Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +4509 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,4509 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 36816,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 1.9972837896566713e-05,
|
| 14 |
+
"loss": 1.0389,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 1.994567579313342e-05,
|
| 20 |
+
"loss": 0.9783,
|
| 21 |
+
"step": 100
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 1.9918513689700133e-05,
|
| 26 |
+
"loss": 0.8949,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.02,
|
| 31 |
+
"learning_rate": 1.9891351586266844e-05,
|
| 32 |
+
"loss": 0.8832,
|
| 33 |
+
"step": 200
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 1.9864189482833552e-05,
|
| 38 |
+
"loss": 0.8377,
|
| 39 |
+
"step": 250
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 1.983702737940026e-05,
|
| 44 |
+
"loss": 0.8297,
|
| 45 |
+
"step": 300
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03,
|
| 49 |
+
"learning_rate": 1.9809865275966972e-05,
|
| 50 |
+
"loss": 0.7661,
|
| 51 |
+
"step": 350
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 1.978270317253368e-05,
|
| 56 |
+
"loss": 0.803,
|
| 57 |
+
"step": 400
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.04,
|
| 61 |
+
"learning_rate": 1.9755541069100392e-05,
|
| 62 |
+
"loss": 0.7971,
|
| 63 |
+
"step": 450
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.04,
|
| 67 |
+
"learning_rate": 1.9728378965667103e-05,
|
| 68 |
+
"loss": 0.7882,
|
| 69 |
+
"step": 500
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"learning_rate": 1.9701216862233815e-05,
|
| 74 |
+
"loss": 0.7406,
|
| 75 |
+
"step": 550
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.05,
|
| 79 |
+
"learning_rate": 1.9674054758800523e-05,
|
| 80 |
+
"loss": 0.7539,
|
| 81 |
+
"step": 600
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.05,
|
| 85 |
+
"learning_rate": 1.9646892655367235e-05,
|
| 86 |
+
"loss": 0.7562,
|
| 87 |
+
"step": 650
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06,
|
| 91 |
+
"learning_rate": 1.9619730551933943e-05,
|
| 92 |
+
"loss": 0.7525,
|
| 93 |
+
"step": 700
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.06,
|
| 97 |
+
"learning_rate": 1.9592568448500654e-05,
|
| 98 |
+
"loss": 0.7378,
|
| 99 |
+
"step": 750
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07,
|
| 103 |
+
"learning_rate": 1.9565406345067362e-05,
|
| 104 |
+
"loss": 0.738,
|
| 105 |
+
"step": 800
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.07,
|
| 109 |
+
"learning_rate": 1.9538244241634074e-05,
|
| 110 |
+
"loss": 0.727,
|
| 111 |
+
"step": 850
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.07,
|
| 115 |
+
"learning_rate": 1.9511082138200782e-05,
|
| 116 |
+
"loss": 0.7254,
|
| 117 |
+
"step": 900
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.08,
|
| 121 |
+
"learning_rate": 1.9483920034767494e-05,
|
| 122 |
+
"loss": 0.7433,
|
| 123 |
+
"step": 950
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.08,
|
| 127 |
+
"learning_rate": 1.9456757931334205e-05,
|
| 128 |
+
"loss": 0.7547,
|
| 129 |
+
"step": 1000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09,
|
| 133 |
+
"learning_rate": 1.9429595827900913e-05,
|
| 134 |
+
"loss": 0.7251,
|
| 135 |
+
"step": 1050
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.09,
|
| 139 |
+
"learning_rate": 1.9402433724467625e-05,
|
| 140 |
+
"loss": 0.7203,
|
| 141 |
+
"step": 1100
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.09,
|
| 145 |
+
"learning_rate": 1.9375271621034336e-05,
|
| 146 |
+
"loss": 0.704,
|
| 147 |
+
"step": 1150
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.1,
|
| 151 |
+
"learning_rate": 1.9348109517601044e-05,
|
| 152 |
+
"loss": 0.7271,
|
| 153 |
+
"step": 1200
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.1,
|
| 157 |
+
"learning_rate": 1.9320947414167756e-05,
|
| 158 |
+
"loss": 0.6955,
|
| 159 |
+
"step": 1250
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.11,
|
| 163 |
+
"learning_rate": 1.9293785310734464e-05,
|
| 164 |
+
"loss": 0.6565,
|
| 165 |
+
"step": 1300
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.11,
|
| 169 |
+
"learning_rate": 1.9266623207301176e-05,
|
| 170 |
+
"loss": 0.7266,
|
| 171 |
+
"step": 1350
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.11,
|
| 175 |
+
"learning_rate": 1.9239461103867884e-05,
|
| 176 |
+
"loss": 0.6809,
|
| 177 |
+
"step": 1400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.12,
|
| 181 |
+
"learning_rate": 1.9212299000434595e-05,
|
| 182 |
+
"loss": 0.7345,
|
| 183 |
+
"step": 1450
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.12,
|
| 187 |
+
"learning_rate": 1.9185136897001307e-05,
|
| 188 |
+
"loss": 0.7052,
|
| 189 |
+
"step": 1500
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.13,
|
| 193 |
+
"learning_rate": 1.9157974793568015e-05,
|
| 194 |
+
"loss": 0.7097,
|
| 195 |
+
"step": 1550
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.13,
|
| 199 |
+
"learning_rate": 1.9130812690134726e-05,
|
| 200 |
+
"loss": 0.7224,
|
| 201 |
+
"step": 1600
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.13,
|
| 205 |
+
"learning_rate": 1.9103650586701438e-05,
|
| 206 |
+
"loss": 0.6998,
|
| 207 |
+
"step": 1650
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.14,
|
| 211 |
+
"learning_rate": 1.9076488483268146e-05,
|
| 212 |
+
"loss": 0.6799,
|
| 213 |
+
"step": 1700
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.14,
|
| 217 |
+
"learning_rate": 1.9049326379834858e-05,
|
| 218 |
+
"loss": 0.6957,
|
| 219 |
+
"step": 1750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.15,
|
| 223 |
+
"learning_rate": 1.9022164276401566e-05,
|
| 224 |
+
"loss": 0.7111,
|
| 225 |
+
"step": 1800
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.15,
|
| 229 |
+
"learning_rate": 1.8995002172968274e-05,
|
| 230 |
+
"loss": 0.7109,
|
| 231 |
+
"step": 1850
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.15,
|
| 235 |
+
"learning_rate": 1.8967840069534985e-05,
|
| 236 |
+
"loss": 0.7088,
|
| 237 |
+
"step": 1900
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.16,
|
| 241 |
+
"learning_rate": 1.8940677966101697e-05,
|
| 242 |
+
"loss": 0.6694,
|
| 243 |
+
"step": 1950
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.16,
|
| 247 |
+
"learning_rate": 1.8913515862668405e-05,
|
| 248 |
+
"loss": 0.6598,
|
| 249 |
+
"step": 2000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.17,
|
| 253 |
+
"learning_rate": 1.8886353759235117e-05,
|
| 254 |
+
"loss": 0.6834,
|
| 255 |
+
"step": 2050
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.17,
|
| 259 |
+
"learning_rate": 1.8859191655801828e-05,
|
| 260 |
+
"loss": 0.7251,
|
| 261 |
+
"step": 2100
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.18,
|
| 265 |
+
"learning_rate": 1.8832029552368536e-05,
|
| 266 |
+
"loss": 0.6978,
|
| 267 |
+
"step": 2150
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.18,
|
| 271 |
+
"learning_rate": 1.8804867448935248e-05,
|
| 272 |
+
"loss": 0.6761,
|
| 273 |
+
"step": 2200
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.18,
|
| 277 |
+
"learning_rate": 1.8777705345501956e-05,
|
| 278 |
+
"loss": 0.6638,
|
| 279 |
+
"step": 2250
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.19,
|
| 283 |
+
"learning_rate": 1.8750543242068668e-05,
|
| 284 |
+
"loss": 0.6987,
|
| 285 |
+
"step": 2300
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.19,
|
| 289 |
+
"learning_rate": 1.8723381138635376e-05,
|
| 290 |
+
"loss": 0.6922,
|
| 291 |
+
"step": 2350
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.2,
|
| 295 |
+
"learning_rate": 1.8696219035202087e-05,
|
| 296 |
+
"loss": 0.6764,
|
| 297 |
+
"step": 2400
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.2,
|
| 301 |
+
"learning_rate": 1.86690569317688e-05,
|
| 302 |
+
"loss": 0.6812,
|
| 303 |
+
"step": 2450
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.2,
|
| 307 |
+
"learning_rate": 1.8641894828335507e-05,
|
| 308 |
+
"loss": 0.6764,
|
| 309 |
+
"step": 2500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.21,
|
| 313 |
+
"learning_rate": 1.861473272490222e-05,
|
| 314 |
+
"loss": 0.6856,
|
| 315 |
+
"step": 2550
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.21,
|
| 319 |
+
"learning_rate": 1.858757062146893e-05,
|
| 320 |
+
"loss": 0.7037,
|
| 321 |
+
"step": 2600
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.22,
|
| 325 |
+
"learning_rate": 1.8560408518035638e-05,
|
| 326 |
+
"loss": 0.6843,
|
| 327 |
+
"step": 2650
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.22,
|
| 331 |
+
"learning_rate": 1.853324641460235e-05,
|
| 332 |
+
"loss": 0.6761,
|
| 333 |
+
"step": 2700
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.22,
|
| 337 |
+
"learning_rate": 1.8506084311169058e-05,
|
| 338 |
+
"loss": 0.6778,
|
| 339 |
+
"step": 2750
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.23,
|
| 343 |
+
"learning_rate": 1.847892220773577e-05,
|
| 344 |
+
"loss": 0.6733,
|
| 345 |
+
"step": 2800
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.23,
|
| 349 |
+
"learning_rate": 1.8451760104302477e-05,
|
| 350 |
+
"loss": 0.6711,
|
| 351 |
+
"step": 2850
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.24,
|
| 355 |
+
"learning_rate": 1.842459800086919e-05,
|
| 356 |
+
"loss": 0.684,
|
| 357 |
+
"step": 2900
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.24,
|
| 361 |
+
"learning_rate": 1.8397435897435897e-05,
|
| 362 |
+
"loss": 0.6783,
|
| 363 |
+
"step": 2950
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.24,
|
| 367 |
+
"learning_rate": 1.837027379400261e-05,
|
| 368 |
+
"loss": 0.6788,
|
| 369 |
+
"step": 3000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.25,
|
| 373 |
+
"learning_rate": 1.834311169056932e-05,
|
| 374 |
+
"loss": 0.6497,
|
| 375 |
+
"step": 3050
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.25,
|
| 379 |
+
"learning_rate": 1.8315949587136028e-05,
|
| 380 |
+
"loss": 0.6246,
|
| 381 |
+
"step": 3100
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.26,
|
| 385 |
+
"learning_rate": 1.828878748370274e-05,
|
| 386 |
+
"loss": 0.6497,
|
| 387 |
+
"step": 3150
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.26,
|
| 391 |
+
"learning_rate": 1.826162538026945e-05,
|
| 392 |
+
"loss": 0.663,
|
| 393 |
+
"step": 3200
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.26,
|
| 397 |
+
"learning_rate": 1.823446327683616e-05,
|
| 398 |
+
"loss": 0.6542,
|
| 399 |
+
"step": 3250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.27,
|
| 403 |
+
"learning_rate": 1.820730117340287e-05,
|
| 404 |
+
"loss": 0.6734,
|
| 405 |
+
"step": 3300
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.27,
|
| 409 |
+
"learning_rate": 1.818013906996958e-05,
|
| 410 |
+
"loss": 0.6548,
|
| 411 |
+
"step": 3350
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.28,
|
| 415 |
+
"learning_rate": 1.815297696653629e-05,
|
| 416 |
+
"loss": 0.6362,
|
| 417 |
+
"step": 3400
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.28,
|
| 421 |
+
"learning_rate": 1.8125814863103e-05,
|
| 422 |
+
"loss": 0.6544,
|
| 423 |
+
"step": 3450
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.29,
|
| 427 |
+
"learning_rate": 1.809865275966971e-05,
|
| 428 |
+
"loss": 0.658,
|
| 429 |
+
"step": 3500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.29,
|
| 433 |
+
"learning_rate": 1.8071490656236422e-05,
|
| 434 |
+
"loss": 0.6443,
|
| 435 |
+
"step": 3550
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.29,
|
| 439 |
+
"learning_rate": 1.804432855280313e-05,
|
| 440 |
+
"loss": 0.6707,
|
| 441 |
+
"step": 3600
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.3,
|
| 445 |
+
"learning_rate": 1.801716644936984e-05,
|
| 446 |
+
"loss": 0.6438,
|
| 447 |
+
"step": 3650
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.3,
|
| 451 |
+
"learning_rate": 1.7990004345936553e-05,
|
| 452 |
+
"loss": 0.6871,
|
| 453 |
+
"step": 3700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.31,
|
| 457 |
+
"learning_rate": 1.796284224250326e-05,
|
| 458 |
+
"loss": 0.6694,
|
| 459 |
+
"step": 3750
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.31,
|
| 463 |
+
"learning_rate": 1.793568013906997e-05,
|
| 464 |
+
"loss": 0.6279,
|
| 465 |
+
"step": 3800
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.31,
|
| 469 |
+
"learning_rate": 1.790851803563668e-05,
|
| 470 |
+
"loss": 0.634,
|
| 471 |
+
"step": 3850
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.32,
|
| 475 |
+
"learning_rate": 1.788135593220339e-05,
|
| 476 |
+
"loss": 0.6535,
|
| 477 |
+
"step": 3900
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.32,
|
| 481 |
+
"learning_rate": 1.78541938287701e-05,
|
| 482 |
+
"loss": 0.6388,
|
| 483 |
+
"step": 3950
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.33,
|
| 487 |
+
"learning_rate": 1.7827031725336812e-05,
|
| 488 |
+
"loss": 0.6227,
|
| 489 |
+
"step": 4000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.33,
|
| 493 |
+
"learning_rate": 1.7799869621903524e-05,
|
| 494 |
+
"loss": 0.6741,
|
| 495 |
+
"step": 4050
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.33,
|
| 499 |
+
"learning_rate": 1.777270751847023e-05,
|
| 500 |
+
"loss": 0.6566,
|
| 501 |
+
"step": 4100
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.34,
|
| 505 |
+
"learning_rate": 1.7745545415036943e-05,
|
| 506 |
+
"loss": 0.6695,
|
| 507 |
+
"step": 4150
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.34,
|
| 511 |
+
"learning_rate": 1.771838331160365e-05,
|
| 512 |
+
"loss": 0.6454,
|
| 513 |
+
"step": 4200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.35,
|
| 517 |
+
"learning_rate": 1.7691221208170363e-05,
|
| 518 |
+
"loss": 0.6531,
|
| 519 |
+
"step": 4250
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.35,
|
| 523 |
+
"learning_rate": 1.766405910473707e-05,
|
| 524 |
+
"loss": 0.6287,
|
| 525 |
+
"step": 4300
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.35,
|
| 529 |
+
"learning_rate": 1.7636897001303783e-05,
|
| 530 |
+
"loss": 0.6356,
|
| 531 |
+
"step": 4350
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.36,
|
| 535 |
+
"learning_rate": 1.760973489787049e-05,
|
| 536 |
+
"loss": 0.6609,
|
| 537 |
+
"step": 4400
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.36,
|
| 541 |
+
"learning_rate": 1.7582572794437202e-05,
|
| 542 |
+
"loss": 0.6502,
|
| 543 |
+
"step": 4450
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.37,
|
| 547 |
+
"learning_rate": 1.7555410691003914e-05,
|
| 548 |
+
"loss": 0.6211,
|
| 549 |
+
"step": 4500
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.37,
|
| 553 |
+
"learning_rate": 1.7528248587570622e-05,
|
| 554 |
+
"loss": 0.6432,
|
| 555 |
+
"step": 4550
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.37,
|
| 559 |
+
"learning_rate": 1.7501086484137333e-05,
|
| 560 |
+
"loss": 0.6416,
|
| 561 |
+
"step": 4600
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.38,
|
| 565 |
+
"learning_rate": 1.7473924380704045e-05,
|
| 566 |
+
"loss": 0.6211,
|
| 567 |
+
"step": 4650
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.38,
|
| 571 |
+
"learning_rate": 1.7446762277270753e-05,
|
| 572 |
+
"loss": 0.6537,
|
| 573 |
+
"step": 4700
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.39,
|
| 577 |
+
"learning_rate": 1.7419600173837465e-05,
|
| 578 |
+
"loss": 0.6001,
|
| 579 |
+
"step": 4750
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.39,
|
| 583 |
+
"learning_rate": 1.7392438070404173e-05,
|
| 584 |
+
"loss": 0.6057,
|
| 585 |
+
"step": 4800
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.4,
|
| 589 |
+
"learning_rate": 1.7365275966970884e-05,
|
| 590 |
+
"loss": 0.6716,
|
| 591 |
+
"step": 4850
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.4,
|
| 595 |
+
"learning_rate": 1.7338113863537592e-05,
|
| 596 |
+
"loss": 0.6447,
|
| 597 |
+
"step": 4900
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.4,
|
| 601 |
+
"learning_rate": 1.7310951760104304e-05,
|
| 602 |
+
"loss": 0.6371,
|
| 603 |
+
"step": 4950
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.41,
|
| 607 |
+
"learning_rate": 1.7283789656671015e-05,
|
| 608 |
+
"loss": 0.6369,
|
| 609 |
+
"step": 5000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.41,
|
| 613 |
+
"eval_accuracy": 0.7400916963830871,
|
| 614 |
+
"eval_loss": 0.6398988962173462,
|
| 615 |
+
"eval_runtime": 16.7037,
|
| 616 |
+
"eval_samples_per_second": 587.595,
|
| 617 |
+
"eval_steps_per_second": 36.758,
|
| 618 |
+
"step": 5000
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.41,
|
| 622 |
+
"learning_rate": 1.7256627553237724e-05,
|
| 623 |
+
"loss": 0.6108,
|
| 624 |
+
"step": 5050
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.42,
|
| 628 |
+
"learning_rate": 1.7229465449804435e-05,
|
| 629 |
+
"loss": 0.6118,
|
| 630 |
+
"step": 5100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.42,
|
| 634 |
+
"learning_rate": 1.7202303346371147e-05,
|
| 635 |
+
"loss": 0.6306,
|
| 636 |
+
"step": 5150
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.42,
|
| 640 |
+
"learning_rate": 1.7175141242937855e-05,
|
| 641 |
+
"loss": 0.6089,
|
| 642 |
+
"step": 5200
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.43,
|
| 646 |
+
"learning_rate": 1.7147979139504566e-05,
|
| 647 |
+
"loss": 0.6283,
|
| 648 |
+
"step": 5250
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.43,
|
| 652 |
+
"learning_rate": 1.7120817036071274e-05,
|
| 653 |
+
"loss": 0.6383,
|
| 654 |
+
"step": 5300
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.44,
|
| 658 |
+
"learning_rate": 1.7093654932637983e-05,
|
| 659 |
+
"loss": 0.6179,
|
| 660 |
+
"step": 5350
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.44,
|
| 664 |
+
"learning_rate": 1.7066492829204694e-05,
|
| 665 |
+
"loss": 0.6357,
|
| 666 |
+
"step": 5400
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.44,
|
| 670 |
+
"learning_rate": 1.7039330725771406e-05,
|
| 671 |
+
"loss": 0.634,
|
| 672 |
+
"step": 5450
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.45,
|
| 676 |
+
"learning_rate": 1.7012168622338114e-05,
|
| 677 |
+
"loss": 0.5852,
|
| 678 |
+
"step": 5500
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.45,
|
| 682 |
+
"learning_rate": 1.6985006518904825e-05,
|
| 683 |
+
"loss": 0.6249,
|
| 684 |
+
"step": 5550
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.46,
|
| 688 |
+
"learning_rate": 1.6957844415471537e-05,
|
| 689 |
+
"loss": 0.6043,
|
| 690 |
+
"step": 5600
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.46,
|
| 694 |
+
"learning_rate": 1.6930682312038245e-05,
|
| 695 |
+
"loss": 0.6171,
|
| 696 |
+
"step": 5650
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.46,
|
| 700 |
+
"learning_rate": 1.6903520208604957e-05,
|
| 701 |
+
"loss": 0.6358,
|
| 702 |
+
"step": 5700
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.47,
|
| 706 |
+
"learning_rate": 1.6876358105171665e-05,
|
| 707 |
+
"loss": 0.6129,
|
| 708 |
+
"step": 5750
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.47,
|
| 712 |
+
"learning_rate": 1.6849196001738376e-05,
|
| 713 |
+
"loss": 0.6069,
|
| 714 |
+
"step": 5800
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.48,
|
| 718 |
+
"learning_rate": 1.6822033898305084e-05,
|
| 719 |
+
"loss": 0.6553,
|
| 720 |
+
"step": 5850
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.48,
|
| 724 |
+
"learning_rate": 1.6794871794871796e-05,
|
| 725 |
+
"loss": 0.6372,
|
| 726 |
+
"step": 5900
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.48,
|
| 730 |
+
"learning_rate": 1.6767709691438507e-05,
|
| 731 |
+
"loss": 0.6348,
|
| 732 |
+
"step": 5950
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.49,
|
| 736 |
+
"learning_rate": 1.6740547588005215e-05,
|
| 737 |
+
"loss": 0.596,
|
| 738 |
+
"step": 6000
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.49,
|
| 742 |
+
"learning_rate": 1.6713385484571927e-05,
|
| 743 |
+
"loss": 0.6192,
|
| 744 |
+
"step": 6050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.5,
|
| 748 |
+
"learning_rate": 1.668622338113864e-05,
|
| 749 |
+
"loss": 0.6435,
|
| 750 |
+
"step": 6100
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 0.5,
|
| 754 |
+
"learning_rate": 1.6659061277705347e-05,
|
| 755 |
+
"loss": 0.6135,
|
| 756 |
+
"step": 6150
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.51,
|
| 760 |
+
"learning_rate": 1.6631899174272058e-05,
|
| 761 |
+
"loss": 0.6476,
|
| 762 |
+
"step": 6200
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.51,
|
| 766 |
+
"learning_rate": 1.6604737070838766e-05,
|
| 767 |
+
"loss": 0.646,
|
| 768 |
+
"step": 6250
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.51,
|
| 772 |
+
"learning_rate": 1.6577574967405478e-05,
|
| 773 |
+
"loss": 0.6104,
|
| 774 |
+
"step": 6300
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.52,
|
| 778 |
+
"learning_rate": 1.6550412863972186e-05,
|
| 779 |
+
"loss": 0.6151,
|
| 780 |
+
"step": 6350
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.52,
|
| 784 |
+
"learning_rate": 1.6523250760538898e-05,
|
| 785 |
+
"loss": 0.6314,
|
| 786 |
+
"step": 6400
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.53,
|
| 790 |
+
"learning_rate": 1.6496088657105606e-05,
|
| 791 |
+
"loss": 0.6256,
|
| 792 |
+
"step": 6450
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 0.53,
|
| 796 |
+
"learning_rate": 1.6468926553672317e-05,
|
| 797 |
+
"loss": 0.6284,
|
| 798 |
+
"step": 6500
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 0.53,
|
| 802 |
+
"learning_rate": 1.644176445023903e-05,
|
| 803 |
+
"loss": 0.6134,
|
| 804 |
+
"step": 6550
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.54,
|
| 808 |
+
"learning_rate": 1.6414602346805737e-05,
|
| 809 |
+
"loss": 0.6203,
|
| 810 |
+
"step": 6600
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.54,
|
| 814 |
+
"learning_rate": 1.638744024337245e-05,
|
| 815 |
+
"loss": 0.5652,
|
| 816 |
+
"step": 6650
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 0.55,
|
| 820 |
+
"learning_rate": 1.636027813993916e-05,
|
| 821 |
+
"loss": 0.6353,
|
| 822 |
+
"step": 6700
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.55,
|
| 826 |
+
"learning_rate": 1.6333116036505868e-05,
|
| 827 |
+
"loss": 0.6013,
|
| 828 |
+
"step": 6750
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.55,
|
| 832 |
+
"learning_rate": 1.630595393307258e-05,
|
| 833 |
+
"loss": 0.6112,
|
| 834 |
+
"step": 6800
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.56,
|
| 838 |
+
"learning_rate": 1.6278791829639288e-05,
|
| 839 |
+
"loss": 0.6161,
|
| 840 |
+
"step": 6850
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 0.56,
|
| 844 |
+
"learning_rate": 1.6251629726206e-05,
|
| 845 |
+
"loss": 0.6173,
|
| 846 |
+
"step": 6900
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.57,
|
| 850 |
+
"learning_rate": 1.6224467622772707e-05,
|
| 851 |
+
"loss": 0.6203,
|
| 852 |
+
"step": 6950
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.57,
|
| 856 |
+
"learning_rate": 1.619730551933942e-05,
|
| 857 |
+
"loss": 0.6264,
|
| 858 |
+
"step": 7000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.57,
|
| 862 |
+
"learning_rate": 1.617014341590613e-05,
|
| 863 |
+
"loss": 0.6104,
|
| 864 |
+
"step": 7050
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.58,
|
| 868 |
+
"learning_rate": 1.614298131247284e-05,
|
| 869 |
+
"loss": 0.5963,
|
| 870 |
+
"step": 7100
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.58,
|
| 874 |
+
"learning_rate": 1.611581920903955e-05,
|
| 875 |
+
"loss": 0.6044,
|
| 876 |
+
"step": 7150
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 0.59,
|
| 880 |
+
"learning_rate": 1.608865710560626e-05,
|
| 881 |
+
"loss": 0.601,
|
| 882 |
+
"step": 7200
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 0.59,
|
| 886 |
+
"learning_rate": 1.606149500217297e-05,
|
| 887 |
+
"loss": 0.5882,
|
| 888 |
+
"step": 7250
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.59,
|
| 892 |
+
"learning_rate": 1.6034332898739678e-05,
|
| 893 |
+
"loss": 0.5976,
|
| 894 |
+
"step": 7300
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6,
|
| 898 |
+
"learning_rate": 1.600717079530639e-05,
|
| 899 |
+
"loss": 0.6077,
|
| 900 |
+
"step": 7350
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 0.6,
|
| 904 |
+
"learning_rate": 1.5980008691873098e-05,
|
| 905 |
+
"loss": 0.6079,
|
| 906 |
+
"step": 7400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.61,
|
| 910 |
+
"learning_rate": 1.595284658843981e-05,
|
| 911 |
+
"loss": 0.6181,
|
| 912 |
+
"step": 7450
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.61,
|
| 916 |
+
"learning_rate": 1.592568448500652e-05,
|
| 917 |
+
"loss": 0.6186,
|
| 918 |
+
"step": 7500
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 0.62,
|
| 922 |
+
"learning_rate": 1.589852238157323e-05,
|
| 923 |
+
"loss": 0.6134,
|
| 924 |
+
"step": 7550
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.62,
|
| 928 |
+
"learning_rate": 1.587136027813994e-05,
|
| 929 |
+
"loss": 0.6042,
|
| 930 |
+
"step": 7600
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 0.62,
|
| 934 |
+
"learning_rate": 1.5844198174706652e-05,
|
| 935 |
+
"loss": 0.621,
|
| 936 |
+
"step": 7650
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 0.63,
|
| 940 |
+
"learning_rate": 1.5817036071273363e-05,
|
| 941 |
+
"loss": 0.6249,
|
| 942 |
+
"step": 7700
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 0.63,
|
| 946 |
+
"learning_rate": 1.578987396784007e-05,
|
| 947 |
+
"loss": 0.6148,
|
| 948 |
+
"step": 7750
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.64,
|
| 952 |
+
"learning_rate": 1.576271186440678e-05,
|
| 953 |
+
"loss": 0.6052,
|
| 954 |
+
"step": 7800
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.64,
|
| 958 |
+
"learning_rate": 1.573554976097349e-05,
|
| 959 |
+
"loss": 0.6207,
|
| 960 |
+
"step": 7850
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.64,
|
| 964 |
+
"learning_rate": 1.57083876575402e-05,
|
| 965 |
+
"loss": 0.575,
|
| 966 |
+
"step": 7900
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.65,
|
| 970 |
+
"learning_rate": 1.568122555410691e-05,
|
| 971 |
+
"loss": 0.6081,
|
| 972 |
+
"step": 7950
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 0.65,
|
| 976 |
+
"learning_rate": 1.5654063450673622e-05,
|
| 977 |
+
"loss": 0.6148,
|
| 978 |
+
"step": 8000
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 0.66,
|
| 982 |
+
"learning_rate": 1.562690134724033e-05,
|
| 983 |
+
"loss": 0.603,
|
| 984 |
+
"step": 8050
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.66,
|
| 988 |
+
"learning_rate": 1.5599739243807042e-05,
|
| 989 |
+
"loss": 0.5894,
|
| 990 |
+
"step": 8100
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.66,
|
| 994 |
+
"learning_rate": 1.5572577140373754e-05,
|
| 995 |
+
"loss": 0.6078,
|
| 996 |
+
"step": 8150
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.67,
|
| 1000 |
+
"learning_rate": 1.5545415036940462e-05,
|
| 1001 |
+
"loss": 0.5784,
|
| 1002 |
+
"step": 8200
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.67,
|
| 1006 |
+
"learning_rate": 1.5518252933507173e-05,
|
| 1007 |
+
"loss": 0.5849,
|
| 1008 |
+
"step": 8250
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.68,
|
| 1012 |
+
"learning_rate": 1.549109083007388e-05,
|
| 1013 |
+
"loss": 0.6066,
|
| 1014 |
+
"step": 8300
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.68,
|
| 1018 |
+
"learning_rate": 1.5463928726640593e-05,
|
| 1019 |
+
"loss": 0.599,
|
| 1020 |
+
"step": 8350
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.68,
|
| 1024 |
+
"learning_rate": 1.54367666232073e-05,
|
| 1025 |
+
"loss": 0.6186,
|
| 1026 |
+
"step": 8400
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 0.69,
|
| 1030 |
+
"learning_rate": 1.5409604519774013e-05,
|
| 1031 |
+
"loss": 0.5947,
|
| 1032 |
+
"step": 8450
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.69,
|
| 1036 |
+
"learning_rate": 1.5382442416340724e-05,
|
| 1037 |
+
"loss": 0.6191,
|
| 1038 |
+
"step": 8500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.7,
|
| 1042 |
+
"learning_rate": 1.5355280312907432e-05,
|
| 1043 |
+
"loss": 0.5896,
|
| 1044 |
+
"step": 8550
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7,
|
| 1048 |
+
"learning_rate": 1.5328118209474144e-05,
|
| 1049 |
+
"loss": 0.6241,
|
| 1050 |
+
"step": 8600
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.7,
|
| 1054 |
+
"learning_rate": 1.5300956106040855e-05,
|
| 1055 |
+
"loss": 0.5856,
|
| 1056 |
+
"step": 8650
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 0.71,
|
| 1060 |
+
"learning_rate": 1.5273794002607563e-05,
|
| 1061 |
+
"loss": 0.5966,
|
| 1062 |
+
"step": 8700
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 0.71,
|
| 1066 |
+
"learning_rate": 1.5246631899174273e-05,
|
| 1067 |
+
"loss": 0.602,
|
| 1068 |
+
"step": 8750
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.72,
|
| 1072 |
+
"learning_rate": 1.5219469795740985e-05,
|
| 1073 |
+
"loss": 0.6188,
|
| 1074 |
+
"step": 8800
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.72,
|
| 1078 |
+
"learning_rate": 1.5192307692307693e-05,
|
| 1079 |
+
"loss": 0.6342,
|
| 1080 |
+
"step": 8850
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.73,
|
| 1084 |
+
"learning_rate": 1.5165145588874404e-05,
|
| 1085 |
+
"loss": 0.605,
|
| 1086 |
+
"step": 8900
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 0.73,
|
| 1090 |
+
"learning_rate": 1.5137983485441114e-05,
|
| 1091 |
+
"loss": 0.6138,
|
| 1092 |
+
"step": 8950
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 0.73,
|
| 1096 |
+
"learning_rate": 1.5110821382007822e-05,
|
| 1097 |
+
"loss": 0.6043,
|
| 1098 |
+
"step": 9000
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 0.74,
|
| 1102 |
+
"learning_rate": 1.5083659278574534e-05,
|
| 1103 |
+
"loss": 0.6095,
|
| 1104 |
+
"step": 9050
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.74,
|
| 1108 |
+
"learning_rate": 1.5056497175141245e-05,
|
| 1109 |
+
"loss": 0.5755,
|
| 1110 |
+
"step": 9100
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.75,
|
| 1114 |
+
"learning_rate": 1.5029335071707954e-05,
|
| 1115 |
+
"loss": 0.5987,
|
| 1116 |
+
"step": 9150
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.75,
|
| 1120 |
+
"learning_rate": 1.5002172968274663e-05,
|
| 1121 |
+
"loss": 0.5747,
|
| 1122 |
+
"step": 9200
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.75,
|
| 1126 |
+
"learning_rate": 1.4975010864841375e-05,
|
| 1127 |
+
"loss": 0.5762,
|
| 1128 |
+
"step": 9250
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 0.76,
|
| 1132 |
+
"learning_rate": 1.4947848761408083e-05,
|
| 1133 |
+
"loss": 0.5679,
|
| 1134 |
+
"step": 9300
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.76,
|
| 1138 |
+
"learning_rate": 1.4920686657974795e-05,
|
| 1139 |
+
"loss": 0.5724,
|
| 1140 |
+
"step": 9350
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.77,
|
| 1144 |
+
"learning_rate": 1.4893524554541504e-05,
|
| 1145 |
+
"loss": 0.5789,
|
| 1146 |
+
"step": 9400
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.77,
|
| 1150 |
+
"learning_rate": 1.4866362451108216e-05,
|
| 1151 |
+
"loss": 0.5888,
|
| 1152 |
+
"step": 9450
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 0.77,
|
| 1156 |
+
"learning_rate": 1.4839200347674924e-05,
|
| 1157 |
+
"loss": 0.5821,
|
| 1158 |
+
"step": 9500
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.78,
|
| 1162 |
+
"learning_rate": 1.4812038244241636e-05,
|
| 1163 |
+
"loss": 0.6436,
|
| 1164 |
+
"step": 9550
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.78,
|
| 1168 |
+
"learning_rate": 1.4784876140808346e-05,
|
| 1169 |
+
"loss": 0.5867,
|
| 1170 |
+
"step": 9600
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.79,
|
| 1174 |
+
"learning_rate": 1.4757714037375055e-05,
|
| 1175 |
+
"loss": 0.5537,
|
| 1176 |
+
"step": 9650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.79,
|
| 1180 |
+
"learning_rate": 1.4730551933941765e-05,
|
| 1181 |
+
"loss": 0.5978,
|
| 1182 |
+
"step": 9700
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 0.79,
|
| 1186 |
+
"learning_rate": 1.4703389830508477e-05,
|
| 1187 |
+
"loss": 0.5499,
|
| 1188 |
+
"step": 9750
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 0.8,
|
| 1192 |
+
"learning_rate": 1.4676227727075185e-05,
|
| 1193 |
+
"loss": 0.6219,
|
| 1194 |
+
"step": 9800
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8,
|
| 1198 |
+
"learning_rate": 1.4649065623641896e-05,
|
| 1199 |
+
"loss": 0.5923,
|
| 1200 |
+
"step": 9850
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.81,
|
| 1204 |
+
"learning_rate": 1.4621903520208606e-05,
|
| 1205 |
+
"loss": 0.6236,
|
| 1206 |
+
"step": 9900
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.81,
|
| 1210 |
+
"learning_rate": 1.4594741416775316e-05,
|
| 1211 |
+
"loss": 0.5732,
|
| 1212 |
+
"step": 9950
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 0.81,
|
| 1216 |
+
"learning_rate": 1.4567579313342026e-05,
|
| 1217 |
+
"loss": 0.5945,
|
| 1218 |
+
"step": 10000
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 0.81,
|
| 1222 |
+
"eval_accuracy": 0.7680081507896077,
|
| 1223 |
+
"eval_loss": 0.5745856761932373,
|
| 1224 |
+
"eval_runtime": 16.756,
|
| 1225 |
+
"eval_samples_per_second": 585.761,
|
| 1226 |
+
"eval_steps_per_second": 36.644,
|
| 1227 |
+
"step": 10000
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.82,
|
| 1231 |
+
"learning_rate": 1.4540417209908737e-05,
|
| 1232 |
+
"loss": 0.5909,
|
| 1233 |
+
"step": 10050
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.82,
|
| 1237 |
+
"learning_rate": 1.4513255106475446e-05,
|
| 1238 |
+
"loss": 0.5937,
|
| 1239 |
+
"step": 10100
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.83,
|
| 1243 |
+
"learning_rate": 1.4486093003042157e-05,
|
| 1244 |
+
"loss": 0.5916,
|
| 1245 |
+
"step": 10150
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.83,
|
| 1249 |
+
"learning_rate": 1.4458930899608867e-05,
|
| 1250 |
+
"loss": 0.5777,
|
| 1251 |
+
"step": 10200
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.84,
|
| 1255 |
+
"learning_rate": 1.4431768796175577e-05,
|
| 1256 |
+
"loss": 0.581,
|
| 1257 |
+
"step": 10250
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.84,
|
| 1261 |
+
"learning_rate": 1.4404606692742287e-05,
|
| 1262 |
+
"loss": 0.6225,
|
| 1263 |
+
"step": 10300
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.84,
|
| 1267 |
+
"learning_rate": 1.4377444589308998e-05,
|
| 1268 |
+
"loss": 0.6204,
|
| 1269 |
+
"step": 10350
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.85,
|
| 1273 |
+
"learning_rate": 1.4350282485875708e-05,
|
| 1274 |
+
"loss": 0.5783,
|
| 1275 |
+
"step": 10400
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.85,
|
| 1279 |
+
"learning_rate": 1.4323120382442418e-05,
|
| 1280 |
+
"loss": 0.5715,
|
| 1281 |
+
"step": 10450
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.86,
|
| 1285 |
+
"learning_rate": 1.4295958279009128e-05,
|
| 1286 |
+
"loss": 0.601,
|
| 1287 |
+
"step": 10500
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.86,
|
| 1291 |
+
"learning_rate": 1.4268796175575839e-05,
|
| 1292 |
+
"loss": 0.5964,
|
| 1293 |
+
"step": 10550
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 0.86,
|
| 1297 |
+
"learning_rate": 1.4241634072142547e-05,
|
| 1298 |
+
"loss": 0.6087,
|
| 1299 |
+
"step": 10600
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.87,
|
| 1303 |
+
"learning_rate": 1.4214471968709259e-05,
|
| 1304 |
+
"loss": 0.5917,
|
| 1305 |
+
"step": 10650
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.87,
|
| 1309 |
+
"learning_rate": 1.4187309865275969e-05,
|
| 1310 |
+
"loss": 0.5593,
|
| 1311 |
+
"step": 10700
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.88,
|
| 1315 |
+
"learning_rate": 1.4160147761842677e-05,
|
| 1316 |
+
"loss": 0.6105,
|
| 1317 |
+
"step": 10750
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.88,
|
| 1321 |
+
"learning_rate": 1.4132985658409388e-05,
|
| 1322 |
+
"loss": 0.573,
|
| 1323 |
+
"step": 10800
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.88,
|
| 1327 |
+
"learning_rate": 1.41058235549761e-05,
|
| 1328 |
+
"loss": 0.615,
|
| 1329 |
+
"step": 10850
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.89,
|
| 1333 |
+
"learning_rate": 1.4078661451542808e-05,
|
| 1334 |
+
"loss": 0.5854,
|
| 1335 |
+
"step": 10900
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 0.89,
|
| 1339 |
+
"learning_rate": 1.4051499348109518e-05,
|
| 1340 |
+
"loss": 0.5828,
|
| 1341 |
+
"step": 10950
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.9,
|
| 1345 |
+
"learning_rate": 1.402433724467623e-05,
|
| 1346 |
+
"loss": 0.5892,
|
| 1347 |
+
"step": 11000
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.9,
|
| 1351 |
+
"learning_rate": 1.3997175141242937e-05,
|
| 1352 |
+
"loss": 0.59,
|
| 1353 |
+
"step": 11050
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.9,
|
| 1357 |
+
"learning_rate": 1.3970013037809649e-05,
|
| 1358 |
+
"loss": 0.5789,
|
| 1359 |
+
"step": 11100
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.91,
|
| 1363 |
+
"learning_rate": 1.3942850934376359e-05,
|
| 1364 |
+
"loss": 0.5549,
|
| 1365 |
+
"step": 11150
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 0.91,
|
| 1369 |
+
"learning_rate": 1.391568883094307e-05,
|
| 1370 |
+
"loss": 0.6024,
|
| 1371 |
+
"step": 11200
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 0.92,
|
| 1375 |
+
"learning_rate": 1.3888526727509778e-05,
|
| 1376 |
+
"loss": 0.5597,
|
| 1377 |
+
"step": 11250
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.92,
|
| 1381 |
+
"learning_rate": 1.386136462407649e-05,
|
| 1382 |
+
"loss": 0.5646,
|
| 1383 |
+
"step": 11300
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.92,
|
| 1387 |
+
"learning_rate": 1.38342025206432e-05,
|
| 1388 |
+
"loss": 0.5711,
|
| 1389 |
+
"step": 11350
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.93,
|
| 1393 |
+
"learning_rate": 1.380704041720991e-05,
|
| 1394 |
+
"loss": 0.5555,
|
| 1395 |
+
"step": 11400
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.93,
|
| 1399 |
+
"learning_rate": 1.377987831377662e-05,
|
| 1400 |
+
"loss": 0.5504,
|
| 1401 |
+
"step": 11450
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.94,
|
| 1405 |
+
"learning_rate": 1.3752716210343331e-05,
|
| 1406 |
+
"loss": 0.5712,
|
| 1407 |
+
"step": 11500
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 0.94,
|
| 1411 |
+
"learning_rate": 1.372555410691004e-05,
|
| 1412 |
+
"loss": 0.5768,
|
| 1413 |
+
"step": 11550
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.95,
|
| 1417 |
+
"learning_rate": 1.369839200347675e-05,
|
| 1418 |
+
"loss": 0.5877,
|
| 1419 |
+
"step": 11600
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.95,
|
| 1423 |
+
"learning_rate": 1.367122990004346e-05,
|
| 1424 |
+
"loss": 0.5639,
|
| 1425 |
+
"step": 11650
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.95,
|
| 1429 |
+
"learning_rate": 1.364406779661017e-05,
|
| 1430 |
+
"loss": 0.5803,
|
| 1431 |
+
"step": 11700
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.96,
|
| 1435 |
+
"learning_rate": 1.361690569317688e-05,
|
| 1436 |
+
"loss": 0.602,
|
| 1437 |
+
"step": 11750
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.96,
|
| 1441 |
+
"learning_rate": 1.3589743589743592e-05,
|
| 1442 |
+
"loss": 0.5758,
|
| 1443 |
+
"step": 11800
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 0.97,
|
| 1447 |
+
"learning_rate": 1.35625814863103e-05,
|
| 1448 |
+
"loss": 0.5657,
|
| 1449 |
+
"step": 11850
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.97,
|
| 1453 |
+
"learning_rate": 1.3535419382877011e-05,
|
| 1454 |
+
"loss": 0.5857,
|
| 1455 |
+
"step": 11900
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.97,
|
| 1459 |
+
"learning_rate": 1.3508257279443721e-05,
|
| 1460 |
+
"loss": 0.5922,
|
| 1461 |
+
"step": 11950
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 0.98,
|
| 1465 |
+
"learning_rate": 1.3481095176010431e-05,
|
| 1466 |
+
"loss": 0.6179,
|
| 1467 |
+
"step": 12000
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.98,
|
| 1471 |
+
"learning_rate": 1.3453933072577141e-05,
|
| 1472 |
+
"loss": 0.5464,
|
| 1473 |
+
"step": 12050
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.99,
|
| 1477 |
+
"learning_rate": 1.3426770969143852e-05,
|
| 1478 |
+
"loss": 0.5578,
|
| 1479 |
+
"step": 12100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.99,
|
| 1483 |
+
"learning_rate": 1.3399608865710562e-05,
|
| 1484 |
+
"loss": 0.6018,
|
| 1485 |
+
"step": 12150
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.99,
|
| 1489 |
+
"learning_rate": 1.3372446762277272e-05,
|
| 1490 |
+
"loss": 0.5907,
|
| 1491 |
+
"step": 12200
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 1.0,
|
| 1495 |
+
"learning_rate": 1.3345284658843982e-05,
|
| 1496 |
+
"loss": 0.5403,
|
| 1497 |
+
"step": 12250
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.0,
|
| 1501 |
+
"learning_rate": 1.3318122555410693e-05,
|
| 1502 |
+
"loss": 0.5376,
|
| 1503 |
+
"step": 12300
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 1.01,
|
| 1507 |
+
"learning_rate": 1.3290960451977402e-05,
|
| 1508 |
+
"loss": 0.5479,
|
| 1509 |
+
"step": 12350
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 1.01,
|
| 1513 |
+
"learning_rate": 1.3263798348544113e-05,
|
| 1514 |
+
"loss": 0.5121,
|
| 1515 |
+
"step": 12400
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 1.01,
|
| 1519 |
+
"learning_rate": 1.3236636245110823e-05,
|
| 1520 |
+
"loss": 0.4965,
|
| 1521 |
+
"step": 12450
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.02,
|
| 1525 |
+
"learning_rate": 1.3209474141677531e-05,
|
| 1526 |
+
"loss": 0.481,
|
| 1527 |
+
"step": 12500
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.02,
|
| 1531 |
+
"learning_rate": 1.3182312038244243e-05,
|
| 1532 |
+
"loss": 0.4933,
|
| 1533 |
+
"step": 12550
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 1.03,
|
| 1537 |
+
"learning_rate": 1.3155149934810954e-05,
|
| 1538 |
+
"loss": 0.4907,
|
| 1539 |
+
"step": 12600
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.03,
|
| 1543 |
+
"learning_rate": 1.3127987831377662e-05,
|
| 1544 |
+
"loss": 0.5178,
|
| 1545 |
+
"step": 12650
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 1.03,
|
| 1549 |
+
"learning_rate": 1.3100825727944372e-05,
|
| 1550 |
+
"loss": 0.5114,
|
| 1551 |
+
"step": 12700
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 1.04,
|
| 1555 |
+
"learning_rate": 1.3073663624511084e-05,
|
| 1556 |
+
"loss": 0.4874,
|
| 1557 |
+
"step": 12750
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.04,
|
| 1561 |
+
"learning_rate": 1.3046501521077792e-05,
|
| 1562 |
+
"loss": 0.5061,
|
| 1563 |
+
"step": 12800
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.05,
|
| 1567 |
+
"learning_rate": 1.3019339417644503e-05,
|
| 1568 |
+
"loss": 0.5151,
|
| 1569 |
+
"step": 12850
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.05,
|
| 1573 |
+
"learning_rate": 1.2992177314211213e-05,
|
| 1574 |
+
"loss": 0.5208,
|
| 1575 |
+
"step": 12900
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 1.06,
|
| 1579 |
+
"learning_rate": 1.2965015210777925e-05,
|
| 1580 |
+
"loss": 0.4972,
|
| 1581 |
+
"step": 12950
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 1.06,
|
| 1585 |
+
"learning_rate": 1.2937853107344633e-05,
|
| 1586 |
+
"loss": 0.4916,
|
| 1587 |
+
"step": 13000
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 1.06,
|
| 1591 |
+
"learning_rate": 1.2910691003911344e-05,
|
| 1592 |
+
"loss": 0.5134,
|
| 1593 |
+
"step": 13050
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.07,
|
| 1597 |
+
"learning_rate": 1.2883528900478056e-05,
|
| 1598 |
+
"loss": 0.5218,
|
| 1599 |
+
"step": 13100
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 1.07,
|
| 1603 |
+
"learning_rate": 1.2856366797044764e-05,
|
| 1604 |
+
"loss": 0.5096,
|
| 1605 |
+
"step": 13150
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.08,
|
| 1609 |
+
"learning_rate": 1.2829204693611474e-05,
|
| 1610 |
+
"loss": 0.5038,
|
| 1611 |
+
"step": 13200
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.08,
|
| 1615 |
+
"learning_rate": 1.2802042590178185e-05,
|
| 1616 |
+
"loss": 0.4812,
|
| 1617 |
+
"step": 13250
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 1.08,
|
| 1621 |
+
"learning_rate": 1.2774880486744893e-05,
|
| 1622 |
+
"loss": 0.5195,
|
| 1623 |
+
"step": 13300
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 1.09,
|
| 1627 |
+
"learning_rate": 1.2747718383311605e-05,
|
| 1628 |
+
"loss": 0.5023,
|
| 1629 |
+
"step": 13350
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 1.09,
|
| 1633 |
+
"learning_rate": 1.2720556279878315e-05,
|
| 1634 |
+
"loss": 0.5112,
|
| 1635 |
+
"step": 13400
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 1.1,
|
| 1639 |
+
"learning_rate": 1.2693394176445025e-05,
|
| 1640 |
+
"loss": 0.5131,
|
| 1641 |
+
"step": 13450
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 1.1,
|
| 1645 |
+
"learning_rate": 1.2666232073011735e-05,
|
| 1646 |
+
"loss": 0.5317,
|
| 1647 |
+
"step": 13500
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.1,
|
| 1651 |
+
"learning_rate": 1.2639069969578446e-05,
|
| 1652 |
+
"loss": 0.4836,
|
| 1653 |
+
"step": 13550
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.11,
|
| 1657 |
+
"learning_rate": 1.2611907866145154e-05,
|
| 1658 |
+
"loss": 0.5583,
|
| 1659 |
+
"step": 13600
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 1.11,
|
| 1663 |
+
"learning_rate": 1.2584745762711866e-05,
|
| 1664 |
+
"loss": 0.5195,
|
| 1665 |
+
"step": 13650
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.12,
|
| 1669 |
+
"learning_rate": 1.2557583659278576e-05,
|
| 1670 |
+
"loss": 0.4933,
|
| 1671 |
+
"step": 13700
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 1.12,
|
| 1675 |
+
"learning_rate": 1.2530421555845285e-05,
|
| 1676 |
+
"loss": 0.5127,
|
| 1677 |
+
"step": 13750
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 1.12,
|
| 1681 |
+
"learning_rate": 1.2503259452411995e-05,
|
| 1682 |
+
"loss": 0.4924,
|
| 1683 |
+
"step": 13800
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 1.13,
|
| 1687 |
+
"learning_rate": 1.2476097348978707e-05,
|
| 1688 |
+
"loss": 0.4972,
|
| 1689 |
+
"step": 13850
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.13,
|
| 1693 |
+
"learning_rate": 1.2448935245545417e-05,
|
| 1694 |
+
"loss": 0.5336,
|
| 1695 |
+
"step": 13900
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 1.14,
|
| 1699 |
+
"learning_rate": 1.2421773142112126e-05,
|
| 1700 |
+
"loss": 0.4995,
|
| 1701 |
+
"step": 13950
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 1.14,
|
| 1705 |
+
"learning_rate": 1.2394611038678836e-05,
|
| 1706 |
+
"loss": 0.4911,
|
| 1707 |
+
"step": 14000
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.14,
|
| 1711 |
+
"learning_rate": 1.2367448935245548e-05,
|
| 1712 |
+
"loss": 0.5177,
|
| 1713 |
+
"step": 14050
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 1.15,
|
| 1717 |
+
"learning_rate": 1.2340286831812256e-05,
|
| 1718 |
+
"loss": 0.4603,
|
| 1719 |
+
"step": 14100
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 1.15,
|
| 1723 |
+
"learning_rate": 1.2313124728378967e-05,
|
| 1724 |
+
"loss": 0.5058,
|
| 1725 |
+
"step": 14150
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 1.16,
|
| 1729 |
+
"learning_rate": 1.2285962624945677e-05,
|
| 1730 |
+
"loss": 0.5239,
|
| 1731 |
+
"step": 14200
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.16,
|
| 1735 |
+
"learning_rate": 1.2258800521512385e-05,
|
| 1736 |
+
"loss": 0.5143,
|
| 1737 |
+
"step": 14250
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.17,
|
| 1741 |
+
"learning_rate": 1.2231638418079097e-05,
|
| 1742 |
+
"loss": 0.4909,
|
| 1743 |
+
"step": 14300
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 1.17,
|
| 1747 |
+
"learning_rate": 1.2204476314645808e-05,
|
| 1748 |
+
"loss": 0.5104,
|
| 1749 |
+
"step": 14350
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 1.17,
|
| 1753 |
+
"learning_rate": 1.2177314211212517e-05,
|
| 1754 |
+
"loss": 0.5263,
|
| 1755 |
+
"step": 14400
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 1.18,
|
| 1759 |
+
"learning_rate": 1.2150152107779226e-05,
|
| 1760 |
+
"loss": 0.4828,
|
| 1761 |
+
"step": 14450
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 1.18,
|
| 1765 |
+
"learning_rate": 1.2122990004345938e-05,
|
| 1766 |
+
"loss": 0.4891,
|
| 1767 |
+
"step": 14500
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 1.19,
|
| 1771 |
+
"learning_rate": 1.2095827900912646e-05,
|
| 1772 |
+
"loss": 0.5003,
|
| 1773 |
+
"step": 14550
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.19,
|
| 1777 |
+
"learning_rate": 1.2068665797479358e-05,
|
| 1778 |
+
"loss": 0.4827,
|
| 1779 |
+
"step": 14600
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 1.19,
|
| 1783 |
+
"learning_rate": 1.204150369404607e-05,
|
| 1784 |
+
"loss": 0.5017,
|
| 1785 |
+
"step": 14650
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 1.2,
|
| 1789 |
+
"learning_rate": 1.2014341590612777e-05,
|
| 1790 |
+
"loss": 0.525,
|
| 1791 |
+
"step": 14700
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 1.2,
|
| 1795 |
+
"learning_rate": 1.1987179487179487e-05,
|
| 1796 |
+
"loss": 0.5077,
|
| 1797 |
+
"step": 14750
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 1.21,
|
| 1801 |
+
"learning_rate": 1.1960017383746199e-05,
|
| 1802 |
+
"loss": 0.5062,
|
| 1803 |
+
"step": 14800
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 1.21,
|
| 1807 |
+
"learning_rate": 1.193285528031291e-05,
|
| 1808 |
+
"loss": 0.5098,
|
| 1809 |
+
"step": 14850
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.21,
|
| 1813 |
+
"learning_rate": 1.1905693176879618e-05,
|
| 1814 |
+
"loss": 0.5366,
|
| 1815 |
+
"step": 14900
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.22,
|
| 1819 |
+
"learning_rate": 1.1878531073446328e-05,
|
| 1820 |
+
"loss": 0.5097,
|
| 1821 |
+
"step": 14950
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 1.22,
|
| 1825 |
+
"learning_rate": 1.185136897001304e-05,
|
| 1826 |
+
"loss": 0.4847,
|
| 1827 |
+
"step": 15000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 1.22,
|
| 1831 |
+
"eval_accuracy": 0.7772796739684157,
|
| 1832 |
+
"eval_loss": 0.5816997289657593,
|
| 1833 |
+
"eval_runtime": 16.7546,
|
| 1834 |
+
"eval_samples_per_second": 585.81,
|
| 1835 |
+
"eval_steps_per_second": 36.647,
|
| 1836 |
+
"step": 15000
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 1.23,
|
| 1840 |
+
"learning_rate": 1.1824206866579748e-05,
|
| 1841 |
+
"loss": 0.5372,
|
| 1842 |
+
"step": 15050
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 1.23,
|
| 1846 |
+
"learning_rate": 1.179704476314646e-05,
|
| 1847 |
+
"loss": 0.4814,
|
| 1848 |
+
"step": 15100
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 1.23,
|
| 1852 |
+
"learning_rate": 1.176988265971317e-05,
|
| 1853 |
+
"loss": 0.4823,
|
| 1854 |
+
"step": 15150
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 1.24,
|
| 1858 |
+
"learning_rate": 1.1742720556279879e-05,
|
| 1859 |
+
"loss": 0.5288,
|
| 1860 |
+
"step": 15200
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 1.24,
|
| 1864 |
+
"learning_rate": 1.1715558452846589e-05,
|
| 1865 |
+
"loss": 0.5211,
|
| 1866 |
+
"step": 15250
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 1.25,
|
| 1870 |
+
"learning_rate": 1.16883963494133e-05,
|
| 1871 |
+
"loss": 0.4856,
|
| 1872 |
+
"step": 15300
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 1.25,
|
| 1876 |
+
"learning_rate": 1.1661234245980009e-05,
|
| 1877 |
+
"loss": 0.5162,
|
| 1878 |
+
"step": 15350
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 1.25,
|
| 1882 |
+
"learning_rate": 1.163407214254672e-05,
|
| 1883 |
+
"loss": 0.5032,
|
| 1884 |
+
"step": 15400
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 1.26,
|
| 1888 |
+
"learning_rate": 1.160691003911343e-05,
|
| 1889 |
+
"loss": 0.5181,
|
| 1890 |
+
"step": 15450
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 1.26,
|
| 1894 |
+
"learning_rate": 1.157974793568014e-05,
|
| 1895 |
+
"loss": 0.5042,
|
| 1896 |
+
"step": 15500
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 1.27,
|
| 1900 |
+
"learning_rate": 1.155258583224685e-05,
|
| 1901 |
+
"loss": 0.4901,
|
| 1902 |
+
"step": 15550
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 1.27,
|
| 1906 |
+
"learning_rate": 1.1525423728813561e-05,
|
| 1907 |
+
"loss": 0.5109,
|
| 1908 |
+
"step": 15600
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 1.28,
|
| 1912 |
+
"learning_rate": 1.1498261625380271e-05,
|
| 1913 |
+
"loss": 0.5227,
|
| 1914 |
+
"step": 15650
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 1.28,
|
| 1918 |
+
"learning_rate": 1.147109952194698e-05,
|
| 1919 |
+
"loss": 0.4899,
|
| 1920 |
+
"step": 15700
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 1.28,
|
| 1924 |
+
"learning_rate": 1.144393741851369e-05,
|
| 1925 |
+
"loss": 0.4738,
|
| 1926 |
+
"step": 15750
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 1.29,
|
| 1930 |
+
"learning_rate": 1.1416775315080402e-05,
|
| 1931 |
+
"loss": 0.5072,
|
| 1932 |
+
"step": 15800
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 1.29,
|
| 1936 |
+
"learning_rate": 1.138961321164711e-05,
|
| 1937 |
+
"loss": 0.507,
|
| 1938 |
+
"step": 15850
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 1.3,
|
| 1942 |
+
"learning_rate": 1.1362451108213822e-05,
|
| 1943 |
+
"loss": 0.5224,
|
| 1944 |
+
"step": 15900
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 1.3,
|
| 1948 |
+
"learning_rate": 1.1335289004780532e-05,
|
| 1949 |
+
"loss": 0.5032,
|
| 1950 |
+
"step": 15950
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 1.3,
|
| 1954 |
+
"learning_rate": 1.130812690134724e-05,
|
| 1955 |
+
"loss": 0.4855,
|
| 1956 |
+
"step": 16000
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 1.31,
|
| 1960 |
+
"learning_rate": 1.1280964797913951e-05,
|
| 1961 |
+
"loss": 0.4908,
|
| 1962 |
+
"step": 16050
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 1.31,
|
| 1966 |
+
"learning_rate": 1.1253802694480663e-05,
|
| 1967 |
+
"loss": 0.5129,
|
| 1968 |
+
"step": 16100
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 1.32,
|
| 1972 |
+
"learning_rate": 1.1226640591047371e-05,
|
| 1973 |
+
"loss": 0.5153,
|
| 1974 |
+
"step": 16150
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 1.32,
|
| 1978 |
+
"learning_rate": 1.1199478487614082e-05,
|
| 1979 |
+
"loss": 0.5167,
|
| 1980 |
+
"step": 16200
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 1.32,
|
| 1984 |
+
"learning_rate": 1.1172316384180792e-05,
|
| 1985 |
+
"loss": 0.4818,
|
| 1986 |
+
"step": 16250
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 1.33,
|
| 1990 |
+
"learning_rate": 1.11451542807475e-05,
|
| 1991 |
+
"loss": 0.5057,
|
| 1992 |
+
"step": 16300
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 1.33,
|
| 1996 |
+
"learning_rate": 1.1117992177314212e-05,
|
| 1997 |
+
"loss": 0.4963,
|
| 1998 |
+
"step": 16350
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 1.34,
|
| 2002 |
+
"learning_rate": 1.1090830073880923e-05,
|
| 2003 |
+
"loss": 0.4832,
|
| 2004 |
+
"step": 16400
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 1.34,
|
| 2008 |
+
"learning_rate": 1.1063667970447632e-05,
|
| 2009 |
+
"loss": 0.4967,
|
| 2010 |
+
"step": 16450
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 1.34,
|
| 2014 |
+
"learning_rate": 1.1036505867014341e-05,
|
| 2015 |
+
"loss": 0.4924,
|
| 2016 |
+
"step": 16500
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 1.35,
|
| 2020 |
+
"learning_rate": 1.1009343763581053e-05,
|
| 2021 |
+
"loss": 0.4857,
|
| 2022 |
+
"step": 16550
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 1.35,
|
| 2026 |
+
"learning_rate": 1.0982181660147765e-05,
|
| 2027 |
+
"loss": 0.4981,
|
| 2028 |
+
"step": 16600
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 1.36,
|
| 2032 |
+
"learning_rate": 1.0955019556714473e-05,
|
| 2033 |
+
"loss": 0.484,
|
| 2034 |
+
"step": 16650
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 1.36,
|
| 2038 |
+
"learning_rate": 1.0927857453281182e-05,
|
| 2039 |
+
"loss": 0.5214,
|
| 2040 |
+
"step": 16700
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 1.36,
|
| 2044 |
+
"learning_rate": 1.0900695349847894e-05,
|
| 2045 |
+
"loss": 0.5174,
|
| 2046 |
+
"step": 16750
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 1.37,
|
| 2050 |
+
"learning_rate": 1.0873533246414602e-05,
|
| 2051 |
+
"loss": 0.5203,
|
| 2052 |
+
"step": 16800
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 1.37,
|
| 2056 |
+
"learning_rate": 1.0846371142981314e-05,
|
| 2057 |
+
"loss": 0.4865,
|
| 2058 |
+
"step": 16850
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 1.38,
|
| 2062 |
+
"learning_rate": 1.0819209039548024e-05,
|
| 2063 |
+
"loss": 0.501,
|
| 2064 |
+
"step": 16900
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 1.38,
|
| 2068 |
+
"learning_rate": 1.0792046936114733e-05,
|
| 2069 |
+
"loss": 0.4807,
|
| 2070 |
+
"step": 16950
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 1.39,
|
| 2074 |
+
"learning_rate": 1.0764884832681443e-05,
|
| 2075 |
+
"loss": 0.5103,
|
| 2076 |
+
"step": 17000
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 1.39,
|
| 2080 |
+
"learning_rate": 1.0737722729248155e-05,
|
| 2081 |
+
"loss": 0.5467,
|
| 2082 |
+
"step": 17050
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 1.39,
|
| 2086 |
+
"learning_rate": 1.0710560625814863e-05,
|
| 2087 |
+
"loss": 0.4653,
|
| 2088 |
+
"step": 17100
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 1.4,
|
| 2092 |
+
"learning_rate": 1.0683398522381574e-05,
|
| 2093 |
+
"loss": 0.4821,
|
| 2094 |
+
"step": 17150
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 1.4,
|
| 2098 |
+
"learning_rate": 1.0656236418948284e-05,
|
| 2099 |
+
"loss": 0.5136,
|
| 2100 |
+
"step": 17200
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 1.41,
|
| 2104 |
+
"learning_rate": 1.0629074315514994e-05,
|
| 2105 |
+
"loss": 0.4993,
|
| 2106 |
+
"step": 17250
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 1.41,
|
| 2110 |
+
"learning_rate": 1.0601912212081704e-05,
|
| 2111 |
+
"loss": 0.5035,
|
| 2112 |
+
"step": 17300
|
| 2113 |
+
},
|
| 2114 |
+
{
|
| 2115 |
+
"epoch": 1.41,
|
| 2116 |
+
"learning_rate": 1.0574750108648415e-05,
|
| 2117 |
+
"loss": 0.4993,
|
| 2118 |
+
"step": 17350
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 1.42,
|
| 2122 |
+
"learning_rate": 1.0547588005215125e-05,
|
| 2123 |
+
"loss": 0.5187,
|
| 2124 |
+
"step": 17400
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 1.42,
|
| 2128 |
+
"learning_rate": 1.0520425901781835e-05,
|
| 2129 |
+
"loss": 0.507,
|
| 2130 |
+
"step": 17450
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 1.43,
|
| 2134 |
+
"learning_rate": 1.0493263798348545e-05,
|
| 2135 |
+
"loss": 0.4857,
|
| 2136 |
+
"step": 17500
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 1.43,
|
| 2140 |
+
"learning_rate": 1.0466101694915256e-05,
|
| 2141 |
+
"loss": 0.5085,
|
| 2142 |
+
"step": 17550
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 1.43,
|
| 2146 |
+
"learning_rate": 1.0438939591481965e-05,
|
| 2147 |
+
"loss": 0.5106,
|
| 2148 |
+
"step": 17600
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 1.44,
|
| 2152 |
+
"learning_rate": 1.0411777488048676e-05,
|
| 2153 |
+
"loss": 0.5152,
|
| 2154 |
+
"step": 17650
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 1.44,
|
| 2158 |
+
"learning_rate": 1.0384615384615386e-05,
|
| 2159 |
+
"loss": 0.5066,
|
| 2160 |
+
"step": 17700
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 1.45,
|
| 2164 |
+
"learning_rate": 1.0357453281182096e-05,
|
| 2165 |
+
"loss": 0.5097,
|
| 2166 |
+
"step": 17750
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 1.45,
|
| 2170 |
+
"learning_rate": 1.0330291177748806e-05,
|
| 2171 |
+
"loss": 0.5447,
|
| 2172 |
+
"step": 17800
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 1.45,
|
| 2176 |
+
"learning_rate": 1.0303129074315517e-05,
|
| 2177 |
+
"loss": 0.5142,
|
| 2178 |
+
"step": 17850
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.46,
|
| 2182 |
+
"learning_rate": 1.0275966970882225e-05,
|
| 2183 |
+
"loss": 0.5053,
|
| 2184 |
+
"step": 17900
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 1.46,
|
| 2188 |
+
"learning_rate": 1.0248804867448937e-05,
|
| 2189 |
+
"loss": 0.511,
|
| 2190 |
+
"step": 17950
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 1.47,
|
| 2194 |
+
"learning_rate": 1.0221642764015647e-05,
|
| 2195 |
+
"loss": 0.5134,
|
| 2196 |
+
"step": 18000
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 1.47,
|
| 2200 |
+
"learning_rate": 1.0194480660582355e-05,
|
| 2201 |
+
"loss": 0.5304,
|
| 2202 |
+
"step": 18050
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 1.47,
|
| 2206 |
+
"learning_rate": 1.0167318557149066e-05,
|
| 2207 |
+
"loss": 0.4506,
|
| 2208 |
+
"step": 18100
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 1.48,
|
| 2212 |
+
"learning_rate": 1.0140156453715778e-05,
|
| 2213 |
+
"loss": 0.5318,
|
| 2214 |
+
"step": 18150
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 1.48,
|
| 2218 |
+
"learning_rate": 1.0112994350282486e-05,
|
| 2219 |
+
"loss": 0.4923,
|
| 2220 |
+
"step": 18200
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.49,
|
| 2224 |
+
"learning_rate": 1.0085832246849196e-05,
|
| 2225 |
+
"loss": 0.4903,
|
| 2226 |
+
"step": 18250
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 1.49,
|
| 2230 |
+
"learning_rate": 1.0058670143415907e-05,
|
| 2231 |
+
"loss": 0.4626,
|
| 2232 |
+
"step": 18300
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 1.5,
|
| 2236 |
+
"learning_rate": 1.0031508039982619e-05,
|
| 2237 |
+
"loss": 0.4965,
|
| 2238 |
+
"step": 18350
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 1.5,
|
| 2242 |
+
"learning_rate": 1.0004345936549327e-05,
|
| 2243 |
+
"loss": 0.4839,
|
| 2244 |
+
"step": 18400
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 1.5,
|
| 2248 |
+
"learning_rate": 9.977183833116037e-06,
|
| 2249 |
+
"loss": 0.5238,
|
| 2250 |
+
"step": 18450
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 1.51,
|
| 2254 |
+
"learning_rate": 9.950021729682747e-06,
|
| 2255 |
+
"loss": 0.5056,
|
| 2256 |
+
"step": 18500
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 1.51,
|
| 2260 |
+
"learning_rate": 9.922859626249458e-06,
|
| 2261 |
+
"loss": 0.5162,
|
| 2262 |
+
"step": 18550
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.52,
|
| 2266 |
+
"learning_rate": 9.895697522816168e-06,
|
| 2267 |
+
"loss": 0.4718,
|
| 2268 |
+
"step": 18600
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 1.52,
|
| 2272 |
+
"learning_rate": 9.868535419382878e-06,
|
| 2273 |
+
"loss": 0.5132,
|
| 2274 |
+
"step": 18650
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 1.52,
|
| 2278 |
+
"learning_rate": 9.841373315949588e-06,
|
| 2279 |
+
"loss": 0.5185,
|
| 2280 |
+
"step": 18700
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 1.53,
|
| 2284 |
+
"learning_rate": 9.814211212516298e-06,
|
| 2285 |
+
"loss": 0.4927,
|
| 2286 |
+
"step": 18750
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 1.53,
|
| 2290 |
+
"learning_rate": 9.787049109083007e-06,
|
| 2291 |
+
"loss": 0.4992,
|
| 2292 |
+
"step": 18800
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 1.54,
|
| 2296 |
+
"learning_rate": 9.759887005649719e-06,
|
| 2297 |
+
"loss": 0.4462,
|
| 2298 |
+
"step": 18850
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 1.54,
|
| 2302 |
+
"learning_rate": 9.732724902216429e-06,
|
| 2303 |
+
"loss": 0.4955,
|
| 2304 |
+
"step": 18900
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.54,
|
| 2308 |
+
"learning_rate": 9.705562798783139e-06,
|
| 2309 |
+
"loss": 0.5153,
|
| 2310 |
+
"step": 18950
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 1.55,
|
| 2314 |
+
"learning_rate": 9.678400695349848e-06,
|
| 2315 |
+
"loss": 0.4776,
|
| 2316 |
+
"step": 19000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 1.55,
|
| 2320 |
+
"learning_rate": 9.651238591916558e-06,
|
| 2321 |
+
"loss": 0.518,
|
| 2322 |
+
"step": 19050
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 1.56,
|
| 2326 |
+
"learning_rate": 9.62407648848327e-06,
|
| 2327 |
+
"loss": 0.5094,
|
| 2328 |
+
"step": 19100
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 1.56,
|
| 2332 |
+
"learning_rate": 9.59691438504998e-06,
|
| 2333 |
+
"loss": 0.5092,
|
| 2334 |
+
"step": 19150
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 1.56,
|
| 2338 |
+
"learning_rate": 9.56975228161669e-06,
|
| 2339 |
+
"loss": 0.5016,
|
| 2340 |
+
"step": 19200
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 1.57,
|
| 2344 |
+
"learning_rate": 9.5425901781834e-06,
|
| 2345 |
+
"loss": 0.5357,
|
| 2346 |
+
"step": 19250
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.57,
|
| 2350 |
+
"learning_rate": 9.515428074750109e-06,
|
| 2351 |
+
"loss": 0.521,
|
| 2352 |
+
"step": 19300
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 1.58,
|
| 2356 |
+
"learning_rate": 9.488265971316819e-06,
|
| 2357 |
+
"loss": 0.488,
|
| 2358 |
+
"step": 19350
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 1.58,
|
| 2362 |
+
"learning_rate": 9.46110386788353e-06,
|
| 2363 |
+
"loss": 0.4856,
|
| 2364 |
+
"step": 19400
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 1.58,
|
| 2368 |
+
"learning_rate": 9.43394176445024e-06,
|
| 2369 |
+
"loss": 0.506,
|
| 2370 |
+
"step": 19450
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 1.59,
|
| 2374 |
+
"learning_rate": 9.40677966101695e-06,
|
| 2375 |
+
"loss": 0.5038,
|
| 2376 |
+
"step": 19500
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 1.59,
|
| 2380 |
+
"learning_rate": 9.37961755758366e-06,
|
| 2381 |
+
"loss": 0.471,
|
| 2382 |
+
"step": 19550
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 1.6,
|
| 2386 |
+
"learning_rate": 9.35245545415037e-06,
|
| 2387 |
+
"loss": 0.4886,
|
| 2388 |
+
"step": 19600
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.6,
|
| 2392 |
+
"learning_rate": 9.325293350717081e-06,
|
| 2393 |
+
"loss": 0.5131,
|
| 2394 |
+
"step": 19650
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 1.61,
|
| 2398 |
+
"learning_rate": 9.298131247283791e-06,
|
| 2399 |
+
"loss": 0.5082,
|
| 2400 |
+
"step": 19700
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 1.61,
|
| 2404 |
+
"learning_rate": 9.2709691438505e-06,
|
| 2405 |
+
"loss": 0.5021,
|
| 2406 |
+
"step": 19750
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 1.61,
|
| 2410 |
+
"learning_rate": 9.24380704041721e-06,
|
| 2411 |
+
"loss": 0.4695,
|
| 2412 |
+
"step": 19800
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 1.62,
|
| 2416 |
+
"learning_rate": 9.21664493698392e-06,
|
| 2417 |
+
"loss": 0.4883,
|
| 2418 |
+
"step": 19850
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 1.62,
|
| 2422 |
+
"learning_rate": 9.189482833550632e-06,
|
| 2423 |
+
"loss": 0.4833,
|
| 2424 |
+
"step": 19900
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 1.63,
|
| 2428 |
+
"learning_rate": 9.162320730117342e-06,
|
| 2429 |
+
"loss": 0.4642,
|
| 2430 |
+
"step": 19950
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.63,
|
| 2434 |
+
"learning_rate": 9.13515862668405e-06,
|
| 2435 |
+
"loss": 0.5109,
|
| 2436 |
+
"step": 20000
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 1.63,
|
| 2440 |
+
"eval_accuracy": 0.7790117167600611,
|
| 2441 |
+
"eval_loss": 0.5679929852485657,
|
| 2442 |
+
"eval_runtime": 16.5969,
|
| 2443 |
+
"eval_samples_per_second": 591.374,
|
| 2444 |
+
"eval_steps_per_second": 36.995,
|
| 2445 |
+
"step": 20000
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 1.63,
|
| 2449 |
+
"learning_rate": 9.107996523250762e-06,
|
| 2450 |
+
"loss": 0.5284,
|
| 2451 |
+
"step": 20050
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.64,
|
| 2455 |
+
"learning_rate": 9.080834419817471e-06,
|
| 2456 |
+
"loss": 0.4903,
|
| 2457 |
+
"step": 20100
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 1.64,
|
| 2461 |
+
"learning_rate": 9.053672316384181e-06,
|
| 2462 |
+
"loss": 0.5407,
|
| 2463 |
+
"step": 20150
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 1.65,
|
| 2467 |
+
"learning_rate": 9.026510212950891e-06,
|
| 2468 |
+
"loss": 0.4949,
|
| 2469 |
+
"step": 20200
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.65,
|
| 2473 |
+
"learning_rate": 8.999348109517601e-06,
|
| 2474 |
+
"loss": 0.5127,
|
| 2475 |
+
"step": 20250
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 1.65,
|
| 2479 |
+
"learning_rate": 8.972186006084312e-06,
|
| 2480 |
+
"loss": 0.4523,
|
| 2481 |
+
"step": 20300
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 1.66,
|
| 2485 |
+
"learning_rate": 8.945023902651022e-06,
|
| 2486 |
+
"loss": 0.4655,
|
| 2487 |
+
"step": 20350
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 1.66,
|
| 2491 |
+
"learning_rate": 8.917861799217732e-06,
|
| 2492 |
+
"loss": 0.5071,
|
| 2493 |
+
"step": 20400
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.67,
|
| 2497 |
+
"learning_rate": 8.890699695784442e-06,
|
| 2498 |
+
"loss": 0.4697,
|
| 2499 |
+
"step": 20450
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 1.67,
|
| 2503 |
+
"learning_rate": 8.863537592351152e-06,
|
| 2504 |
+
"loss": 0.499,
|
| 2505 |
+
"step": 20500
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 1.67,
|
| 2509 |
+
"learning_rate": 8.836375488917862e-06,
|
| 2510 |
+
"loss": 0.5156,
|
| 2511 |
+
"step": 20550
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.68,
|
| 2515 |
+
"learning_rate": 8.809213385484573e-06,
|
| 2516 |
+
"loss": 0.5244,
|
| 2517 |
+
"step": 20600
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 1.68,
|
| 2521 |
+
"learning_rate": 8.782051282051283e-06,
|
| 2522 |
+
"loss": 0.5259,
|
| 2523 |
+
"step": 20650
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 1.69,
|
| 2527 |
+
"learning_rate": 8.754889178617993e-06,
|
| 2528 |
+
"loss": 0.4975,
|
| 2529 |
+
"step": 20700
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 1.69,
|
| 2533 |
+
"learning_rate": 8.727727075184703e-06,
|
| 2534 |
+
"loss": 0.4861,
|
| 2535 |
+
"step": 20750
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 1.69,
|
| 2539 |
+
"learning_rate": 8.700564971751413e-06,
|
| 2540 |
+
"loss": 0.5068,
|
| 2541 |
+
"step": 20800
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 1.7,
|
| 2545 |
+
"learning_rate": 8.673402868318124e-06,
|
| 2546 |
+
"loss": 0.477,
|
| 2547 |
+
"step": 20850
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 1.7,
|
| 2551 |
+
"learning_rate": 8.646240764884834e-06,
|
| 2552 |
+
"loss": 0.4909,
|
| 2553 |
+
"step": 20900
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 1.71,
|
| 2557 |
+
"learning_rate": 8.619078661451544e-06,
|
| 2558 |
+
"loss": 0.5286,
|
| 2559 |
+
"step": 20950
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 1.71,
|
| 2563 |
+
"learning_rate": 8.591916558018254e-06,
|
| 2564 |
+
"loss": 0.5003,
|
| 2565 |
+
"step": 21000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 1.72,
|
| 2569 |
+
"learning_rate": 8.564754454584963e-06,
|
| 2570 |
+
"loss": 0.4592,
|
| 2571 |
+
"step": 21050
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 1.72,
|
| 2575 |
+
"learning_rate": 8.537592351151673e-06,
|
| 2576 |
+
"loss": 0.4945,
|
| 2577 |
+
"step": 21100
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 1.72,
|
| 2581 |
+
"learning_rate": 8.510430247718385e-06,
|
| 2582 |
+
"loss": 0.5153,
|
| 2583 |
+
"step": 21150
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 1.73,
|
| 2587 |
+
"learning_rate": 8.483268144285095e-06,
|
| 2588 |
+
"loss": 0.488,
|
| 2589 |
+
"step": 21200
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 1.73,
|
| 2593 |
+
"learning_rate": 8.456106040851804e-06,
|
| 2594 |
+
"loss": 0.5066,
|
| 2595 |
+
"step": 21250
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 1.74,
|
| 2599 |
+
"learning_rate": 8.428943937418514e-06,
|
| 2600 |
+
"loss": 0.506,
|
| 2601 |
+
"step": 21300
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 1.74,
|
| 2605 |
+
"learning_rate": 8.401781833985224e-06,
|
| 2606 |
+
"loss": 0.5118,
|
| 2607 |
+
"step": 21350
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 1.74,
|
| 2611 |
+
"learning_rate": 8.374619730551936e-06,
|
| 2612 |
+
"loss": 0.4863,
|
| 2613 |
+
"step": 21400
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 1.75,
|
| 2617 |
+
"learning_rate": 8.347457627118645e-06,
|
| 2618 |
+
"loss": 0.4974,
|
| 2619 |
+
"step": 21450
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 1.75,
|
| 2623 |
+
"learning_rate": 8.320295523685355e-06,
|
| 2624 |
+
"loss": 0.4872,
|
| 2625 |
+
"step": 21500
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 1.76,
|
| 2629 |
+
"learning_rate": 8.293133420252065e-06,
|
| 2630 |
+
"loss": 0.4901,
|
| 2631 |
+
"step": 21550
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 1.76,
|
| 2635 |
+
"learning_rate": 8.265971316818775e-06,
|
| 2636 |
+
"loss": 0.5113,
|
| 2637 |
+
"step": 21600
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 1.76,
|
| 2641 |
+
"learning_rate": 8.238809213385486e-06,
|
| 2642 |
+
"loss": 0.4773,
|
| 2643 |
+
"step": 21650
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 1.77,
|
| 2647 |
+
"learning_rate": 8.211647109952196e-06,
|
| 2648 |
+
"loss": 0.4777,
|
| 2649 |
+
"step": 21700
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 1.77,
|
| 2653 |
+
"learning_rate": 8.184485006518904e-06,
|
| 2654 |
+
"loss": 0.5108,
|
| 2655 |
+
"step": 21750
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 1.78,
|
| 2659 |
+
"learning_rate": 8.157322903085616e-06,
|
| 2660 |
+
"loss": 0.4692,
|
| 2661 |
+
"step": 21800
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 1.78,
|
| 2665 |
+
"learning_rate": 8.130160799652326e-06,
|
| 2666 |
+
"loss": 0.5289,
|
| 2667 |
+
"step": 21850
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 1.78,
|
| 2671 |
+
"learning_rate": 8.102998696219036e-06,
|
| 2672 |
+
"loss": 0.5053,
|
| 2673 |
+
"step": 21900
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 1.79,
|
| 2677 |
+
"learning_rate": 8.075836592785745e-06,
|
| 2678 |
+
"loss": 0.4812,
|
| 2679 |
+
"step": 21950
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 1.79,
|
| 2683 |
+
"learning_rate": 8.048674489352455e-06,
|
| 2684 |
+
"loss": 0.4772,
|
| 2685 |
+
"step": 22000
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 1.8,
|
| 2689 |
+
"learning_rate": 8.021512385919165e-06,
|
| 2690 |
+
"loss": 0.5205,
|
| 2691 |
+
"step": 22050
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 1.8,
|
| 2695 |
+
"learning_rate": 7.994350282485877e-06,
|
| 2696 |
+
"loss": 0.4581,
|
| 2697 |
+
"step": 22100
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.8,
|
| 2701 |
+
"learning_rate": 7.967188179052586e-06,
|
| 2702 |
+
"loss": 0.4899,
|
| 2703 |
+
"step": 22150
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 1.81,
|
| 2707 |
+
"learning_rate": 7.940026075619296e-06,
|
| 2708 |
+
"loss": 0.5192,
|
| 2709 |
+
"step": 22200
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 1.81,
|
| 2713 |
+
"learning_rate": 7.912863972186006e-06,
|
| 2714 |
+
"loss": 0.4652,
|
| 2715 |
+
"step": 22250
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 1.82,
|
| 2719 |
+
"learning_rate": 7.885701868752716e-06,
|
| 2720 |
+
"loss": 0.4865,
|
| 2721 |
+
"step": 22300
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.82,
|
| 2725 |
+
"learning_rate": 7.858539765319428e-06,
|
| 2726 |
+
"loss": 0.5024,
|
| 2727 |
+
"step": 22350
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 1.83,
|
| 2731 |
+
"learning_rate": 7.831377661886137e-06,
|
| 2732 |
+
"loss": 0.5156,
|
| 2733 |
+
"step": 22400
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.83,
|
| 2737 |
+
"learning_rate": 7.804215558452847e-06,
|
| 2738 |
+
"loss": 0.5068,
|
| 2739 |
+
"step": 22450
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.83,
|
| 2743 |
+
"learning_rate": 7.777053455019557e-06,
|
| 2744 |
+
"loss": 0.4798,
|
| 2745 |
+
"step": 22500
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 1.84,
|
| 2749 |
+
"learning_rate": 7.749891351586267e-06,
|
| 2750 |
+
"loss": 0.4908,
|
| 2751 |
+
"step": 22550
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 1.84,
|
| 2755 |
+
"learning_rate": 7.722729248152978e-06,
|
| 2756 |
+
"loss": 0.4852,
|
| 2757 |
+
"step": 22600
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 1.85,
|
| 2761 |
+
"learning_rate": 7.695567144719688e-06,
|
| 2762 |
+
"loss": 0.5014,
|
| 2763 |
+
"step": 22650
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.85,
|
| 2767 |
+
"learning_rate": 7.668405041286398e-06,
|
| 2768 |
+
"loss": 0.4777,
|
| 2769 |
+
"step": 22700
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 1.85,
|
| 2773 |
+
"learning_rate": 7.641242937853108e-06,
|
| 2774 |
+
"loss": 0.4924,
|
| 2775 |
+
"step": 22750
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.86,
|
| 2779 |
+
"learning_rate": 7.614080834419818e-06,
|
| 2780 |
+
"loss": 0.4686,
|
| 2781 |
+
"step": 22800
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.86,
|
| 2785 |
+
"learning_rate": 7.5869187309865275e-06,
|
| 2786 |
+
"loss": 0.4535,
|
| 2787 |
+
"step": 22850
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 1.87,
|
| 2791 |
+
"learning_rate": 7.559756627553238e-06,
|
| 2792 |
+
"loss": 0.488,
|
| 2793 |
+
"step": 22900
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 1.87,
|
| 2797 |
+
"learning_rate": 7.532594524119948e-06,
|
| 2798 |
+
"loss": 0.4881,
|
| 2799 |
+
"step": 22950
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 1.87,
|
| 2803 |
+
"learning_rate": 7.505432420686659e-06,
|
| 2804 |
+
"loss": 0.4768,
|
| 2805 |
+
"step": 23000
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.88,
|
| 2809 |
+
"learning_rate": 7.4782703172533686e-06,
|
| 2810 |
+
"loss": 0.4947,
|
| 2811 |
+
"step": 23050
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 1.88,
|
| 2815 |
+
"learning_rate": 7.451108213820078e-06,
|
| 2816 |
+
"loss": 0.4965,
|
| 2817 |
+
"step": 23100
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.89,
|
| 2821 |
+
"learning_rate": 7.423946110386789e-06,
|
| 2822 |
+
"loss": 0.5233,
|
| 2823 |
+
"step": 23150
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 1.89,
|
| 2827 |
+
"learning_rate": 7.396784006953499e-06,
|
| 2828 |
+
"loss": 0.4837,
|
| 2829 |
+
"step": 23200
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 1.89,
|
| 2833 |
+
"learning_rate": 7.369621903520209e-06,
|
| 2834 |
+
"loss": 0.4922,
|
| 2835 |
+
"step": 23250
|
| 2836 |
+
},
|
| 2837 |
+
{
|
| 2838 |
+
"epoch": 1.9,
|
| 2839 |
+
"learning_rate": 7.3424598000869194e-06,
|
| 2840 |
+
"loss": 0.5411,
|
| 2841 |
+
"step": 23300
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 1.9,
|
| 2845 |
+
"learning_rate": 7.315297696653629e-06,
|
| 2846 |
+
"loss": 0.5368,
|
| 2847 |
+
"step": 23350
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 1.91,
|
| 2851 |
+
"learning_rate": 7.288135593220339e-06,
|
| 2852 |
+
"loss": 0.479,
|
| 2853 |
+
"step": 23400
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 1.91,
|
| 2857 |
+
"learning_rate": 7.26097348978705e-06,
|
| 2858 |
+
"loss": 0.438,
|
| 2859 |
+
"step": 23450
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 1.91,
|
| 2863 |
+
"learning_rate": 7.23381138635376e-06,
|
| 2864 |
+
"loss": 0.4863,
|
| 2865 |
+
"step": 23500
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 1.92,
|
| 2869 |
+
"learning_rate": 7.20664928292047e-06,
|
| 2870 |
+
"loss": 0.4888,
|
| 2871 |
+
"step": 23550
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 1.92,
|
| 2875 |
+
"learning_rate": 7.17948717948718e-06,
|
| 2876 |
+
"loss": 0.4937,
|
| 2877 |
+
"step": 23600
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 1.93,
|
| 2881 |
+
"learning_rate": 7.15232507605389e-06,
|
| 2882 |
+
"loss": 0.4749,
|
| 2883 |
+
"step": 23650
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 1.93,
|
| 2887 |
+
"learning_rate": 7.125162972620601e-06,
|
| 2888 |
+
"loss": 0.4869,
|
| 2889 |
+
"step": 23700
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 1.94,
|
| 2893 |
+
"learning_rate": 7.0980008691873105e-06,
|
| 2894 |
+
"loss": 0.4876,
|
| 2895 |
+
"step": 23750
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 1.94,
|
| 2899 |
+
"learning_rate": 7.07083876575402e-06,
|
| 2900 |
+
"loss": 0.4814,
|
| 2901 |
+
"step": 23800
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.94,
|
| 2905 |
+
"learning_rate": 7.043676662320731e-06,
|
| 2906 |
+
"loss": 0.4893,
|
| 2907 |
+
"step": 23850
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 1.95,
|
| 2911 |
+
"learning_rate": 7.016514558887441e-06,
|
| 2912 |
+
"loss": 0.4679,
|
| 2913 |
+
"step": 23900
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 1.95,
|
| 2917 |
+
"learning_rate": 6.9893524554541515e-06,
|
| 2918 |
+
"loss": 0.46,
|
| 2919 |
+
"step": 23950
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 1.96,
|
| 2923 |
+
"learning_rate": 6.962190352020861e-06,
|
| 2924 |
+
"loss": 0.4786,
|
| 2925 |
+
"step": 24000
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 1.96,
|
| 2929 |
+
"learning_rate": 6.935028248587571e-06,
|
| 2930 |
+
"loss": 0.5118,
|
| 2931 |
+
"step": 24050
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 1.96,
|
| 2935 |
+
"learning_rate": 6.907866145154282e-06,
|
| 2936 |
+
"loss": 0.4952,
|
| 2937 |
+
"step": 24100
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 1.97,
|
| 2941 |
+
"learning_rate": 6.880704041720992e-06,
|
| 2942 |
+
"loss": 0.4985,
|
| 2943 |
+
"step": 24150
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.97,
|
| 2947 |
+
"learning_rate": 6.8535419382877015e-06,
|
| 2948 |
+
"loss": 0.49,
|
| 2949 |
+
"step": 24200
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 1.98,
|
| 2953 |
+
"learning_rate": 6.826379834854412e-06,
|
| 2954 |
+
"loss": 0.4873,
|
| 2955 |
+
"step": 24250
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 1.98,
|
| 2959 |
+
"learning_rate": 6.799217731421122e-06,
|
| 2960 |
+
"loss": 0.4672,
|
| 2961 |
+
"step": 24300
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 1.98,
|
| 2965 |
+
"learning_rate": 6.772055627987833e-06,
|
| 2966 |
+
"loss": 0.4655,
|
| 2967 |
+
"step": 24350
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 1.99,
|
| 2971 |
+
"learning_rate": 6.7448935245545425e-06,
|
| 2972 |
+
"loss": 0.482,
|
| 2973 |
+
"step": 24400
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 1.99,
|
| 2977 |
+
"learning_rate": 6.7177314211212515e-06,
|
| 2978 |
+
"loss": 0.4529,
|
| 2979 |
+
"step": 24450
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 2.0,
|
| 2983 |
+
"learning_rate": 6.690569317687963e-06,
|
| 2984 |
+
"loss": 0.5119,
|
| 2985 |
+
"step": 24500
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 2.0,
|
| 2989 |
+
"learning_rate": 6.663407214254672e-06,
|
| 2990 |
+
"loss": 0.4338,
|
| 2991 |
+
"step": 24550
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 2.0,
|
| 2995 |
+
"learning_rate": 6.636245110821382e-06,
|
| 2996 |
+
"loss": 0.4019,
|
| 2997 |
+
"step": 24600
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 2.01,
|
| 3001 |
+
"learning_rate": 6.6090830073880925e-06,
|
| 3002 |
+
"loss": 0.3948,
|
| 3003 |
+
"step": 24650
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 2.01,
|
| 3007 |
+
"learning_rate": 6.581920903954802e-06,
|
| 3008 |
+
"loss": 0.395,
|
| 3009 |
+
"step": 24700
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 2.02,
|
| 3013 |
+
"learning_rate": 6.554758800521513e-06,
|
| 3014 |
+
"loss": 0.4065,
|
| 3015 |
+
"step": 24750
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 2.02,
|
| 3019 |
+
"learning_rate": 6.527596697088223e-06,
|
| 3020 |
+
"loss": 0.3928,
|
| 3021 |
+
"step": 24800
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 2.02,
|
| 3025 |
+
"learning_rate": 6.500434593654933e-06,
|
| 3026 |
+
"loss": 0.3977,
|
| 3027 |
+
"step": 24850
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 2.03,
|
| 3031 |
+
"learning_rate": 6.473272490221643e-06,
|
| 3032 |
+
"loss": 0.3697,
|
| 3033 |
+
"step": 24900
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 2.03,
|
| 3037 |
+
"learning_rate": 6.446110386788353e-06,
|
| 3038 |
+
"loss": 0.4261,
|
| 3039 |
+
"step": 24950
|
| 3040 |
+
},
|
| 3041 |
+
{
|
| 3042 |
+
"epoch": 2.04,
|
| 3043 |
+
"learning_rate": 6.418948283355063e-06,
|
| 3044 |
+
"loss": 0.3754,
|
| 3045 |
+
"step": 25000
|
| 3046 |
+
},
|
| 3047 |
+
{
|
| 3048 |
+
"epoch": 2.04,
|
| 3049 |
+
"eval_accuracy": 0.7889964340295467,
|
| 3050 |
+
"eval_loss": 0.5795576572418213,
|
| 3051 |
+
"eval_runtime": 16.708,
|
| 3052 |
+
"eval_samples_per_second": 587.442,
|
| 3053 |
+
"eval_steps_per_second": 36.749,
|
| 3054 |
+
"step": 25000
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 2.04,
|
| 3058 |
+
"learning_rate": 6.391786179921774e-06,
|
| 3059 |
+
"loss": 0.4236,
|
| 3060 |
+
"step": 25050
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 2.05,
|
| 3064 |
+
"learning_rate": 6.364624076488484e-06,
|
| 3065 |
+
"loss": 0.4392,
|
| 3066 |
+
"step": 25100
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 2.05,
|
| 3070 |
+
"learning_rate": 6.337461973055193e-06,
|
| 3071 |
+
"loss": 0.3784,
|
| 3072 |
+
"step": 25150
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 2.05,
|
| 3076 |
+
"learning_rate": 6.310299869621904e-06,
|
| 3077 |
+
"loss": 0.3771,
|
| 3078 |
+
"step": 25200
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 2.06,
|
| 3082 |
+
"learning_rate": 6.283137766188614e-06,
|
| 3083 |
+
"loss": 0.4049,
|
| 3084 |
+
"step": 25250
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 2.06,
|
| 3088 |
+
"learning_rate": 6.255975662755325e-06,
|
| 3089 |
+
"loss": 0.4441,
|
| 3090 |
+
"step": 25300
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 2.07,
|
| 3094 |
+
"learning_rate": 6.2288135593220344e-06,
|
| 3095 |
+
"loss": 0.3952,
|
| 3096 |
+
"step": 25350
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 2.07,
|
| 3100 |
+
"learning_rate": 6.201651455888744e-06,
|
| 3101 |
+
"loss": 0.4196,
|
| 3102 |
+
"step": 25400
|
| 3103 |
+
},
|
| 3104 |
+
{
|
| 3105 |
+
"epoch": 2.07,
|
| 3106 |
+
"learning_rate": 6.174489352455455e-06,
|
| 3107 |
+
"loss": 0.3848,
|
| 3108 |
+
"step": 25450
|
| 3109 |
+
},
|
| 3110 |
+
{
|
| 3111 |
+
"epoch": 2.08,
|
| 3112 |
+
"learning_rate": 6.147327249022165e-06,
|
| 3113 |
+
"loss": 0.396,
|
| 3114 |
+
"step": 25500
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 2.08,
|
| 3118 |
+
"learning_rate": 6.120165145588875e-06,
|
| 3119 |
+
"loss": 0.4099,
|
| 3120 |
+
"step": 25550
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 2.09,
|
| 3124 |
+
"learning_rate": 6.093003042155585e-06,
|
| 3125 |
+
"loss": 0.3977,
|
| 3126 |
+
"step": 25600
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 2.09,
|
| 3130 |
+
"learning_rate": 6.065840938722295e-06,
|
| 3131 |
+
"loss": 0.4118,
|
| 3132 |
+
"step": 25650
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 2.09,
|
| 3136 |
+
"learning_rate": 6.038678835289006e-06,
|
| 3137 |
+
"loss": 0.378,
|
| 3138 |
+
"step": 25700
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 2.1,
|
| 3142 |
+
"learning_rate": 6.011516731855716e-06,
|
| 3143 |
+
"loss": 0.3767,
|
| 3144 |
+
"step": 25750
|
| 3145 |
+
},
|
| 3146 |
+
{
|
| 3147 |
+
"epoch": 2.1,
|
| 3148 |
+
"learning_rate": 5.9843546284224255e-06,
|
| 3149 |
+
"loss": 0.4023,
|
| 3150 |
+
"step": 25800
|
| 3151 |
+
},
|
| 3152 |
+
{
|
| 3153 |
+
"epoch": 2.11,
|
| 3154 |
+
"learning_rate": 5.957192524989136e-06,
|
| 3155 |
+
"loss": 0.437,
|
| 3156 |
+
"step": 25850
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 2.11,
|
| 3160 |
+
"learning_rate": 5.930030421555846e-06,
|
| 3161 |
+
"loss": 0.4221,
|
| 3162 |
+
"step": 25900
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 2.11,
|
| 3166 |
+
"learning_rate": 5.902868318122556e-06,
|
| 3167 |
+
"loss": 0.4053,
|
| 3168 |
+
"step": 25950
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 2.12,
|
| 3172 |
+
"learning_rate": 5.8757062146892665e-06,
|
| 3173 |
+
"loss": 0.3794,
|
| 3174 |
+
"step": 26000
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 2.12,
|
| 3178 |
+
"learning_rate": 5.848544111255976e-06,
|
| 3179 |
+
"loss": 0.3832,
|
| 3180 |
+
"step": 26050
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 2.13,
|
| 3184 |
+
"learning_rate": 5.821382007822687e-06,
|
| 3185 |
+
"loss": 0.373,
|
| 3186 |
+
"step": 26100
|
| 3187 |
+
},
|
| 3188 |
+
{
|
| 3189 |
+
"epoch": 2.13,
|
| 3190 |
+
"learning_rate": 5.794219904389397e-06,
|
| 3191 |
+
"loss": 0.3986,
|
| 3192 |
+
"step": 26150
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 2.13,
|
| 3196 |
+
"learning_rate": 5.767057800956106e-06,
|
| 3197 |
+
"loss": 0.4046,
|
| 3198 |
+
"step": 26200
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 2.14,
|
| 3202 |
+
"learning_rate": 5.739895697522817e-06,
|
| 3203 |
+
"loss": 0.404,
|
| 3204 |
+
"step": 26250
|
| 3205 |
+
},
|
| 3206 |
+
{
|
| 3207 |
+
"epoch": 2.14,
|
| 3208 |
+
"learning_rate": 5.712733594089526e-06,
|
| 3209 |
+
"loss": 0.3856,
|
| 3210 |
+
"step": 26300
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 2.15,
|
| 3214 |
+
"learning_rate": 5.685571490656236e-06,
|
| 3215 |
+
"loss": 0.3753,
|
| 3216 |
+
"step": 26350
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 2.15,
|
| 3220 |
+
"learning_rate": 5.658409387222948e-06,
|
| 3221 |
+
"loss": 0.4296,
|
| 3222 |
+
"step": 26400
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 2.16,
|
| 3226 |
+
"learning_rate": 5.631247283789657e-06,
|
| 3227 |
+
"loss": 0.3694,
|
| 3228 |
+
"step": 26450
|
| 3229 |
+
},
|
| 3230 |
+
{
|
| 3231 |
+
"epoch": 2.16,
|
| 3232 |
+
"learning_rate": 5.6040851803563665e-06,
|
| 3233 |
+
"loss": 0.4106,
|
| 3234 |
+
"step": 26500
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 2.16,
|
| 3238 |
+
"learning_rate": 5.576923076923077e-06,
|
| 3239 |
+
"loss": 0.3862,
|
| 3240 |
+
"step": 26550
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 2.17,
|
| 3244 |
+
"learning_rate": 5.549760973489787e-06,
|
| 3245 |
+
"loss": 0.4276,
|
| 3246 |
+
"step": 26600
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 2.17,
|
| 3250 |
+
"learning_rate": 5.522598870056498e-06,
|
| 3251 |
+
"loss": 0.4068,
|
| 3252 |
+
"step": 26650
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 2.18,
|
| 3256 |
+
"learning_rate": 5.4954367666232076e-06,
|
| 3257 |
+
"loss": 0.3879,
|
| 3258 |
+
"step": 26700
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 2.18,
|
| 3262 |
+
"learning_rate": 5.468274663189917e-06,
|
| 3263 |
+
"loss": 0.422,
|
| 3264 |
+
"step": 26750
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 2.18,
|
| 3268 |
+
"learning_rate": 5.441112559756628e-06,
|
| 3269 |
+
"loss": 0.3924,
|
| 3270 |
+
"step": 26800
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 2.19,
|
| 3274 |
+
"learning_rate": 5.413950456323338e-06,
|
| 3275 |
+
"loss": 0.4107,
|
| 3276 |
+
"step": 26850
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 2.19,
|
| 3280 |
+
"learning_rate": 5.386788352890048e-06,
|
| 3281 |
+
"loss": 0.4056,
|
| 3282 |
+
"step": 26900
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 2.2,
|
| 3286 |
+
"learning_rate": 5.359626249456758e-06,
|
| 3287 |
+
"loss": 0.3873,
|
| 3288 |
+
"step": 26950
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 2.2,
|
| 3292 |
+
"learning_rate": 5.332464146023468e-06,
|
| 3293 |
+
"loss": 0.409,
|
| 3294 |
+
"step": 27000
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 2.2,
|
| 3298 |
+
"learning_rate": 5.305302042590179e-06,
|
| 3299 |
+
"loss": 0.4301,
|
| 3300 |
+
"step": 27050
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 2.21,
|
| 3304 |
+
"learning_rate": 5.278139939156889e-06,
|
| 3305 |
+
"loss": 0.409,
|
| 3306 |
+
"step": 27100
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 2.21,
|
| 3310 |
+
"learning_rate": 5.250977835723599e-06,
|
| 3311 |
+
"loss": 0.4243,
|
| 3312 |
+
"step": 27150
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 2.22,
|
| 3316 |
+
"learning_rate": 5.223815732290309e-06,
|
| 3317 |
+
"loss": 0.4252,
|
| 3318 |
+
"step": 27200
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 2.22,
|
| 3322 |
+
"learning_rate": 5.196653628857019e-06,
|
| 3323 |
+
"loss": 0.405,
|
| 3324 |
+
"step": 27250
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 2.22,
|
| 3328 |
+
"learning_rate": 5.169491525423729e-06,
|
| 3329 |
+
"loss": 0.374,
|
| 3330 |
+
"step": 27300
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 2.23,
|
| 3334 |
+
"learning_rate": 5.14232942199044e-06,
|
| 3335 |
+
"loss": 0.4005,
|
| 3336 |
+
"step": 27350
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 2.23,
|
| 3340 |
+
"learning_rate": 5.1151673185571495e-06,
|
| 3341 |
+
"loss": 0.4009,
|
| 3342 |
+
"step": 27400
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 2.24,
|
| 3346 |
+
"learning_rate": 5.08800521512386e-06,
|
| 3347 |
+
"loss": 0.413,
|
| 3348 |
+
"step": 27450
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 2.24,
|
| 3352 |
+
"learning_rate": 5.06084311169057e-06,
|
| 3353 |
+
"loss": 0.4043,
|
| 3354 |
+
"step": 27500
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 2.24,
|
| 3358 |
+
"learning_rate": 5.03368100825728e-06,
|
| 3359 |
+
"loss": 0.3998,
|
| 3360 |
+
"step": 27550
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 2.25,
|
| 3364 |
+
"learning_rate": 5.0065189048239905e-06,
|
| 3365 |
+
"loss": 0.441,
|
| 3366 |
+
"step": 27600
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 2.25,
|
| 3370 |
+
"learning_rate": 4.9793568013907e-06,
|
| 3371 |
+
"loss": 0.4,
|
| 3372 |
+
"step": 27650
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 2.26,
|
| 3376 |
+
"learning_rate": 4.95219469795741e-06,
|
| 3377 |
+
"loss": 0.3921,
|
| 3378 |
+
"step": 27700
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 2.26,
|
| 3382 |
+
"learning_rate": 4.92503259452412e-06,
|
| 3383 |
+
"loss": 0.3924,
|
| 3384 |
+
"step": 27750
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 2.27,
|
| 3388 |
+
"learning_rate": 4.897870491090831e-06,
|
| 3389 |
+
"loss": 0.4186,
|
| 3390 |
+
"step": 27800
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 2.27,
|
| 3394 |
+
"learning_rate": 4.8707083876575405e-06,
|
| 3395 |
+
"loss": 0.3972,
|
| 3396 |
+
"step": 27850
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 2.27,
|
| 3400 |
+
"learning_rate": 4.843546284224251e-06,
|
| 3401 |
+
"loss": 0.425,
|
| 3402 |
+
"step": 27900
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 2.28,
|
| 3406 |
+
"learning_rate": 4.816384180790961e-06,
|
| 3407 |
+
"loss": 0.4067,
|
| 3408 |
+
"step": 27950
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 2.28,
|
| 3412 |
+
"learning_rate": 4.789222077357671e-06,
|
| 3413 |
+
"loss": 0.4474,
|
| 3414 |
+
"step": 28000
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 2.29,
|
| 3418 |
+
"learning_rate": 4.7620599739243815e-06,
|
| 3419 |
+
"loss": 0.4214,
|
| 3420 |
+
"step": 28050
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 2.29,
|
| 3424 |
+
"learning_rate": 4.734897870491091e-06,
|
| 3425 |
+
"loss": 0.3861,
|
| 3426 |
+
"step": 28100
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 2.29,
|
| 3430 |
+
"learning_rate": 4.707735767057801e-06,
|
| 3431 |
+
"loss": 0.4119,
|
| 3432 |
+
"step": 28150
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 2.3,
|
| 3436 |
+
"learning_rate": 4.680573663624511e-06,
|
| 3437 |
+
"loss": 0.4046,
|
| 3438 |
+
"step": 28200
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 2.3,
|
| 3442 |
+
"learning_rate": 4.653411560191222e-06,
|
| 3443 |
+
"loss": 0.3846,
|
| 3444 |
+
"step": 28250
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 2.31,
|
| 3448 |
+
"learning_rate": 4.6262494567579315e-06,
|
| 3449 |
+
"loss": 0.4105,
|
| 3450 |
+
"step": 28300
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 2.31,
|
| 3454 |
+
"learning_rate": 4.599087353324641e-06,
|
| 3455 |
+
"loss": 0.4202,
|
| 3456 |
+
"step": 28350
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 2.31,
|
| 3460 |
+
"learning_rate": 4.571925249891352e-06,
|
| 3461 |
+
"loss": 0.3995,
|
| 3462 |
+
"step": 28400
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 2.32,
|
| 3466 |
+
"learning_rate": 4.544763146458062e-06,
|
| 3467 |
+
"loss": 0.4221,
|
| 3468 |
+
"step": 28450
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 2.32,
|
| 3472 |
+
"learning_rate": 4.5176010430247726e-06,
|
| 3473 |
+
"loss": 0.4233,
|
| 3474 |
+
"step": 28500
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 2.33,
|
| 3478 |
+
"learning_rate": 4.490438939591482e-06,
|
| 3479 |
+
"loss": 0.4016,
|
| 3480 |
+
"step": 28550
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 2.33,
|
| 3484 |
+
"learning_rate": 4.463276836158192e-06,
|
| 3485 |
+
"loss": 0.3899,
|
| 3486 |
+
"step": 28600
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 2.33,
|
| 3490 |
+
"learning_rate": 4.436114732724903e-06,
|
| 3491 |
+
"loss": 0.4105,
|
| 3492 |
+
"step": 28650
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 2.34,
|
| 3496 |
+
"learning_rate": 4.408952629291613e-06,
|
| 3497 |
+
"loss": 0.3675,
|
| 3498 |
+
"step": 28700
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 2.34,
|
| 3502 |
+
"learning_rate": 4.381790525858323e-06,
|
| 3503 |
+
"loss": 0.3944,
|
| 3504 |
+
"step": 28750
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 2.35,
|
| 3508 |
+
"learning_rate": 4.354628422425033e-06,
|
| 3509 |
+
"loss": 0.4149,
|
| 3510 |
+
"step": 28800
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 2.35,
|
| 3514 |
+
"learning_rate": 4.327466318991743e-06,
|
| 3515 |
+
"loss": 0.3901,
|
| 3516 |
+
"step": 28850
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 2.35,
|
| 3520 |
+
"learning_rate": 4.300304215558454e-06,
|
| 3521 |
+
"loss": 0.3982,
|
| 3522 |
+
"step": 28900
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 2.36,
|
| 3526 |
+
"learning_rate": 4.273142112125163e-06,
|
| 3527 |
+
"loss": 0.4135,
|
| 3528 |
+
"step": 28950
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 2.36,
|
| 3532 |
+
"learning_rate": 4.2459800086918734e-06,
|
| 3533 |
+
"loss": 0.4122,
|
| 3534 |
+
"step": 29000
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 2.37,
|
| 3538 |
+
"learning_rate": 4.218817905258584e-06,
|
| 3539 |
+
"loss": 0.3816,
|
| 3540 |
+
"step": 29050
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 2.37,
|
| 3544 |
+
"learning_rate": 4.191655801825294e-06,
|
| 3545 |
+
"loss": 0.3875,
|
| 3546 |
+
"step": 29100
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 2.38,
|
| 3550 |
+
"learning_rate": 4.164493698392004e-06,
|
| 3551 |
+
"loss": 0.4247,
|
| 3552 |
+
"step": 29150
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 2.38,
|
| 3556 |
+
"learning_rate": 4.137331594958714e-06,
|
| 3557 |
+
"loss": 0.4301,
|
| 3558 |
+
"step": 29200
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 2.38,
|
| 3562 |
+
"learning_rate": 4.110169491525424e-06,
|
| 3563 |
+
"loss": 0.4442,
|
| 3564 |
+
"step": 29250
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 2.39,
|
| 3568 |
+
"learning_rate": 4.083007388092134e-06,
|
| 3569 |
+
"loss": 0.4154,
|
| 3570 |
+
"step": 29300
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 2.39,
|
| 3574 |
+
"learning_rate": 4.055845284658844e-06,
|
| 3575 |
+
"loss": 0.4089,
|
| 3576 |
+
"step": 29350
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 2.4,
|
| 3580 |
+
"learning_rate": 4.028683181225555e-06,
|
| 3581 |
+
"loss": 0.3797,
|
| 3582 |
+
"step": 29400
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 2.4,
|
| 3586 |
+
"learning_rate": 4.0015210777922645e-06,
|
| 3587 |
+
"loss": 0.4433,
|
| 3588 |
+
"step": 29450
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 2.4,
|
| 3592 |
+
"learning_rate": 3.974358974358974e-06,
|
| 3593 |
+
"loss": 0.3922,
|
| 3594 |
+
"step": 29500
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 2.41,
|
| 3598 |
+
"learning_rate": 3.947196870925685e-06,
|
| 3599 |
+
"loss": 0.4504,
|
| 3600 |
+
"step": 29550
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 2.41,
|
| 3604 |
+
"learning_rate": 3.920034767492395e-06,
|
| 3605 |
+
"loss": 0.3998,
|
| 3606 |
+
"step": 29600
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 2.42,
|
| 3610 |
+
"learning_rate": 3.8928726640591055e-06,
|
| 3611 |
+
"loss": 0.3771,
|
| 3612 |
+
"step": 29650
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 2.42,
|
| 3616 |
+
"learning_rate": 3.865710560625815e-06,
|
| 3617 |
+
"loss": 0.4342,
|
| 3618 |
+
"step": 29700
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 2.42,
|
| 3622 |
+
"learning_rate": 3.838548457192525e-06,
|
| 3623 |
+
"loss": 0.4114,
|
| 3624 |
+
"step": 29750
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 2.43,
|
| 3628 |
+
"learning_rate": 3.8113863537592354e-06,
|
| 3629 |
+
"loss": 0.3966,
|
| 3630 |
+
"step": 29800
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 2.43,
|
| 3634 |
+
"learning_rate": 3.7842242503259457e-06,
|
| 3635 |
+
"loss": 0.411,
|
| 3636 |
+
"step": 29850
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 2.44,
|
| 3640 |
+
"learning_rate": 3.7570621468926555e-06,
|
| 3641 |
+
"loss": 0.4119,
|
| 3642 |
+
"step": 29900
|
| 3643 |
+
},
|
| 3644 |
+
{
|
| 3645 |
+
"epoch": 2.44,
|
| 3646 |
+
"learning_rate": 3.7299000434593658e-06,
|
| 3647 |
+
"loss": 0.4471,
|
| 3648 |
+
"step": 29950
|
| 3649 |
+
},
|
| 3650 |
+
{
|
| 3651 |
+
"epoch": 2.44,
|
| 3652 |
+
"learning_rate": 3.702737940026076e-06,
|
| 3653 |
+
"loss": 0.3989,
|
| 3654 |
+
"step": 30000
|
| 3655 |
+
},
|
| 3656 |
+
{
|
| 3657 |
+
"epoch": 2.44,
|
| 3658 |
+
"eval_accuracy": 0.7892002037697402,
|
| 3659 |
+
"eval_loss": 0.5580592751502991,
|
| 3660 |
+
"eval_runtime": 16.7363,
|
| 3661 |
+
"eval_samples_per_second": 586.449,
|
| 3662 |
+
"eval_steps_per_second": 36.687,
|
| 3663 |
+
"step": 30000
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 2.45,
|
| 3667 |
+
"learning_rate": 3.6755758365927863e-06,
|
| 3668 |
+
"loss": 0.3757,
|
| 3669 |
+
"step": 30050
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 2.45,
|
| 3673 |
+
"learning_rate": 3.648413733159496e-06,
|
| 3674 |
+
"loss": 0.3962,
|
| 3675 |
+
"step": 30100
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 2.46,
|
| 3679 |
+
"learning_rate": 3.6212516297262064e-06,
|
| 3680 |
+
"loss": 0.4181,
|
| 3681 |
+
"step": 30150
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 2.46,
|
| 3685 |
+
"learning_rate": 3.5940895262929166e-06,
|
| 3686 |
+
"loss": 0.4102,
|
| 3687 |
+
"step": 30200
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 2.46,
|
| 3691 |
+
"learning_rate": 3.566927422859627e-06,
|
| 3692 |
+
"loss": 0.4232,
|
| 3693 |
+
"step": 30250
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 2.47,
|
| 3697 |
+
"learning_rate": 3.5397653194263363e-06,
|
| 3698 |
+
"loss": 0.4225,
|
| 3699 |
+
"step": 30300
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 2.47,
|
| 3703 |
+
"learning_rate": 3.5126032159930466e-06,
|
| 3704 |
+
"loss": 0.4211,
|
| 3705 |
+
"step": 30350
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 2.48,
|
| 3709 |
+
"learning_rate": 3.4854411125597572e-06,
|
| 3710 |
+
"loss": 0.4046,
|
| 3711 |
+
"step": 30400
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 2.48,
|
| 3715 |
+
"learning_rate": 3.4582790091264675e-06,
|
| 3716 |
+
"loss": 0.424,
|
| 3717 |
+
"step": 30450
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 2.49,
|
| 3721 |
+
"learning_rate": 3.431116905693177e-06,
|
| 3722 |
+
"loss": 0.4039,
|
| 3723 |
+
"step": 30500
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 2.49,
|
| 3727 |
+
"learning_rate": 3.403954802259887e-06,
|
| 3728 |
+
"loss": 0.4225,
|
| 3729 |
+
"step": 30550
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 2.49,
|
| 3733 |
+
"learning_rate": 3.3767926988265974e-06,
|
| 3734 |
+
"loss": 0.3794,
|
| 3735 |
+
"step": 30600
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 2.5,
|
| 3739 |
+
"learning_rate": 3.3496305953933073e-06,
|
| 3740 |
+
"loss": 0.4136,
|
| 3741 |
+
"step": 30650
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 2.5,
|
| 3745 |
+
"learning_rate": 3.3224684919600175e-06,
|
| 3746 |
+
"loss": 0.3917,
|
| 3747 |
+
"step": 30700
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 2.51,
|
| 3751 |
+
"learning_rate": 3.2953063885267278e-06,
|
| 3752 |
+
"loss": 0.3823,
|
| 3753 |
+
"step": 30750
|
| 3754 |
+
},
|
| 3755 |
+
{
|
| 3756 |
+
"epoch": 2.51,
|
| 3757 |
+
"learning_rate": 3.268144285093438e-06,
|
| 3758 |
+
"loss": 0.3955,
|
| 3759 |
+
"step": 30800
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 2.51,
|
| 3763 |
+
"learning_rate": 3.240982181660148e-06,
|
| 3764 |
+
"loss": 0.4004,
|
| 3765 |
+
"step": 30850
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 2.52,
|
| 3769 |
+
"learning_rate": 3.213820078226858e-06,
|
| 3770 |
+
"loss": 0.4052,
|
| 3771 |
+
"step": 30900
|
| 3772 |
+
},
|
| 3773 |
+
{
|
| 3774 |
+
"epoch": 2.52,
|
| 3775 |
+
"learning_rate": 3.1866579747935684e-06,
|
| 3776 |
+
"loss": 0.4154,
|
| 3777 |
+
"step": 30950
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 2.53,
|
| 3781 |
+
"learning_rate": 3.1594958713602786e-06,
|
| 3782 |
+
"loss": 0.3727,
|
| 3783 |
+
"step": 31000
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 2.53,
|
| 3787 |
+
"learning_rate": 3.1323337679269885e-06,
|
| 3788 |
+
"loss": 0.3971,
|
| 3789 |
+
"step": 31050
|
| 3790 |
+
},
|
| 3791 |
+
{
|
| 3792 |
+
"epoch": 2.53,
|
| 3793 |
+
"learning_rate": 3.1051716644936987e-06,
|
| 3794 |
+
"loss": 0.3984,
|
| 3795 |
+
"step": 31100
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 2.54,
|
| 3799 |
+
"learning_rate": 3.078009561060409e-06,
|
| 3800 |
+
"loss": 0.3918,
|
| 3801 |
+
"step": 31150
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 2.54,
|
| 3805 |
+
"learning_rate": 3.0508474576271192e-06,
|
| 3806 |
+
"loss": 0.4079,
|
| 3807 |
+
"step": 31200
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 2.55,
|
| 3811 |
+
"learning_rate": 3.0236853541938286e-06,
|
| 3812 |
+
"loss": 0.3879,
|
| 3813 |
+
"step": 31250
|
| 3814 |
+
},
|
| 3815 |
+
{
|
| 3816 |
+
"epoch": 2.55,
|
| 3817 |
+
"learning_rate": 2.996523250760539e-06,
|
| 3818 |
+
"loss": 0.3808,
|
| 3819 |
+
"step": 31300
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 2.55,
|
| 3823 |
+
"learning_rate": 2.969361147327249e-06,
|
| 3824 |
+
"loss": 0.4068,
|
| 3825 |
+
"step": 31350
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 2.56,
|
| 3829 |
+
"learning_rate": 2.9421990438939594e-06,
|
| 3830 |
+
"loss": 0.4029,
|
| 3831 |
+
"step": 31400
|
| 3832 |
+
},
|
| 3833 |
+
{
|
| 3834 |
+
"epoch": 2.56,
|
| 3835 |
+
"learning_rate": 2.9150369404606692e-06,
|
| 3836 |
+
"loss": 0.3524,
|
| 3837 |
+
"step": 31450
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 2.57,
|
| 3841 |
+
"learning_rate": 2.8878748370273795e-06,
|
| 3842 |
+
"loss": 0.4021,
|
| 3843 |
+
"step": 31500
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 2.57,
|
| 3847 |
+
"learning_rate": 2.8607127335940898e-06,
|
| 3848 |
+
"loss": 0.3924,
|
| 3849 |
+
"step": 31550
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 2.57,
|
| 3853 |
+
"learning_rate": 2.8335506301608e-06,
|
| 3854 |
+
"loss": 0.4303,
|
| 3855 |
+
"step": 31600
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 2.58,
|
| 3859 |
+
"learning_rate": 2.80638852672751e-06,
|
| 3860 |
+
"loss": 0.3993,
|
| 3861 |
+
"step": 31650
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 2.58,
|
| 3865 |
+
"learning_rate": 2.77922642329422e-06,
|
| 3866 |
+
"loss": 0.3992,
|
| 3867 |
+
"step": 31700
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 2.59,
|
| 3871 |
+
"learning_rate": 2.7520643198609304e-06,
|
| 3872 |
+
"loss": 0.4215,
|
| 3873 |
+
"step": 31750
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 2.59,
|
| 3877 |
+
"learning_rate": 2.7249022164276406e-06,
|
| 3878 |
+
"loss": 0.4107,
|
| 3879 |
+
"step": 31800
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 2.6,
|
| 3883 |
+
"learning_rate": 2.6977401129943505e-06,
|
| 3884 |
+
"loss": 0.3876,
|
| 3885 |
+
"step": 31850
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 2.6,
|
| 3889 |
+
"learning_rate": 2.6705780095610607e-06,
|
| 3890 |
+
"loss": 0.3873,
|
| 3891 |
+
"step": 31900
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 2.6,
|
| 3895 |
+
"learning_rate": 2.643415906127771e-06,
|
| 3896 |
+
"loss": 0.4126,
|
| 3897 |
+
"step": 31950
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 2.61,
|
| 3901 |
+
"learning_rate": 2.6162538026944812e-06,
|
| 3902 |
+
"loss": 0.4173,
|
| 3903 |
+
"step": 32000
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 2.61,
|
| 3907 |
+
"learning_rate": 2.589091699261191e-06,
|
| 3908 |
+
"loss": 0.409,
|
| 3909 |
+
"step": 32050
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 2.62,
|
| 3913 |
+
"learning_rate": 2.5619295958279013e-06,
|
| 3914 |
+
"loss": 0.4129,
|
| 3915 |
+
"step": 32100
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 2.62,
|
| 3919 |
+
"learning_rate": 2.5347674923946116e-06,
|
| 3920 |
+
"loss": 0.3676,
|
| 3921 |
+
"step": 32150
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 2.62,
|
| 3925 |
+
"learning_rate": 2.507605388961321e-06,
|
| 3926 |
+
"loss": 0.3939,
|
| 3927 |
+
"step": 32200
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 2.63,
|
| 3931 |
+
"learning_rate": 2.4804432855280312e-06,
|
| 3932 |
+
"loss": 0.3828,
|
| 3933 |
+
"step": 32250
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 2.63,
|
| 3937 |
+
"learning_rate": 2.4532811820947415e-06,
|
| 3938 |
+
"loss": 0.418,
|
| 3939 |
+
"step": 32300
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 2.64,
|
| 3943 |
+
"learning_rate": 2.4261190786614517e-06,
|
| 3944 |
+
"loss": 0.4142,
|
| 3945 |
+
"step": 32350
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 2.64,
|
| 3949 |
+
"learning_rate": 2.398956975228162e-06,
|
| 3950 |
+
"loss": 0.398,
|
| 3951 |
+
"step": 32400
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 2.64,
|
| 3955 |
+
"learning_rate": 2.371794871794872e-06,
|
| 3956 |
+
"loss": 0.4077,
|
| 3957 |
+
"step": 32450
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 2.65,
|
| 3961 |
+
"learning_rate": 2.344632768361582e-06,
|
| 3962 |
+
"loss": 0.412,
|
| 3963 |
+
"step": 32500
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 2.65,
|
| 3967 |
+
"learning_rate": 2.3174706649282924e-06,
|
| 3968 |
+
"loss": 0.4163,
|
| 3969 |
+
"step": 32550
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 2.66,
|
| 3973 |
+
"learning_rate": 2.2903085614950026e-06,
|
| 3974 |
+
"loss": 0.4202,
|
| 3975 |
+
"step": 32600
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 2.66,
|
| 3979 |
+
"learning_rate": 2.2631464580617124e-06,
|
| 3980 |
+
"loss": 0.4121,
|
| 3981 |
+
"step": 32650
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 2.66,
|
| 3985 |
+
"learning_rate": 2.2359843546284227e-06,
|
| 3986 |
+
"loss": 0.4,
|
| 3987 |
+
"step": 32700
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 2.67,
|
| 3991 |
+
"learning_rate": 2.2088222511951325e-06,
|
| 3992 |
+
"loss": 0.4008,
|
| 3993 |
+
"step": 32750
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 2.67,
|
| 3997 |
+
"learning_rate": 2.1816601477618428e-06,
|
| 3998 |
+
"loss": 0.4103,
|
| 3999 |
+
"step": 32800
|
| 4000 |
+
},
|
| 4001 |
+
{
|
| 4002 |
+
"epoch": 2.68,
|
| 4003 |
+
"learning_rate": 2.154498044328553e-06,
|
| 4004 |
+
"loss": 0.3995,
|
| 4005 |
+
"step": 32850
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 2.68,
|
| 4009 |
+
"learning_rate": 2.1273359408952633e-06,
|
| 4010 |
+
"loss": 0.3924,
|
| 4011 |
+
"step": 32900
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 2.68,
|
| 4015 |
+
"learning_rate": 2.100173837461973e-06,
|
| 4016 |
+
"loss": 0.431,
|
| 4017 |
+
"step": 32950
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 2.69,
|
| 4021 |
+
"learning_rate": 2.0730117340286834e-06,
|
| 4022 |
+
"loss": 0.4029,
|
| 4023 |
+
"step": 33000
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 2.69,
|
| 4027 |
+
"learning_rate": 2.0458496305953932e-06,
|
| 4028 |
+
"loss": 0.3988,
|
| 4029 |
+
"step": 33050
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 2.7,
|
| 4033 |
+
"learning_rate": 2.018687527162104e-06,
|
| 4034 |
+
"loss": 0.4012,
|
| 4035 |
+
"step": 33100
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 2.7,
|
| 4039 |
+
"learning_rate": 1.9915254237288137e-06,
|
| 4040 |
+
"loss": 0.3842,
|
| 4041 |
+
"step": 33150
|
| 4042 |
+
},
|
| 4043 |
+
{
|
| 4044 |
+
"epoch": 2.71,
|
| 4045 |
+
"learning_rate": 1.964363320295524e-06,
|
| 4046 |
+
"loss": 0.3976,
|
| 4047 |
+
"step": 33200
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 2.71,
|
| 4051 |
+
"learning_rate": 1.937201216862234e-06,
|
| 4052 |
+
"loss": 0.4081,
|
| 4053 |
+
"step": 33250
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 2.71,
|
| 4057 |
+
"learning_rate": 1.910039113428944e-06,
|
| 4058 |
+
"loss": 0.3973,
|
| 4059 |
+
"step": 33300
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 2.72,
|
| 4063 |
+
"learning_rate": 1.8828770099956541e-06,
|
| 4064 |
+
"loss": 0.4087,
|
| 4065 |
+
"step": 33350
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 2.72,
|
| 4069 |
+
"learning_rate": 1.8557149065623644e-06,
|
| 4070 |
+
"loss": 0.387,
|
| 4071 |
+
"step": 33400
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 2.73,
|
| 4075 |
+
"learning_rate": 1.8285528031290744e-06,
|
| 4076 |
+
"loss": 0.3997,
|
| 4077 |
+
"step": 33450
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 2.73,
|
| 4081 |
+
"learning_rate": 1.8013906996957845e-06,
|
| 4082 |
+
"loss": 0.3971,
|
| 4083 |
+
"step": 33500
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 2.73,
|
| 4087 |
+
"learning_rate": 1.7742285962624947e-06,
|
| 4088 |
+
"loss": 0.3768,
|
| 4089 |
+
"step": 33550
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 2.74,
|
| 4093 |
+
"learning_rate": 1.7470664928292048e-06,
|
| 4094 |
+
"loss": 0.364,
|
| 4095 |
+
"step": 33600
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 2.74,
|
| 4099 |
+
"learning_rate": 1.719904389395915e-06,
|
| 4100 |
+
"loss": 0.4112,
|
| 4101 |
+
"step": 33650
|
| 4102 |
+
},
|
| 4103 |
+
{
|
| 4104 |
+
"epoch": 2.75,
|
| 4105 |
+
"learning_rate": 1.6927422859626249e-06,
|
| 4106 |
+
"loss": 0.4038,
|
| 4107 |
+
"step": 33700
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 2.75,
|
| 4111 |
+
"learning_rate": 1.6655801825293353e-06,
|
| 4112 |
+
"loss": 0.365,
|
| 4113 |
+
"step": 33750
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 2.75,
|
| 4117 |
+
"learning_rate": 1.6384180790960452e-06,
|
| 4118 |
+
"loss": 0.4264,
|
| 4119 |
+
"step": 33800
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 2.76,
|
| 4123 |
+
"learning_rate": 1.6112559756627554e-06,
|
| 4124 |
+
"loss": 0.4099,
|
| 4125 |
+
"step": 33850
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 2.76,
|
| 4129 |
+
"learning_rate": 1.5840938722294655e-06,
|
| 4130 |
+
"loss": 0.3823,
|
| 4131 |
+
"step": 33900
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 2.77,
|
| 4135 |
+
"learning_rate": 1.5569317687961757e-06,
|
| 4136 |
+
"loss": 0.427,
|
| 4137 |
+
"step": 33950
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 2.77,
|
| 4141 |
+
"learning_rate": 1.5297696653628858e-06,
|
| 4142 |
+
"loss": 0.3966,
|
| 4143 |
+
"step": 34000
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 2.77,
|
| 4147 |
+
"learning_rate": 1.502607561929596e-06,
|
| 4148 |
+
"loss": 0.4186,
|
| 4149 |
+
"step": 34050
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 2.78,
|
| 4153 |
+
"learning_rate": 1.475445458496306e-06,
|
| 4154 |
+
"loss": 0.3757,
|
| 4155 |
+
"step": 34100
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 2.78,
|
| 4159 |
+
"learning_rate": 1.4482833550630163e-06,
|
| 4160 |
+
"loss": 0.3756,
|
| 4161 |
+
"step": 34150
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 2.79,
|
| 4165 |
+
"learning_rate": 1.4211212516297262e-06,
|
| 4166 |
+
"loss": 0.3846,
|
| 4167 |
+
"step": 34200
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 2.79,
|
| 4171 |
+
"learning_rate": 1.3939591481964364e-06,
|
| 4172 |
+
"loss": 0.425,
|
| 4173 |
+
"step": 34250
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 2.79,
|
| 4177 |
+
"learning_rate": 1.3667970447631465e-06,
|
| 4178 |
+
"loss": 0.4,
|
| 4179 |
+
"step": 34300
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 2.8,
|
| 4183 |
+
"learning_rate": 1.3396349413298567e-06,
|
| 4184 |
+
"loss": 0.4014,
|
| 4185 |
+
"step": 34350
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 2.8,
|
| 4189 |
+
"learning_rate": 1.3124728378965668e-06,
|
| 4190 |
+
"loss": 0.4189,
|
| 4191 |
+
"step": 34400
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 2.81,
|
| 4195 |
+
"learning_rate": 1.285310734463277e-06,
|
| 4196 |
+
"loss": 0.4223,
|
| 4197 |
+
"step": 34450
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 2.81,
|
| 4201 |
+
"learning_rate": 1.258148631029987e-06,
|
| 4202 |
+
"loss": 0.3971,
|
| 4203 |
+
"step": 34500
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 2.82,
|
| 4207 |
+
"learning_rate": 1.2309865275966971e-06,
|
| 4208 |
+
"loss": 0.4106,
|
| 4209 |
+
"step": 34550
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 2.82,
|
| 4213 |
+
"learning_rate": 1.2038244241634074e-06,
|
| 4214 |
+
"loss": 0.4021,
|
| 4215 |
+
"step": 34600
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 2.82,
|
| 4219 |
+
"learning_rate": 1.1766623207301174e-06,
|
| 4220 |
+
"loss": 0.4022,
|
| 4221 |
+
"step": 34650
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 2.83,
|
| 4225 |
+
"learning_rate": 1.1495002172968275e-06,
|
| 4226 |
+
"loss": 0.3845,
|
| 4227 |
+
"step": 34700
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 2.83,
|
| 4231 |
+
"learning_rate": 1.1223381138635377e-06,
|
| 4232 |
+
"loss": 0.3963,
|
| 4233 |
+
"step": 34750
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 2.84,
|
| 4237 |
+
"learning_rate": 1.0951760104302478e-06,
|
| 4238 |
+
"loss": 0.4259,
|
| 4239 |
+
"step": 34800
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 2.84,
|
| 4243 |
+
"learning_rate": 1.068013906996958e-06,
|
| 4244 |
+
"loss": 0.434,
|
| 4245 |
+
"step": 34850
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 2.84,
|
| 4249 |
+
"learning_rate": 1.040851803563668e-06,
|
| 4250 |
+
"loss": 0.3829,
|
| 4251 |
+
"step": 34900
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 2.85,
|
| 4255 |
+
"learning_rate": 1.0136897001303781e-06,
|
| 4256 |
+
"loss": 0.3953,
|
| 4257 |
+
"step": 34950
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 2.85,
|
| 4261 |
+
"learning_rate": 9.865275966970884e-07,
|
| 4262 |
+
"loss": 0.4013,
|
| 4263 |
+
"step": 35000
|
| 4264 |
+
},
|
| 4265 |
+
{
|
| 4266 |
+
"epoch": 2.85,
|
| 4267 |
+
"eval_accuracy": 0.7955170657157412,
|
| 4268 |
+
"eval_loss": 0.5501273274421692,
|
| 4269 |
+
"eval_runtime": 16.8148,
|
| 4270 |
+
"eval_samples_per_second": 583.712,
|
| 4271 |
+
"eval_steps_per_second": 36.515,
|
| 4272 |
+
"step": 35000
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 2.86,
|
| 4276 |
+
"learning_rate": 9.593654932637984e-07,
|
| 4277 |
+
"loss": 0.3919,
|
| 4278 |
+
"step": 35050
|
| 4279 |
+
},
|
| 4280 |
+
{
|
| 4281 |
+
"epoch": 2.86,
|
| 4282 |
+
"learning_rate": 9.322033898305086e-07,
|
| 4283 |
+
"loss": 0.4299,
|
| 4284 |
+
"step": 35100
|
| 4285 |
+
},
|
| 4286 |
+
{
|
| 4287 |
+
"epoch": 2.86,
|
| 4288 |
+
"learning_rate": 9.050412863972187e-07,
|
| 4289 |
+
"loss": 0.4219,
|
| 4290 |
+
"step": 35150
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 2.87,
|
| 4294 |
+
"learning_rate": 8.778791829639289e-07,
|
| 4295 |
+
"loss": 0.3997,
|
| 4296 |
+
"step": 35200
|
| 4297 |
+
},
|
| 4298 |
+
{
|
| 4299 |
+
"epoch": 2.87,
|
| 4300 |
+
"learning_rate": 8.507170795306389e-07,
|
| 4301 |
+
"loss": 0.4138,
|
| 4302 |
+
"step": 35250
|
| 4303 |
+
},
|
| 4304 |
+
{
|
| 4305 |
+
"epoch": 2.88,
|
| 4306 |
+
"learning_rate": 8.235549760973491e-07,
|
| 4307 |
+
"loss": 0.3842,
|
| 4308 |
+
"step": 35300
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 2.88,
|
| 4312 |
+
"learning_rate": 7.963928726640592e-07,
|
| 4313 |
+
"loss": 0.4033,
|
| 4314 |
+
"step": 35350
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 2.88,
|
| 4318 |
+
"learning_rate": 7.692307692307694e-07,
|
| 4319 |
+
"loss": 0.4052,
|
| 4320 |
+
"step": 35400
|
| 4321 |
+
},
|
| 4322 |
+
{
|
| 4323 |
+
"epoch": 2.89,
|
| 4324 |
+
"learning_rate": 7.420686657974795e-07,
|
| 4325 |
+
"loss": 0.4042,
|
| 4326 |
+
"step": 35450
|
| 4327 |
+
},
|
| 4328 |
+
{
|
| 4329 |
+
"epoch": 2.89,
|
| 4330 |
+
"learning_rate": 7.149065623641896e-07,
|
| 4331 |
+
"loss": 0.4045,
|
| 4332 |
+
"step": 35500
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 2.9,
|
| 4336 |
+
"learning_rate": 6.877444589308997e-07,
|
| 4337 |
+
"loss": 0.4073,
|
| 4338 |
+
"step": 35550
|
| 4339 |
+
},
|
| 4340 |
+
{
|
| 4341 |
+
"epoch": 2.9,
|
| 4342 |
+
"learning_rate": 6.605823554976099e-07,
|
| 4343 |
+
"loss": 0.4266,
|
| 4344 |
+
"step": 35600
|
| 4345 |
+
},
|
| 4346 |
+
{
|
| 4347 |
+
"epoch": 2.9,
|
| 4348 |
+
"learning_rate": 6.3342025206432e-07,
|
| 4349 |
+
"loss": 0.3942,
|
| 4350 |
+
"step": 35650
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 2.91,
|
| 4354 |
+
"learning_rate": 6.0625814863103e-07,
|
| 4355 |
+
"loss": 0.4103,
|
| 4356 |
+
"step": 35700
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 2.91,
|
| 4360 |
+
"learning_rate": 5.790960451977402e-07,
|
| 4361 |
+
"loss": 0.4249,
|
| 4362 |
+
"step": 35750
|
| 4363 |
+
},
|
| 4364 |
+
{
|
| 4365 |
+
"epoch": 2.92,
|
| 4366 |
+
"learning_rate": 5.519339417644502e-07,
|
| 4367 |
+
"loss": 0.4074,
|
| 4368 |
+
"step": 35800
|
| 4369 |
+
},
|
| 4370 |
+
{
|
| 4371 |
+
"epoch": 2.92,
|
| 4372 |
+
"learning_rate": 5.247718383311604e-07,
|
| 4373 |
+
"loss": 0.4009,
|
| 4374 |
+
"step": 35850
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 2.93,
|
| 4378 |
+
"learning_rate": 4.976097348978705e-07,
|
| 4379 |
+
"loss": 0.3856,
|
| 4380 |
+
"step": 35900
|
| 4381 |
+
},
|
| 4382 |
+
{
|
| 4383 |
+
"epoch": 2.93,
|
| 4384 |
+
"learning_rate": 4.704476314645807e-07,
|
| 4385 |
+
"loss": 0.3941,
|
| 4386 |
+
"step": 35950
|
| 4387 |
+
},
|
| 4388 |
+
{
|
| 4389 |
+
"epoch": 2.93,
|
| 4390 |
+
"learning_rate": 4.432855280312908e-07,
|
| 4391 |
+
"loss": 0.4133,
|
| 4392 |
+
"step": 36000
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 2.94,
|
| 4396 |
+
"learning_rate": 4.1612342459800084e-07,
|
| 4397 |
+
"loss": 0.4005,
|
| 4398 |
+
"step": 36050
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 2.94,
|
| 4402 |
+
"learning_rate": 3.88961321164711e-07,
|
| 4403 |
+
"loss": 0.368,
|
| 4404 |
+
"step": 36100
|
| 4405 |
+
},
|
| 4406 |
+
{
|
| 4407 |
+
"epoch": 2.95,
|
| 4408 |
+
"learning_rate": 3.6179921773142114e-07,
|
| 4409 |
+
"loss": 0.4314,
|
| 4410 |
+
"step": 36150
|
| 4411 |
+
},
|
| 4412 |
+
{
|
| 4413 |
+
"epoch": 2.95,
|
| 4414 |
+
"learning_rate": 3.3463711429813124e-07,
|
| 4415 |
+
"loss": 0.4145,
|
| 4416 |
+
"step": 36200
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 2.95,
|
| 4420 |
+
"learning_rate": 3.074750108648414e-07,
|
| 4421 |
+
"loss": 0.4122,
|
| 4422 |
+
"step": 36250
|
| 4423 |
+
},
|
| 4424 |
+
{
|
| 4425 |
+
"epoch": 2.96,
|
| 4426 |
+
"learning_rate": 2.803129074315515e-07,
|
| 4427 |
+
"loss": 0.4063,
|
| 4428 |
+
"step": 36300
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 2.96,
|
| 4432 |
+
"learning_rate": 2.5315080399826164e-07,
|
| 4433 |
+
"loss": 0.415,
|
| 4434 |
+
"step": 36350
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 2.97,
|
| 4438 |
+
"learning_rate": 2.2598870056497177e-07,
|
| 4439 |
+
"loss": 0.3831,
|
| 4440 |
+
"step": 36400
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 2.97,
|
| 4444 |
+
"learning_rate": 1.988265971316819e-07,
|
| 4445 |
+
"loss": 0.3935,
|
| 4446 |
+
"step": 36450
|
| 4447 |
+
},
|
| 4448 |
+
{
|
| 4449 |
+
"epoch": 2.97,
|
| 4450 |
+
"learning_rate": 1.7166449369839202e-07,
|
| 4451 |
+
"loss": 0.3924,
|
| 4452 |
+
"step": 36500
|
| 4453 |
+
},
|
| 4454 |
+
{
|
| 4455 |
+
"epoch": 2.98,
|
| 4456 |
+
"learning_rate": 1.4450239026510214e-07,
|
| 4457 |
+
"loss": 0.375,
|
| 4458 |
+
"step": 36550
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 2.98,
|
| 4462 |
+
"learning_rate": 1.1734028683181226e-07,
|
| 4463 |
+
"loss": 0.4107,
|
| 4464 |
+
"step": 36600
|
| 4465 |
+
},
|
| 4466 |
+
{
|
| 4467 |
+
"epoch": 2.99,
|
| 4468 |
+
"learning_rate": 9.017818339852239e-08,
|
| 4469 |
+
"loss": 0.4141,
|
| 4470 |
+
"step": 36650
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 2.99,
|
| 4474 |
+
"learning_rate": 6.301607996523251e-08,
|
| 4475 |
+
"loss": 0.4065,
|
| 4476 |
+
"step": 36700
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 2.99,
|
| 4480 |
+
"learning_rate": 3.585397653194264e-08,
|
| 4481 |
+
"loss": 0.3686,
|
| 4482 |
+
"step": 36750
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 3.0,
|
| 4486 |
+
"learning_rate": 8.691873098652761e-09,
|
| 4487 |
+
"loss": 0.4018,
|
| 4488 |
+
"step": 36800
|
| 4489 |
+
},
|
| 4490 |
+
{
|
| 4491 |
+
"epoch": 3.0,
|
| 4492 |
+
"step": 36816,
|
| 4493 |
+
"total_flos": 7.237278554846584e+16,
|
| 4494 |
+
"train_loss": 0.5138862587508302,
|
| 4495 |
+
"train_runtime": 6461.7512,
|
| 4496 |
+
"train_samples_per_second": 182.32,
|
| 4497 |
+
"train_steps_per_second": 5.698
|
| 4498 |
+
}
|
| 4499 |
+
],
|
| 4500 |
+
"logging_steps": 50,
|
| 4501 |
+
"max_steps": 36816,
|
| 4502 |
+
"num_input_tokens_seen": 0,
|
| 4503 |
+
"num_train_epochs": 3,
|
| 4504 |
+
"save_steps": 5000,
|
| 4505 |
+
"total_flos": 7.237278554846584e+16,
|
| 4506 |
+
"train_batch_size": 32,
|
| 4507 |
+
"trial_name": null,
|
| 4508 |
+
"trial_params": null
|
| 4509 |
+
}
|