
|
| 1843 |
+
"loss": 1.7145,
|
| 1844 |
+
"step": 15300
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.49,
|
| 1848 |
+
"learning_rate": 0.00015224843524313912,
|
| 1849 |
+
"loss": 1.6672,
|
| 1850 |
+
"step": 15350
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.49,
|
| 1854 |
+
"learning_rate": 0.00015176697159364466,
|
| 1855 |
+
"loss": 1.6871,
|
| 1856 |
+
"step": 15400
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 0.5,
|
| 1860 |
+
"learning_rate": 0.00015128550794415023,
|
| 1861 |
+
"loss": 1.5669,
|
| 1862 |
+
"step": 15450
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.5,
|
| 1866 |
+
"learning_rate": 0.00015080404429465574,
|
| 1867 |
+
"loss": 1.6625,
|
| 1868 |
+
"step": 15500
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 0.5,
|
| 1872 |
+
"learning_rate": 0.00015032258064516128,
|
| 1873 |
+
"loss": 1.585,
|
| 1874 |
+
"step": 15550
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.5,
|
| 1878 |
+
"learning_rate": 0.00014984111699566682,
|
| 1879 |
+
"loss": 1.7518,
|
| 1880 |
+
"step": 15600
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.5,
|
| 1884 |
+
"learning_rate": 0.00014935965334617236,
|
| 1885 |
+
"loss": 1.6921,
|
| 1886 |
+
"step": 15650
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.5,
|
| 1890 |
+
"learning_rate": 0.00014887818969667787,
|
| 1891 |
+
"loss": 1.6763,
|
| 1892 |
+
"step": 15700
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.51,
|
| 1896 |
+
"learning_rate": 0.00014839672604718344,
|
| 1897 |
+
"loss": 1.6484,
|
| 1898 |
+
"step": 15750
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.51,
|
| 1902 |
+
"learning_rate": 0.00014791526239768895,
|
| 1903 |
+
"loss": 1.6529,
|
| 1904 |
+
"step": 15800
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.51,
|
| 1908 |
+
"learning_rate": 0.0001474337987481945,
|
| 1909 |
+
"loss": 1.7107,
|
| 1910 |
+
"step": 15850
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.51,
|
| 1914 |
+
"learning_rate": 0.00014695233509870003,
|
| 1915 |
+
"loss": 1.6711,
|
| 1916 |
+
"step": 15900
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.51,
|
| 1920 |
+
"learning_rate": 0.00014647087144920557,
|
| 1921 |
+
"loss": 1.5795,
|
| 1922 |
+
"step": 15950
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.51,
|
| 1926 |
+
"learning_rate": 0.00014598940779971111,
|
| 1927 |
+
"loss": 1.6493,
|
| 1928 |
+
"step": 16000
|
| 1929 |
+
}
|
| 1930 |
+
],
|
| 1931 |
+
"max_steps": 31155,
|
| 1932 |
+
"num_train_epochs": 1,
|
| 1933 |
+
"total_flos": 2.3287744559130624e+17,
|
| 1934 |
+
"trial_name": null,
|
| 1935 |
+
"trial_params": null
|
| 1936 |
+
}
|