Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +3430 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3430 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9998451612903225,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4843,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.002064516129032258,
|
| 13 |
+
"grad_norm": 2.7178148753284264,
|
| 14 |
+
"learning_rate": 2.0618556701030925e-08,
|
| 15 |
+
"loss": 0.4276,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.004129032258064516,
|
| 20 |
+
"grad_norm": 2.5625384878679336,
|
| 21 |
+
"learning_rate": 4.123711340206185e-08,
|
| 22 |
+
"loss": 0.4189,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.006193548387096774,
|
| 27 |
+
"grad_norm": 2.267409247602629,
|
| 28 |
+
"learning_rate": 6.185567010309278e-08,
|
| 29 |
+
"loss": 0.4065,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.008258064516129033,
|
| 34 |
+
"grad_norm": 1.9532754064444633,
|
| 35 |
+
"learning_rate": 8.24742268041237e-08,
|
| 36 |
+
"loss": 0.3945,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.01032258064516129,
|
| 41 |
+
"grad_norm": 1.585906241951902,
|
| 42 |
+
"learning_rate": 1.0309278350515462e-07,
|
| 43 |
+
"loss": 0.376,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.012387096774193548,
|
| 48 |
+
"grad_norm": 1.2402093228097,
|
| 49 |
+
"learning_rate": 1.2371134020618556e-07,
|
| 50 |
+
"loss": 0.3589,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.014451612903225806,
|
| 55 |
+
"grad_norm": 1.0130242074966187,
|
| 56 |
+
"learning_rate": 1.4432989690721648e-07,
|
| 57 |
+
"loss": 0.3489,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.016516129032258065,
|
| 62 |
+
"grad_norm": 0.8517244513870358,
|
| 63 |
+
"learning_rate": 1.649484536082474e-07,
|
| 64 |
+
"loss": 0.3286,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.01858064516129032,
|
| 69 |
+
"grad_norm": 0.6911175913558477,
|
| 70 |
+
"learning_rate": 1.8556701030927835e-07,
|
| 71 |
+
"loss": 0.3177,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.02064516129032258,
|
| 76 |
+
"grad_norm": 0.604894441447739,
|
| 77 |
+
"learning_rate": 2.0618556701030925e-07,
|
| 78 |
+
"loss": 0.3146,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.02270967741935484,
|
| 83 |
+
"grad_norm": 0.5731491989954443,
|
| 84 |
+
"learning_rate": 2.268041237113402e-07,
|
| 85 |
+
"loss": 0.3156,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.024774193548387096,
|
| 90 |
+
"grad_norm": 0.5622949683109659,
|
| 91 |
+
"learning_rate": 2.474226804123711e-07,
|
| 92 |
+
"loss": 0.3053,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.026838709677419356,
|
| 97 |
+
"grad_norm": 0.5173328385018767,
|
| 98 |
+
"learning_rate": 2.6804123711340204e-07,
|
| 99 |
+
"loss": 0.3021,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.02890322580645161,
|
| 104 |
+
"grad_norm": 0.502383112182359,
|
| 105 |
+
"learning_rate": 2.8865979381443296e-07,
|
| 106 |
+
"loss": 0.3024,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.03096774193548387,
|
| 111 |
+
"grad_norm": 0.5056646956646865,
|
| 112 |
+
"learning_rate": 3.0927835051546394e-07,
|
| 113 |
+
"loss": 0.299,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.03303225806451613,
|
| 118 |
+
"grad_norm": 0.48413086309965675,
|
| 119 |
+
"learning_rate": 3.298969072164948e-07,
|
| 120 |
+
"loss": 0.2947,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.03509677419354839,
|
| 125 |
+
"grad_norm": 0.46867700011688224,
|
| 126 |
+
"learning_rate": 3.5051546391752573e-07,
|
| 127 |
+
"loss": 0.295,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.03716129032258064,
|
| 132 |
+
"grad_norm": 0.463440589569122,
|
| 133 |
+
"learning_rate": 3.711340206185567e-07,
|
| 134 |
+
"loss": 0.2938,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.0392258064516129,
|
| 139 |
+
"grad_norm": 0.46043148110607285,
|
| 140 |
+
"learning_rate": 3.917525773195876e-07,
|
| 141 |
+
"loss": 0.2829,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.04129032258064516,
|
| 146 |
+
"grad_norm": 0.4452081751770784,
|
| 147 |
+
"learning_rate": 4.123711340206185e-07,
|
| 148 |
+
"loss": 0.2914,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.04335483870967742,
|
| 153 |
+
"grad_norm": 0.4389480699132875,
|
| 154 |
+
"learning_rate": 4.3298969072164947e-07,
|
| 155 |
+
"loss": 0.2875,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.04541935483870968,
|
| 160 |
+
"grad_norm": 0.44451224796794,
|
| 161 |
+
"learning_rate": 4.536082474226804e-07,
|
| 162 |
+
"loss": 0.2891,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.04748387096774193,
|
| 167 |
+
"grad_norm": 0.44134552800104987,
|
| 168 |
+
"learning_rate": 4.742268041237113e-07,
|
| 169 |
+
"loss": 0.2824,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.04954838709677419,
|
| 174 |
+
"grad_norm": 0.4447117928290257,
|
| 175 |
+
"learning_rate": 4.948453608247422e-07,
|
| 176 |
+
"loss": 0.2833,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.05161290322580645,
|
| 181 |
+
"grad_norm": 0.42778123230774645,
|
| 182 |
+
"learning_rate": 5.154639175257731e-07,
|
| 183 |
+
"loss": 0.288,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.05367741935483871,
|
| 188 |
+
"grad_norm": 0.4310604168679218,
|
| 189 |
+
"learning_rate": 5.360824742268041e-07,
|
| 190 |
+
"loss": 0.2848,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.05574193548387097,
|
| 195 |
+
"grad_norm": 0.44369498025891935,
|
| 196 |
+
"learning_rate": 5.56701030927835e-07,
|
| 197 |
+
"loss": 0.2781,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.05780645161290322,
|
| 202 |
+
"grad_norm": 0.4352776941276996,
|
| 203 |
+
"learning_rate": 5.773195876288659e-07,
|
| 204 |
+
"loss": 0.2765,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.05987096774193548,
|
| 209 |
+
"grad_norm": 0.43125432618608434,
|
| 210 |
+
"learning_rate": 5.979381443298969e-07,
|
| 211 |
+
"loss": 0.2868,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.06193548387096774,
|
| 216 |
+
"grad_norm": 0.4088432228941457,
|
| 217 |
+
"learning_rate": 6.185567010309279e-07,
|
| 218 |
+
"loss": 0.2767,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.064,
|
| 223 |
+
"grad_norm": 0.40838698266390966,
|
| 224 |
+
"learning_rate": 6.391752577319586e-07,
|
| 225 |
+
"loss": 0.2805,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.06606451612903226,
|
| 230 |
+
"grad_norm": 0.450287079898162,
|
| 231 |
+
"learning_rate": 6.597938144329896e-07,
|
| 232 |
+
"loss": 0.2861,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.06812903225806452,
|
| 237 |
+
"grad_norm": 0.44464718825256144,
|
| 238 |
+
"learning_rate": 6.804123711340206e-07,
|
| 239 |
+
"loss": 0.2806,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.07019354838709678,
|
| 244 |
+
"grad_norm": 0.4173477445677872,
|
| 245 |
+
"learning_rate": 7.010309278350515e-07,
|
| 246 |
+
"loss": 0.2775,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.07225806451612904,
|
| 251 |
+
"grad_norm": 0.4247873010801217,
|
| 252 |
+
"learning_rate": 7.216494845360824e-07,
|
| 253 |
+
"loss": 0.2841,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.07432258064516128,
|
| 258 |
+
"grad_norm": 0.42839718840105573,
|
| 259 |
+
"learning_rate": 7.422680412371134e-07,
|
| 260 |
+
"loss": 0.2805,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.07638709677419354,
|
| 265 |
+
"grad_norm": 0.43388241770267927,
|
| 266 |
+
"learning_rate": 7.628865979381443e-07,
|
| 267 |
+
"loss": 0.2716,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.0784516129032258,
|
| 272 |
+
"grad_norm": 0.4326840727745175,
|
| 273 |
+
"learning_rate": 7.835051546391752e-07,
|
| 274 |
+
"loss": 0.2791,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.08051612903225806,
|
| 279 |
+
"grad_norm": 0.4446793301639306,
|
| 280 |
+
"learning_rate": 8.041237113402062e-07,
|
| 281 |
+
"loss": 0.2754,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.08258064516129032,
|
| 286 |
+
"grad_norm": 0.4088616982234986,
|
| 287 |
+
"learning_rate": 8.24742268041237e-07,
|
| 288 |
+
"loss": 0.2776,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.08464516129032258,
|
| 293 |
+
"grad_norm": 0.4305234378372906,
|
| 294 |
+
"learning_rate": 8.45360824742268e-07,
|
| 295 |
+
"loss": 0.275,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.08670967741935484,
|
| 300 |
+
"grad_norm": 0.4178542921488476,
|
| 301 |
+
"learning_rate": 8.659793814432989e-07,
|
| 302 |
+
"loss": 0.2712,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.0887741935483871,
|
| 307 |
+
"grad_norm": 0.4295628808236906,
|
| 308 |
+
"learning_rate": 8.865979381443298e-07,
|
| 309 |
+
"loss": 0.2783,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.09083870967741936,
|
| 314 |
+
"grad_norm": 0.42059850195120757,
|
| 315 |
+
"learning_rate": 9.072164948453608e-07,
|
| 316 |
+
"loss": 0.2778,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.09290322580645162,
|
| 321 |
+
"grad_norm": 0.41033574022465774,
|
| 322 |
+
"learning_rate": 9.278350515463918e-07,
|
| 323 |
+
"loss": 0.2791,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.09496774193548387,
|
| 328 |
+
"grad_norm": 0.44298494519621506,
|
| 329 |
+
"learning_rate": 9.484536082474226e-07,
|
| 330 |
+
"loss": 0.2791,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.09703225806451612,
|
| 335 |
+
"grad_norm": 0.4272735948398401,
|
| 336 |
+
"learning_rate": 9.690721649484535e-07,
|
| 337 |
+
"loss": 0.277,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.09909677419354838,
|
| 342 |
+
"grad_norm": 0.4425115953914259,
|
| 343 |
+
"learning_rate": 9.896907216494845e-07,
|
| 344 |
+
"loss": 0.2718,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.10116129032258064,
|
| 349 |
+
"grad_norm": 0.43188220880067546,
|
| 350 |
+
"learning_rate": 9.999967520836107e-07,
|
| 351 |
+
"loss": 0.2747,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.1032258064516129,
|
| 356 |
+
"grad_norm": 0.45764678748045984,
|
| 357 |
+
"learning_rate": 9.999707690056706e-07,
|
| 358 |
+
"loss": 0.2829,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.10529032258064516,
|
| 363 |
+
"grad_norm": 0.4165688525515349,
|
| 364 |
+
"learning_rate": 9.999188042000401e-07,
|
| 365 |
+
"loss": 0.2766,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.10735483870967742,
|
| 370 |
+
"grad_norm": 0.3946488524350012,
|
| 371 |
+
"learning_rate": 9.998408603671476e-07,
|
| 372 |
+
"loss": 0.2777,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.10941935483870968,
|
| 377 |
+
"grad_norm": 0.4371786129352002,
|
| 378 |
+
"learning_rate": 9.997369415574612e-07,
|
| 379 |
+
"loss": 0.2698,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.11148387096774194,
|
| 384 |
+
"grad_norm": 0.42687492357393675,
|
| 385 |
+
"learning_rate": 9.996070531712766e-07,
|
| 386 |
+
"loss": 0.272,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.1135483870967742,
|
| 391 |
+
"grad_norm": 0.4388895511669077,
|
| 392 |
+
"learning_rate": 9.99451201958438e-07,
|
| 393 |
+
"loss": 0.2752,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.11561290322580645,
|
| 398 |
+
"grad_norm": 0.4078057120806041,
|
| 399 |
+
"learning_rate": 9.992693960179864e-07,
|
| 400 |
+
"loss": 0.2714,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.1176774193548387,
|
| 405 |
+
"grad_norm": 0.43339623355818996,
|
| 406 |
+
"learning_rate": 9.99061644797739e-07,
|
| 407 |
+
"loss": 0.2794,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.11974193548387096,
|
| 412 |
+
"grad_norm": 0.417329955629982,
|
| 413 |
+
"learning_rate": 9.988279590937983e-07,
|
| 414 |
+
"loss": 0.2748,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.12180645161290322,
|
| 419 |
+
"grad_norm": 0.40923444925031505,
|
| 420 |
+
"learning_rate": 9.985683510499907e-07,
|
| 421 |
+
"loss": 0.2689,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.12387096774193548,
|
| 426 |
+
"grad_norm": 0.42272165886149954,
|
| 427 |
+
"learning_rate": 9.982828341572362e-07,
|
| 428 |
+
"loss": 0.2728,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.12593548387096773,
|
| 433 |
+
"grad_norm": 0.40035391344182425,
|
| 434 |
+
"learning_rate": 9.979714232528463e-07,
|
| 435 |
+
"loss": 0.2676,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.128,
|
| 440 |
+
"grad_norm": 0.41930822994047545,
|
| 441 |
+
"learning_rate": 9.97634134519754e-07,
|
| 442 |
+
"loss": 0.2754,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.13006451612903225,
|
| 447 |
+
"grad_norm": 0.41319545992748746,
|
| 448 |
+
"learning_rate": 9.972709854856716e-07,
|
| 449 |
+
"loss": 0.2747,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.13212903225806452,
|
| 454 |
+
"grad_norm": 0.42917202090046985,
|
| 455 |
+
"learning_rate": 9.968819950221812e-07,
|
| 456 |
+
"loss": 0.2665,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.13419354838709677,
|
| 461 |
+
"grad_norm": 0.41071088182465415,
|
| 462 |
+
"learning_rate": 9.964671833437533e-07,
|
| 463 |
+
"loss": 0.2735,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.13625806451612904,
|
| 468 |
+
"grad_norm": 0.42770202579067185,
|
| 469 |
+
"learning_rate": 9.960265720066961e-07,
|
| 470 |
+
"loss": 0.2655,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.13832258064516129,
|
| 475 |
+
"grad_norm": 0.43110284554769446,
|
| 476 |
+
"learning_rate": 9.95560183908036e-07,
|
| 477 |
+
"loss": 0.2676,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.14038709677419356,
|
| 482 |
+
"grad_norm": 0.4085540494229492,
|
| 483 |
+
"learning_rate": 9.950680432843267e-07,
|
| 484 |
+
"loss": 0.2711,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.1424516129032258,
|
| 489 |
+
"grad_norm": 0.4399277307424629,
|
| 490 |
+
"learning_rate": 9.94550175710391e-07,
|
| 491 |
+
"loss": 0.2714,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.14451612903225808,
|
| 496 |
+
"grad_norm": 0.4022098269924505,
|
| 497 |
+
"learning_rate": 9.940066080979909e-07,
|
| 498 |
+
"loss": 0.2692,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.14658064516129032,
|
| 503 |
+
"grad_norm": 0.4138221392891085,
|
| 504 |
+
"learning_rate": 9.934373686944286e-07,
|
| 505 |
+
"loss": 0.2706,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.14864516129032257,
|
| 510 |
+
"grad_norm": 0.3973539996922592,
|
| 511 |
+
"learning_rate": 9.928424870810804e-07,
|
| 512 |
+
"loss": 0.2678,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.15070967741935484,
|
| 517 |
+
"grad_norm": 0.43845088028338314,
|
| 518 |
+
"learning_rate": 9.922219941718576e-07,
|
| 519 |
+
"loss": 0.2686,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.1527741935483871,
|
| 524 |
+
"grad_norm": 0.4127024501370914,
|
| 525 |
+
"learning_rate": 9.915759222116008e-07,
|
| 526 |
+
"loss": 0.2714,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.15483870967741936,
|
| 531 |
+
"grad_norm": 0.4164519302523563,
|
| 532 |
+
"learning_rate": 9.909043047744044e-07,
|
| 533 |
+
"loss": 0.2761,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.1569032258064516,
|
| 538 |
+
"grad_norm": 0.41330928847821996,
|
| 539 |
+
"learning_rate": 9.902071767618715e-07,
|
| 540 |
+
"loss": 0.27,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.15896774193548388,
|
| 545 |
+
"grad_norm": 0.4049309153933818,
|
| 546 |
+
"learning_rate": 9.894845744013002e-07,
|
| 547 |
+
"loss": 0.2738,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.16103225806451613,
|
| 552 |
+
"grad_norm": 0.4074346893357789,
|
| 553 |
+
"learning_rate": 9.887365352438015e-07,
|
| 554 |
+
"loss": 0.2662,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.1630967741935484,
|
| 559 |
+
"grad_norm": 0.43051997768812966,
|
| 560 |
+
"learning_rate": 9.879630981623475e-07,
|
| 561 |
+
"loss": 0.2762,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.16516129032258065,
|
| 566 |
+
"grad_norm": 0.4649279575587385,
|
| 567 |
+
"learning_rate": 9.871643033497512e-07,
|
| 568 |
+
"loss": 0.2711,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.1672258064516129,
|
| 573 |
+
"grad_norm": 0.40771616358626656,
|
| 574 |
+
"learning_rate": 9.863401923165778e-07,
|
| 575 |
+
"loss": 0.2712,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.16929032258064516,
|
| 580 |
+
"grad_norm": 0.43542803830446203,
|
| 581 |
+
"learning_rate": 9.854908078889881e-07,
|
| 582 |
+
"loss": 0.2696,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.1713548387096774,
|
| 587 |
+
"grad_norm": 0.41224099984054274,
|
| 588 |
+
"learning_rate": 9.846161942065123e-07,
|
| 589 |
+
"loss": 0.27,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.17341935483870968,
|
| 594 |
+
"grad_norm": 0.40015212561755775,
|
| 595 |
+
"learning_rate": 9.837163967197567e-07,
|
| 596 |
+
"loss": 0.2697,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.17548387096774193,
|
| 601 |
+
"grad_norm": 0.4145538327172374,
|
| 602 |
+
"learning_rate": 9.827914621880412e-07,
|
| 603 |
+
"loss": 0.2719,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.1775483870967742,
|
| 608 |
+
"grad_norm": 0.4343877153779256,
|
| 609 |
+
"learning_rate": 9.818414386769702e-07,
|
| 610 |
+
"loss": 0.2724,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.17961290322580645,
|
| 615 |
+
"grad_norm": 0.417050446508891,
|
| 616 |
+
"learning_rate": 9.808663755559345e-07,
|
| 617 |
+
"loss": 0.2685,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.18167741935483872,
|
| 622 |
+
"grad_norm": 0.38732244862938475,
|
| 623 |
+
"learning_rate": 9.798663234955452e-07,
|
| 624 |
+
"loss": 0.2712,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.18374193548387097,
|
| 629 |
+
"grad_norm": 0.4004094471829026,
|
| 630 |
+
"learning_rate": 9.788413344650013e-07,
|
| 631 |
+
"loss": 0.2685,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.18580645161290324,
|
| 636 |
+
"grad_norm": 0.43063415997314575,
|
| 637 |
+
"learning_rate": 9.777914617293884e-07,
|
| 638 |
+
"loss": 0.2625,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.18787096774193549,
|
| 643 |
+
"grad_norm": 0.420456518146896,
|
| 644 |
+
"learning_rate": 9.767167598469108e-07,
|
| 645 |
+
"loss": 0.2697,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.18993548387096773,
|
| 650 |
+
"grad_norm": 0.40480539094398127,
|
| 651 |
+
"learning_rate": 9.756172846660576e-07,
|
| 652 |
+
"loss": 0.2707,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.192,
|
| 657 |
+
"grad_norm": 0.40767383426144205,
|
| 658 |
+
"learning_rate": 9.744930933226977e-07,
|
| 659 |
+
"loss": 0.2725,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.19406451612903225,
|
| 664 |
+
"grad_norm": 0.40584494188825865,
|
| 665 |
+
"learning_rate": 9.73344244237114e-07,
|
| 666 |
+
"loss": 0.27,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.19612903225806452,
|
| 671 |
+
"grad_norm": 0.3945980426880921,
|
| 672 |
+
"learning_rate": 9.721707971109645e-07,
|
| 673 |
+
"loss": 0.2677,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.19819354838709677,
|
| 678 |
+
"grad_norm": 0.4092447523280703,
|
| 679 |
+
"learning_rate": 9.709728129241817e-07,
|
| 680 |
+
"loss": 0.2644,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.20025806451612904,
|
| 685 |
+
"grad_norm": 0.4269916844465542,
|
| 686 |
+
"learning_rate": 9.69750353931803e-07,
|
| 687 |
+
"loss": 0.2666,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.2023225806451613,
|
| 692 |
+
"grad_norm": 0.4269986054049296,
|
| 693 |
+
"learning_rate": 9.685034836607358e-07,
|
| 694 |
+
"loss": 0.274,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.20438709677419356,
|
| 699 |
+
"grad_norm": 0.39811569845801287,
|
| 700 |
+
"learning_rate": 9.672322669064552e-07,
|
| 701 |
+
"loss": 0.2713,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.2064516129032258,
|
| 706 |
+
"grad_norm": 0.4061545807740594,
|
| 707 |
+
"learning_rate": 9.659367697296393e-07,
|
| 708 |
+
"loss": 0.2703,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.20851612903225805,
|
| 713 |
+
"grad_norm": 0.4275991643602954,
|
| 714 |
+
"learning_rate": 9.646170594527327e-07,
|
| 715 |
+
"loss": 0.2667,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.21058064516129033,
|
| 720 |
+
"grad_norm": 0.41304510032617703,
|
| 721 |
+
"learning_rate": 9.632732046564517e-07,
|
| 722 |
+
"loss": 0.2783,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.21264516129032257,
|
| 727 |
+
"grad_norm": 0.4321186888978969,
|
| 728 |
+
"learning_rate": 9.619052751762172e-07,
|
| 729 |
+
"loss": 0.2697,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.21470967741935484,
|
| 734 |
+
"grad_norm": 0.4143847837898666,
|
| 735 |
+
"learning_rate": 9.605133420985276e-07,
|
| 736 |
+
"loss": 0.2646,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.2167741935483871,
|
| 741 |
+
"grad_norm": 0.39735153219116426,
|
| 742 |
+
"learning_rate": 9.590974777572643e-07,
|
| 743 |
+
"loss": 0.2643,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.21883870967741936,
|
| 748 |
+
"grad_norm": 0.4471117773040206,
|
| 749 |
+
"learning_rate": 9.576577557299323e-07,
|
| 750 |
+
"loss": 0.2684,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.2209032258064516,
|
| 755 |
+
"grad_norm": 0.41005490851887794,
|
| 756 |
+
"learning_rate": 9.561942508338368e-07,
|
| 757 |
+
"loss": 0.2724,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.22296774193548388,
|
| 762 |
+
"grad_norm": 0.39495815975173343,
|
| 763 |
+
"learning_rate": 9.547070391221953e-07,
|
| 764 |
+
"loss": 0.2672,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.22503225806451613,
|
| 769 |
+
"grad_norm": 0.41286773837253543,
|
| 770 |
+
"learning_rate": 9.531961978801855e-07,
|
| 771 |
+
"loss": 0.2736,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.2270967741935484,
|
| 776 |
+
"grad_norm": 0.40880461152364334,
|
| 777 |
+
"learning_rate": 9.516618056209291e-07,
|
| 778 |
+
"loss": 0.2629,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.22916129032258065,
|
| 783 |
+
"grad_norm": 0.4082631780670623,
|
| 784 |
+
"learning_rate": 9.501039420814111e-07,
|
| 785 |
+
"loss": 0.2697,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.2312258064516129,
|
| 790 |
+
"grad_norm": 0.42142149331893225,
|
| 791 |
+
"learning_rate": 9.485226882183372e-07,
|
| 792 |
+
"loss": 0.2726,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.23329032258064517,
|
| 797 |
+
"grad_norm": 0.4228180963201584,
|
| 798 |
+
"learning_rate": 9.469181262039256e-07,
|
| 799 |
+
"loss": 0.2735,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.2353548387096774,
|
| 804 |
+
"grad_norm": 0.4068752038231123,
|
| 805 |
+
"learning_rate": 9.45290339421638e-07,
|
| 806 |
+
"loss": 0.2633,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.23741935483870968,
|
| 811 |
+
"grad_norm": 0.4525962035081602,
|
| 812 |
+
"learning_rate": 9.436394124618454e-07,
|
| 813 |
+
"loss": 0.2756,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.23948387096774193,
|
| 818 |
+
"grad_norm": 0.4317406005985711,
|
| 819 |
+
"learning_rate": 9.419654311174329e-07,
|
| 820 |
+
"loss": 0.2649,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.2415483870967742,
|
| 825 |
+
"grad_norm": 0.40937736054209056,
|
| 826 |
+
"learning_rate": 9.40268482379341e-07,
|
| 827 |
+
"loss": 0.2726,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.24361290322580645,
|
| 832 |
+
"grad_norm": 0.4032924973019577,
|
| 833 |
+
"learning_rate": 9.385486544320451e-07,
|
| 834 |
+
"loss": 0.2682,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.24567741935483872,
|
| 839 |
+
"grad_norm": 0.41250674679202737,
|
| 840 |
+
"learning_rate": 9.368060366489732e-07,
|
| 841 |
+
"loss": 0.2636,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.24774193548387097,
|
| 846 |
+
"grad_norm": 0.4110118676412479,
|
| 847 |
+
"learning_rate": 9.350407195878607e-07,
|
| 848 |
+
"loss": 0.2692,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.2498064516129032,
|
| 853 |
+
"grad_norm": 0.4071563287492707,
|
| 854 |
+
"learning_rate": 9.332527949860451e-07,
|
| 855 |
+
"loss": 0.2721,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.25187096774193546,
|
| 860 |
+
"grad_norm": 0.4207844628857985,
|
| 861 |
+
"learning_rate": 9.314423557556986e-07,
|
| 862 |
+
"loss": 0.2707,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.25393548387096776,
|
| 867 |
+
"grad_norm": 0.39530786208260144,
|
| 868 |
+
"learning_rate": 9.296094959789994e-07,
|
| 869 |
+
"loss": 0.2625,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.256,
|
| 874 |
+
"grad_norm": 0.41291391579777315,
|
| 875 |
+
"learning_rate": 9.277543109032433e-07,
|
| 876 |
+
"loss": 0.268,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.25806451612903225,
|
| 881 |
+
"grad_norm": 0.4069311969256028,
|
| 882 |
+
"learning_rate": 9.258768969358933e-07,
|
| 883 |
+
"loss": 0.2691,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.2601290322580645,
|
| 888 |
+
"grad_norm": 0.4266996746716126,
|
| 889 |
+
"learning_rate": 9.2397735163957e-07,
|
| 890 |
+
"loss": 0.264,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.2621935483870968,
|
| 895 |
+
"grad_norm": 0.4168110125071047,
|
| 896 |
+
"learning_rate": 9.220557737269816e-07,
|
| 897 |
+
"loss": 0.2671,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.26425806451612904,
|
| 902 |
+
"grad_norm": 0.402938966442535,
|
| 903 |
+
"learning_rate": 9.201122630557943e-07,
|
| 904 |
+
"loss": 0.2721,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.2663225806451613,
|
| 909 |
+
"grad_norm": 0.4433703754564369,
|
| 910 |
+
"learning_rate": 9.181469206234422e-07,
|
| 911 |
+
"loss": 0.2724,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.26838709677419353,
|
| 916 |
+
"grad_norm": 0.4146953645023786,
|
| 917 |
+
"learning_rate": 9.161598485618803e-07,
|
| 918 |
+
"loss": 0.2686,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.2704516129032258,
|
| 923 |
+
"grad_norm": 0.40920394467675797,
|
| 924 |
+
"learning_rate": 9.141511501322758e-07,
|
| 925 |
+
"loss": 0.2654,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.2725161290322581,
|
| 930 |
+
"grad_norm": 0.4291563961337551,
|
| 931 |
+
"learning_rate": 9.121209297196424e-07,
|
| 932 |
+
"loss": 0.2698,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.2745806451612903,
|
| 937 |
+
"grad_norm": 0.4197741898993113,
|
| 938 |
+
"learning_rate": 9.100692928274158e-07,
|
| 939 |
+
"loss": 0.2678,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.27664516129032257,
|
| 944 |
+
"grad_norm": 0.40761323746825606,
|
| 945 |
+
"learning_rate": 9.079963460719714e-07,
|
| 946 |
+
"loss": 0.2606,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.2787096774193548,
|
| 951 |
+
"grad_norm": 0.4102466153208199,
|
| 952 |
+
"learning_rate": 9.059021971770828e-07,
|
| 953 |
+
"loss": 0.2686,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.2807741935483871,
|
| 958 |
+
"grad_norm": 0.42536876796556666,
|
| 959 |
+
"learning_rate": 9.037869549683252e-07,
|
| 960 |
+
"loss": 0.2619,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.28283870967741936,
|
| 965 |
+
"grad_norm": 0.3915639897213394,
|
| 966 |
+
"learning_rate": 9.016507293674187e-07,
|
| 967 |
+
"loss": 0.27,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.2849032258064516,
|
| 972 |
+
"grad_norm": 0.40399941997235145,
|
| 973 |
+
"learning_rate": 8.994936313865171e-07,
|
| 974 |
+
"loss": 0.2728,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.28696774193548386,
|
| 979 |
+
"grad_norm": 0.39944877551284785,
|
| 980 |
+
"learning_rate": 8.973157731224385e-07,
|
| 981 |
+
"loss": 0.274,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.28903225806451616,
|
| 986 |
+
"grad_norm": 0.4092856245914817,
|
| 987 |
+
"learning_rate": 8.9511726775084e-07,
|
| 988 |
+
"loss": 0.2681,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.2910967741935484,
|
| 993 |
+
"grad_norm": 0.481471630902478,
|
| 994 |
+
"learning_rate": 8.928982295203367e-07,
|
| 995 |
+
"loss": 0.2653,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.29316129032258065,
|
| 1000 |
+
"grad_norm": 0.4198523321425174,
|
| 1001 |
+
"learning_rate": 8.906587737465642e-07,
|
| 1002 |
+
"loss": 0.2683,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.2952258064516129,
|
| 1007 |
+
"grad_norm": 0.4293784170337724,
|
| 1008 |
+
"learning_rate": 8.883990168061863e-07,
|
| 1009 |
+
"loss": 0.2667,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.29729032258064514,
|
| 1014 |
+
"grad_norm": 0.409021082775393,
|
| 1015 |
+
"learning_rate": 8.861190761308472e-07,
|
| 1016 |
+
"loss": 0.2669,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.29935483870967744,
|
| 1021 |
+
"grad_norm": 0.43214579094992417,
|
| 1022 |
+
"learning_rate": 8.838190702010693e-07,
|
| 1023 |
+
"loss": 0.268,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.3014193548387097,
|
| 1028 |
+
"grad_norm": 0.4323768086041837,
|
| 1029 |
+
"learning_rate": 8.814991185400951e-07,
|
| 1030 |
+
"loss": 0.2657,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.30348387096774193,
|
| 1035 |
+
"grad_norm": 0.4140826556545066,
|
| 1036 |
+
"learning_rate": 8.791593417076781e-07,
|
| 1037 |
+
"loss": 0.2616,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.3055483870967742,
|
| 1042 |
+
"grad_norm": 0.412829397246139,
|
| 1043 |
+
"learning_rate": 8.767998612938152e-07,
|
| 1044 |
+
"loss": 0.2679,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.3076129032258065,
|
| 1049 |
+
"grad_norm": 0.42525391109992733,
|
| 1050 |
+
"learning_rate": 8.744207999124301e-07,
|
| 1051 |
+
"loss": 0.272,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.3096774193548387,
|
| 1056 |
+
"grad_norm": 0.41049108158888165,
|
| 1057 |
+
"learning_rate": 8.720222811950004e-07,
|
| 1058 |
+
"loss": 0.262,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.31174193548387097,
|
| 1063 |
+
"grad_norm": 0.40142790659401983,
|
| 1064 |
+
"learning_rate": 8.696044297841334e-07,
|
| 1065 |
+
"loss": 0.2637,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.3138064516129032,
|
| 1070 |
+
"grad_norm": 0.40612502217256863,
|
| 1071 |
+
"learning_rate": 8.671673713270886e-07,
|
| 1072 |
+
"loss": 0.2623,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.31587096774193546,
|
| 1077 |
+
"grad_norm": 0.4174546028220698,
|
| 1078 |
+
"learning_rate": 8.647112324692481e-07,
|
| 1079 |
+
"loss": 0.2648,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.31793548387096776,
|
| 1084 |
+
"grad_norm": 0.4075955189469561,
|
| 1085 |
+
"learning_rate": 8.622361408475361e-07,
|
| 1086 |
+
"loss": 0.2671,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.32,
|
| 1091 |
+
"grad_norm": 0.42757900653877373,
|
| 1092 |
+
"learning_rate": 8.597422250837848e-07,
|
| 1093 |
+
"loss": 0.2692,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.32206451612903225,
|
| 1098 |
+
"grad_norm": 0.4104754908146659,
|
| 1099 |
+
"learning_rate": 8.572296147780515e-07,
|
| 1100 |
+
"loss": 0.2627,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.3241290322580645,
|
| 1105 |
+
"grad_norm": 0.43397940214884867,
|
| 1106 |
+
"learning_rate": 8.546984405018834e-07,
|
| 1107 |
+
"loss": 0.2711,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.3261935483870968,
|
| 1112 |
+
"grad_norm": 0.43370385059956773,
|
| 1113 |
+
"learning_rate": 8.521488337915318e-07,
|
| 1114 |
+
"loss": 0.2731,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.32825806451612904,
|
| 1119 |
+
"grad_norm": 0.44403794562813276,
|
| 1120 |
+
"learning_rate": 8.49580927141117e-07,
|
| 1121 |
+
"loss": 0.2681,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.3303225806451613,
|
| 1126 |
+
"grad_norm": 0.40739732247633187,
|
| 1127 |
+
"learning_rate": 8.469948539957432e-07,
|
| 1128 |
+
"loss": 0.2704,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.33238709677419354,
|
| 1133 |
+
"grad_norm": 0.43367593462825726,
|
| 1134 |
+
"learning_rate": 8.443907487445634e-07,
|
| 1135 |
+
"loss": 0.2622,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.3344516129032258,
|
| 1140 |
+
"grad_norm": 0.40927201300965355,
|
| 1141 |
+
"learning_rate": 8.417687467137964e-07,
|
| 1142 |
+
"loss": 0.2682,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.3365161290322581,
|
| 1147 |
+
"grad_norm": 0.417183568246897,
|
| 1148 |
+
"learning_rate": 8.391289841596934e-07,
|
| 1149 |
+
"loss": 0.2623,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.33858064516129033,
|
| 1154 |
+
"grad_norm": 0.4099669038108638,
|
| 1155 |
+
"learning_rate": 8.364715982614577e-07,
|
| 1156 |
+
"loss": 0.2679,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.3406451612903226,
|
| 1161 |
+
"grad_norm": 0.40232008744229203,
|
| 1162 |
+
"learning_rate": 8.337967271141164e-07,
|
| 1163 |
+
"loss": 0.2686,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.3427096774193548,
|
| 1168 |
+
"grad_norm": 0.39988507473787327,
|
| 1169 |
+
"learning_rate": 8.311045097213431e-07,
|
| 1170 |
+
"loss": 0.274,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.3447741935483871,
|
| 1175 |
+
"grad_norm": 0.4279758729822823,
|
| 1176 |
+
"learning_rate": 8.283950859882358e-07,
|
| 1177 |
+
"loss": 0.2744,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.34683870967741937,
|
| 1182 |
+
"grad_norm": 0.4113391707725698,
|
| 1183 |
+
"learning_rate": 8.256685967140449e-07,
|
| 1184 |
+
"loss": 0.2658,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.3489032258064516,
|
| 1189 |
+
"grad_norm": 0.41525159484662594,
|
| 1190 |
+
"learning_rate": 8.229251835848573e-07,
|
| 1191 |
+
"loss": 0.2655,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.35096774193548386,
|
| 1196 |
+
"grad_norm": 0.4107278132335921,
|
| 1197 |
+
"learning_rate": 8.201649891662334e-07,
|
| 1198 |
+
"loss": 0.269,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.3530322580645161,
|
| 1203 |
+
"grad_norm": 0.43928129973031016,
|
| 1204 |
+
"learning_rate": 8.173881568957986e-07,
|
| 1205 |
+
"loss": 0.2654,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.3550967741935484,
|
| 1210 |
+
"grad_norm": 0.43005570640611046,
|
| 1211 |
+
"learning_rate": 8.145948310757886e-07,
|
| 1212 |
+
"loss": 0.2644,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.35716129032258065,
|
| 1217 |
+
"grad_norm": 0.41366298799946843,
|
| 1218 |
+
"learning_rate": 8.117851568655517e-07,
|
| 1219 |
+
"loss": 0.2664,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.3592258064516129,
|
| 1224 |
+
"grad_norm": 0.4107479182292912,
|
| 1225 |
+
"learning_rate": 8.089592802740039e-07,
|
| 1226 |
+
"loss": 0.2668,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.36129032258064514,
|
| 1231 |
+
"grad_norm": 0.4356758167861173,
|
| 1232 |
+
"learning_rate": 8.061173481520429e-07,
|
| 1233 |
+
"loss": 0.2636,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.36335483870967744,
|
| 1238 |
+
"grad_norm": 0.37618623784174804,
|
| 1239 |
+
"learning_rate": 8.032595081849154e-07,
|
| 1240 |
+
"loss": 0.2614,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.3654193548387097,
|
| 1245 |
+
"grad_norm": 0.41446607257101076,
|
| 1246 |
+
"learning_rate": 8.003859088845436e-07,
|
| 1247 |
+
"loss": 0.2642,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.36748387096774193,
|
| 1252 |
+
"grad_norm": 0.42523528400660005,
|
| 1253 |
+
"learning_rate": 7.974966995818066e-07,
|
| 1254 |
+
"loss": 0.2729,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.3695483870967742,
|
| 1259 |
+
"grad_norm": 0.3875366408745518,
|
| 1260 |
+
"learning_rate": 7.945920304187806e-07,
|
| 1261 |
+
"loss": 0.2667,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.3716129032258065,
|
| 1266 |
+
"grad_norm": 0.4057888088552019,
|
| 1267 |
+
"learning_rate": 7.916720523409366e-07,
|
| 1268 |
+
"loss": 0.2658,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.3736774193548387,
|
| 1273 |
+
"grad_norm": 0.4246975366858289,
|
| 1274 |
+
"learning_rate": 7.887369170892964e-07,
|
| 1275 |
+
"loss": 0.2704,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.37574193548387097,
|
| 1280 |
+
"grad_norm": 0.405560542887072,
|
| 1281 |
+
"learning_rate": 7.857867771925468e-07,
|
| 1282 |
+
"loss": 0.2645,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.3778064516129032,
|
| 1287 |
+
"grad_norm": 0.3976055398193886,
|
| 1288 |
+
"learning_rate": 7.828217859591133e-07,
|
| 1289 |
+
"loss": 0.266,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.37987096774193546,
|
| 1294 |
+
"grad_norm": 0.4071477013255938,
|
| 1295 |
+
"learning_rate": 7.798420974691935e-07,
|
| 1296 |
+
"loss": 0.267,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.38193548387096776,
|
| 1301 |
+
"grad_norm": 0.42926938061270814,
|
| 1302 |
+
"learning_rate": 7.7684786656675e-07,
|
| 1303 |
+
"loss": 0.267,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.384,
|
| 1308 |
+
"grad_norm": 0.40068409678383476,
|
| 1309 |
+
"learning_rate": 7.738392488514627e-07,
|
| 1310 |
+
"loss": 0.2638,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.38606451612903225,
|
| 1315 |
+
"grad_norm": 0.41545866046107405,
|
| 1316 |
+
"learning_rate": 7.70816400670645e-07,
|
| 1317 |
+
"loss": 0.2648,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.3881290322580645,
|
| 1322 |
+
"grad_norm": 0.4023140436496999,
|
| 1323 |
+
"learning_rate": 7.677794791111168e-07,
|
| 1324 |
+
"loss": 0.2637,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.3901935483870968,
|
| 1329 |
+
"grad_norm": 0.4196388286712214,
|
| 1330 |
+
"learning_rate": 7.647286419910426e-07,
|
| 1331 |
+
"loss": 0.2647,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.39225806451612905,
|
| 1336 |
+
"grad_norm": 0.40124208669384503,
|
| 1337 |
+
"learning_rate": 7.616640478517293e-07,
|
| 1338 |
+
"loss": 0.2654,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.3943225806451613,
|
| 1343 |
+
"grad_norm": 0.41126451275454146,
|
| 1344 |
+
"learning_rate": 7.585858559493885e-07,
|
| 1345 |
+
"loss": 0.2631,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.39638709677419354,
|
| 1350 |
+
"grad_norm": 0.41848510984577836,
|
| 1351 |
+
"learning_rate": 7.554942262468593e-07,
|
| 1352 |
+
"loss": 0.2682,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.3984516129032258,
|
| 1357 |
+
"grad_norm": 0.43454563702411014,
|
| 1358 |
+
"learning_rate": 7.523893194052966e-07,
|
| 1359 |
+
"loss": 0.2619,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.4005161290322581,
|
| 1364 |
+
"grad_norm": 0.43881054463213787,
|
| 1365 |
+
"learning_rate": 7.49271296775821e-07,
|
| 1366 |
+
"loss": 0.2664,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.40258064516129033,
|
| 1371 |
+
"grad_norm": 0.4027456476422296,
|
| 1372 |
+
"learning_rate": 7.461403203911355e-07,
|
| 1373 |
+
"loss": 0.2659,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.4046451612903226,
|
| 1378 |
+
"grad_norm": 0.4130317888357524,
|
| 1379 |
+
"learning_rate": 7.429965529571036e-07,
|
| 1380 |
+
"loss": 0.261,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.4067096774193548,
|
| 1385 |
+
"grad_norm": 0.40488753602787214,
|
| 1386 |
+
"learning_rate": 7.398401578442952e-07,
|
| 1387 |
+
"loss": 0.263,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.4087741935483871,
|
| 1392 |
+
"grad_norm": 0.40658072770043274,
|
| 1393 |
+
"learning_rate": 7.366712990794961e-07,
|
| 1394 |
+
"loss": 0.2706,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.41083870967741937,
|
| 1399 |
+
"grad_norm": 0.4417207769552733,
|
| 1400 |
+
"learning_rate": 7.334901413371847e-07,
|
| 1401 |
+
"loss": 0.2695,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.4129032258064516,
|
| 1406 |
+
"grad_norm": 0.40600956756974876,
|
| 1407 |
+
"learning_rate": 7.302968499309737e-07,
|
| 1408 |
+
"loss": 0.265,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.41496774193548386,
|
| 1413 |
+
"grad_norm": 0.421884057260553,
|
| 1414 |
+
"learning_rate": 7.270915908050204e-07,
|
| 1415 |
+
"loss": 0.2631,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.4170322580645161,
|
| 1420 |
+
"grad_norm": 0.4060635608602976,
|
| 1421 |
+
"learning_rate": 7.238745305254019e-07,
|
| 1422 |
+
"loss": 0.2697,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.4190967741935484,
|
| 1427 |
+
"grad_norm": 0.44059203724336116,
|
| 1428 |
+
"learning_rate": 7.206458362714602e-07,
|
| 1429 |
+
"loss": 0.2565,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.42116129032258065,
|
| 1434 |
+
"grad_norm": 0.3968336903067748,
|
| 1435 |
+
"learning_rate": 7.17405675827114e-07,
|
| 1436 |
+
"loss": 0.2717,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.4232258064516129,
|
| 1441 |
+
"grad_norm": 0.41171391357598364,
|
| 1442 |
+
"learning_rate": 7.1415421757214e-07,
|
| 1443 |
+
"loss": 0.2645,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.42529032258064514,
|
| 1448 |
+
"grad_norm": 0.3876345602992099,
|
| 1449 |
+
"learning_rate": 7.10891630473422e-07,
|
| 1450 |
+
"loss": 0.2652,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.42735483870967744,
|
| 1455 |
+
"grad_norm": 0.39450051059803076,
|
| 1456 |
+
"learning_rate": 7.076180840761714e-07,
|
| 1457 |
+
"loss": 0.2632,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.4294193548387097,
|
| 1462 |
+
"grad_norm": 0.4358288881727055,
|
| 1463 |
+
"learning_rate": 7.043337484951154e-07,
|
| 1464 |
+
"loss": 0.2642,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.43148387096774193,
|
| 1469 |
+
"grad_norm": 0.41920035477123707,
|
| 1470 |
+
"learning_rate": 7.010387944056576e-07,
|
| 1471 |
+
"loss": 0.261,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.4335483870967742,
|
| 1476 |
+
"grad_norm": 0.4059103398768672,
|
| 1477 |
+
"learning_rate": 6.97733393035008e-07,
|
| 1478 |
+
"loss": 0.2628,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.4356129032258064,
|
| 1483 |
+
"grad_norm": 0.3982345695440609,
|
| 1484 |
+
"learning_rate": 6.944177161532851e-07,
|
| 1485 |
+
"loss": 0.2597,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.4376774193548387,
|
| 1490 |
+
"grad_norm": 0.4144966185638575,
|
| 1491 |
+
"learning_rate": 6.910919360645902e-07,
|
| 1492 |
+
"loss": 0.27,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.43974193548387097,
|
| 1497 |
+
"grad_norm": 0.4233480513748313,
|
| 1498 |
+
"learning_rate": 6.877562255980519e-07,
|
| 1499 |
+
"loss": 0.2681,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.4418064516129032,
|
| 1504 |
+
"grad_norm": 0.40910710385731447,
|
| 1505 |
+
"learning_rate": 6.844107580988471e-07,
|
| 1506 |
+
"loss": 0.2662,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.44387096774193546,
|
| 1511 |
+
"grad_norm": 0.4069330452268534,
|
| 1512 |
+
"learning_rate": 6.810557074191899e-07,
|
| 1513 |
+
"loss": 0.2633,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.44593548387096776,
|
| 1518 |
+
"grad_norm": 0.4033815804757132,
|
| 1519 |
+
"learning_rate": 6.776912479093001e-07,
|
| 1520 |
+
"loss": 0.2606,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.448,
|
| 1525 |
+
"grad_norm": 0.41960546237510354,
|
| 1526 |
+
"learning_rate": 6.743175544083403e-07,
|
| 1527 |
+
"loss": 0.2623,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.45006451612903225,
|
| 1532 |
+
"grad_norm": 0.41168810646722503,
|
| 1533 |
+
"learning_rate": 6.709348022353318e-07,
|
| 1534 |
+
"loss": 0.264,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.4521290322580645,
|
| 1539 |
+
"grad_norm": 0.40385762651618623,
|
| 1540 |
+
"learning_rate": 6.675431671800436e-07,
|
| 1541 |
+
"loss": 0.2651,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.4541935483870968,
|
| 1546 |
+
"grad_norm": 0.4032346492996413,
|
| 1547 |
+
"learning_rate": 6.64142825493856e-07,
|
| 1548 |
+
"loss": 0.2632,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.45625806451612905,
|
| 1553 |
+
"grad_norm": 0.4053652025972668,
|
| 1554 |
+
"learning_rate": 6.607339538806034e-07,
|
| 1555 |
+
"loss": 0.2608,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.4583225806451613,
|
| 1560 |
+
"grad_norm": 0.41770642881401504,
|
| 1561 |
+
"learning_rate": 6.573167294873898e-07,
|
| 1562 |
+
"loss": 0.2558,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.46038709677419354,
|
| 1567 |
+
"grad_norm": 0.4112896752142016,
|
| 1568 |
+
"learning_rate": 6.538913298953845e-07,
|
| 1569 |
+
"loss": 0.2646,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.4624516129032258,
|
| 1574 |
+
"grad_norm": 0.41305858313874766,
|
| 1575 |
+
"learning_rate": 6.504579331105928e-07,
|
| 1576 |
+
"loss": 0.2661,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.4645161290322581,
|
| 1581 |
+
"grad_norm": 0.3990060429263245,
|
| 1582 |
+
"learning_rate": 6.470167175546063e-07,
|
| 1583 |
+
"loss": 0.2552,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.46658064516129033,
|
| 1588 |
+
"grad_norm": 0.4068327568812493,
|
| 1589 |
+
"learning_rate": 6.435678620553306e-07,
|
| 1590 |
+
"loss": 0.2625,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.4686451612903226,
|
| 1595 |
+
"grad_norm": 0.4122018568807427,
|
| 1596 |
+
"learning_rate": 6.401115458376924e-07,
|
| 1597 |
+
"loss": 0.2561,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.4707096774193548,
|
| 1602 |
+
"grad_norm": 0.4075890652073717,
|
| 1603 |
+
"learning_rate": 6.366479485143257e-07,
|
| 1604 |
+
"loss": 0.2655,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.4727741935483871,
|
| 1609 |
+
"grad_norm": 0.4041531603835199,
|
| 1610 |
+
"learning_rate": 6.331772500762382e-07,
|
| 1611 |
+
"loss": 0.2608,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.47483870967741937,
|
| 1616 |
+
"grad_norm": 0.401347899960604,
|
| 1617 |
+
"learning_rate": 6.296996308834575e-07,
|
| 1618 |
+
"loss": 0.2626,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.4769032258064516,
|
| 1623 |
+
"grad_norm": 0.4046833913093797,
|
| 1624 |
+
"learning_rate": 6.262152716556586e-07,
|
| 1625 |
+
"loss": 0.2645,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.47896774193548386,
|
| 1630 |
+
"grad_norm": 0.4112501171173604,
|
| 1631 |
+
"learning_rate": 6.227243534627724e-07,
|
| 1632 |
+
"loss": 0.2627,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.4810322580645161,
|
| 1637 |
+
"grad_norm": 0.4119800419244549,
|
| 1638 |
+
"learning_rate": 6.192270577155764e-07,
|
| 1639 |
+
"loss": 0.2666,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.4830967741935484,
|
| 1644 |
+
"grad_norm": 0.4067394550798155,
|
| 1645 |
+
"learning_rate": 6.157235661562672e-07,
|
| 1646 |
+
"loss": 0.2609,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.48516129032258065,
|
| 1651 |
+
"grad_norm": 0.4161737935572665,
|
| 1652 |
+
"learning_rate": 6.122140608490157e-07,
|
| 1653 |
+
"loss": 0.2631,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.4872258064516129,
|
| 1658 |
+
"grad_norm": 0.41837478258205407,
|
| 1659 |
+
"learning_rate": 6.086987241705066e-07,
|
| 1660 |
+
"loss": 0.2613,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.48929032258064514,
|
| 1665 |
+
"grad_norm": 0.37786501702152653,
|
| 1666 |
+
"learning_rate": 6.051777388004603e-07,
|
| 1667 |
+
"loss": 0.2659,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.49135483870967744,
|
| 1672 |
+
"grad_norm": 0.4026726334273472,
|
| 1673 |
+
"learning_rate": 6.0165128771214e-07,
|
| 1674 |
+
"loss": 0.2668,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.4934193548387097,
|
| 1679 |
+
"grad_norm": 0.43109902087561547,
|
| 1680 |
+
"learning_rate": 5.981195541628431e-07,
|
| 1681 |
+
"loss": 0.2665,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.49548387096774194,
|
| 1686 |
+
"grad_norm": 0.40800799018925554,
|
| 1687 |
+
"learning_rate": 5.945827216843779e-07,
|
| 1688 |
+
"loss": 0.2637,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.4975483870967742,
|
| 1693 |
+
"grad_norm": 0.40953889079215977,
|
| 1694 |
+
"learning_rate": 5.910409740735263e-07,
|
| 1695 |
+
"loss": 0.267,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.4996129032258064,
|
| 1700 |
+
"grad_norm": 0.40899543891195417,
|
| 1701 |
+
"learning_rate": 5.874944953824919e-07,
|
| 1702 |
+
"loss": 0.2583,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.5016774193548387,
|
| 1707 |
+
"grad_norm": 0.4251510009606146,
|
| 1708 |
+
"learning_rate": 5.83943469909337e-07,
|
| 1709 |
+
"loss": 0.268,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.5037419354838709,
|
| 1714 |
+
"grad_norm": 0.4058826730393694,
|
| 1715 |
+
"learning_rate": 5.803880821884032e-07,
|
| 1716 |
+
"loss": 0.2579,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.5058064516129033,
|
| 1721 |
+
"grad_norm": 0.4240346499686985,
|
| 1722 |
+
"learning_rate": 5.768285169807233e-07,
|
| 1723 |
+
"loss": 0.2618,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.5078709677419355,
|
| 1728 |
+
"grad_norm": 0.4054651824039119,
|
| 1729 |
+
"learning_rate": 5.73264959264419e-07,
|
| 1730 |
+
"loss": 0.2587,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.5099354838709678,
|
| 1735 |
+
"grad_norm": 0.4264297129067109,
|
| 1736 |
+
"learning_rate": 5.696975942250896e-07,
|
| 1737 |
+
"loss": 0.2608,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.512,
|
| 1742 |
+
"grad_norm": 0.38763232793736896,
|
| 1743 |
+
"learning_rate": 5.661266072461866e-07,
|
| 1744 |
+
"loss": 0.2624,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.5140645161290323,
|
| 1749 |
+
"grad_norm": 0.39972658320821614,
|
| 1750 |
+
"learning_rate": 5.625521838993814e-07,
|
| 1751 |
+
"loss": 0.2628,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.5161290322580645,
|
| 1756 |
+
"grad_norm": 0.4094650932180596,
|
| 1757 |
+
"learning_rate": 5.589745099349219e-07,
|
| 1758 |
+
"loss": 0.2654,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.5181935483870967,
|
| 1763 |
+
"grad_norm": 0.40303191617270684,
|
| 1764 |
+
"learning_rate": 5.55393771271978e-07,
|
| 1765 |
+
"loss": 0.2626,
|
| 1766 |
+
"step": 2510
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.520258064516129,
|
| 1770 |
+
"grad_norm": 0.4432634297156684,
|
| 1771 |
+
"learning_rate": 5.518101539889828e-07,
|
| 1772 |
+
"loss": 0.2631,
|
| 1773 |
+
"step": 2520
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.5223225806451612,
|
| 1777 |
+
"grad_norm": 0.4019002372120617,
|
| 1778 |
+
"learning_rate": 5.482238443139597e-07,
|
| 1779 |
+
"loss": 0.2641,
|
| 1780 |
+
"step": 2530
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.5243870967741936,
|
| 1784 |
+
"grad_norm": 0.3922554382852571,
|
| 1785 |
+
"learning_rate": 5.446350286148471e-07,
|
| 1786 |
+
"loss": 0.2588,
|
| 1787 |
+
"step": 2540
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.5264516129032258,
|
| 1791 |
+
"grad_norm": 0.41838300381220644,
|
| 1792 |
+
"learning_rate": 5.410438933898127e-07,
|
| 1793 |
+
"loss": 0.2603,
|
| 1794 |
+
"step": 2550
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.5285161290322581,
|
| 1798 |
+
"grad_norm": 0.4056835013387469,
|
| 1799 |
+
"learning_rate": 5.374506252575612e-07,
|
| 1800 |
+
"loss": 0.2655,
|
| 1801 |
+
"step": 2560
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.5305806451612903,
|
| 1805 |
+
"grad_norm": 0.40888463409102277,
|
| 1806 |
+
"learning_rate": 5.338554109476379e-07,
|
| 1807 |
+
"loss": 0.2661,
|
| 1808 |
+
"step": 2570
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.5326451612903226,
|
| 1812 |
+
"grad_norm": 0.4221212778601083,
|
| 1813 |
+
"learning_rate": 5.302584372907236e-07,
|
| 1814 |
+
"loss": 0.2643,
|
| 1815 |
+
"step": 2580
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.5347096774193548,
|
| 1819 |
+
"grad_norm": 0.40948851318864904,
|
| 1820 |
+
"learning_rate": 5.266598912089259e-07,
|
| 1821 |
+
"loss": 0.2646,
|
| 1822 |
+
"step": 2590
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.5367741935483871,
|
| 1826 |
+
"grad_norm": 0.3945682130480978,
|
| 1827 |
+
"learning_rate": 5.230599597060666e-07,
|
| 1828 |
+
"loss": 0.2622,
|
| 1829 |
+
"step": 2600
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.5388387096774193,
|
| 1833 |
+
"grad_norm": 0.3959558379054268,
|
| 1834 |
+
"learning_rate": 5.194588298579623e-07,
|
| 1835 |
+
"loss": 0.2631,
|
| 1836 |
+
"step": 2610
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.5409032258064516,
|
| 1840 |
+
"grad_norm": 0.4081097575523178,
|
| 1841 |
+
"learning_rate": 5.158566888027038e-07,
|
| 1842 |
+
"loss": 0.2596,
|
| 1843 |
+
"step": 2620
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.5429677419354839,
|
| 1847 |
+
"grad_norm": 0.39268286624553117,
|
| 1848 |
+
"learning_rate": 5.122537237309305e-07,
|
| 1849 |
+
"loss": 0.2604,
|
| 1850 |
+
"step": 2630
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.5450322580645162,
|
| 1854 |
+
"grad_norm": 0.39342604284402627,
|
| 1855 |
+
"learning_rate": 5.086501218761033e-07,
|
| 1856 |
+
"loss": 0.2586,
|
| 1857 |
+
"step": 2640
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.5470967741935484,
|
| 1861 |
+
"grad_norm": 0.40285421355756235,
|
| 1862 |
+
"learning_rate": 5.050460705047742e-07,
|
| 1863 |
+
"loss": 0.2572,
|
| 1864 |
+
"step": 2650
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.5491612903225807,
|
| 1868 |
+
"grad_norm": 0.4020063207626797,
|
| 1869 |
+
"learning_rate": 5.014417569068554e-07,
|
| 1870 |
+
"loss": 0.2643,
|
| 1871 |
+
"step": 2660
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.5512258064516129,
|
| 1875 |
+
"grad_norm": 0.42131953987445797,
|
| 1876 |
+
"learning_rate": 4.978373683858858e-07,
|
| 1877 |
+
"loss": 0.2643,
|
| 1878 |
+
"step": 2670
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.5532903225806451,
|
| 1882 |
+
"grad_norm": 0.3945931863557437,
|
| 1883 |
+
"learning_rate": 4.942330922492975e-07,
|
| 1884 |
+
"loss": 0.2645,
|
| 1885 |
+
"step": 2680
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.5553548387096774,
|
| 1889 |
+
"grad_norm": 0.40345218634673974,
|
| 1890 |
+
"learning_rate": 4.906291157986834e-07,
|
| 1891 |
+
"loss": 0.2629,
|
| 1892 |
+
"step": 2690
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.5574193548387096,
|
| 1896 |
+
"grad_norm": 0.4053650394538655,
|
| 1897 |
+
"learning_rate": 4.870256263200616e-07,
|
| 1898 |
+
"loss": 0.2596,
|
| 1899 |
+
"step": 2700
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.5594838709677419,
|
| 1903 |
+
"grad_norm": 0.38082232410527134,
|
| 1904 |
+
"learning_rate": 4.834228110741447e-07,
|
| 1905 |
+
"loss": 0.2686,
|
| 1906 |
+
"step": 2710
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.5615483870967742,
|
| 1910 |
+
"grad_norm": 0.4244897702153668,
|
| 1911 |
+
"learning_rate": 4.798208572866074e-07,
|
| 1912 |
+
"loss": 0.2654,
|
| 1913 |
+
"step": 2720
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.5636129032258065,
|
| 1917 |
+
"grad_norm": 0.4168979030949182,
|
| 1918 |
+
"learning_rate": 4.762199521383575e-07,
|
| 1919 |
+
"loss": 0.2656,
|
| 1920 |
+
"step": 2730
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.5656774193548387,
|
| 1924 |
+
"grad_norm": 0.4008885860052325,
|
| 1925 |
+
"learning_rate": 4.72620282755809e-07,
|
| 1926 |
+
"loss": 0.2641,
|
| 1927 |
+
"step": 2740
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.567741935483871,
|
| 1931 |
+
"grad_norm": 0.4133977944936882,
|
| 1932 |
+
"learning_rate": 4.690220362011573e-07,
|
| 1933 |
+
"loss": 0.2624,
|
| 1934 |
+
"step": 2750
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.5698064516129032,
|
| 1938 |
+
"grad_norm": 0.39400301829128825,
|
| 1939 |
+
"learning_rate": 4.6542539946265823e-07,
|
| 1940 |
+
"loss": 0.2598,
|
| 1941 |
+
"step": 2760
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.5718709677419355,
|
| 1945 |
+
"grad_norm": 0.40929967482891283,
|
| 1946 |
+
"learning_rate": 4.618305594449113e-07,
|
| 1947 |
+
"loss": 0.263,
|
| 1948 |
+
"step": 2770
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.5739354838709677,
|
| 1952 |
+
"grad_norm": 0.4056920391576304,
|
| 1953 |
+
"learning_rate": 4.5823770295914706e-07,
|
| 1954 |
+
"loss": 0.2595,
|
| 1955 |
+
"step": 2780
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.576,
|
| 1959 |
+
"grad_norm": 0.4107265334626862,
|
| 1960 |
+
"learning_rate": 4.5464701671351815e-07,
|
| 1961 |
+
"loss": 0.2647,
|
| 1962 |
+
"step": 2790
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.5780645161290323,
|
| 1966 |
+
"grad_norm": 0.3913927583282506,
|
| 1967 |
+
"learning_rate": 4.5105868730339785e-07,
|
| 1968 |
+
"loss": 0.2597,
|
| 1969 |
+
"step": 2800
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.5801290322580646,
|
| 1973 |
+
"grad_norm": 0.41022478158711445,
|
| 1974 |
+
"learning_rate": 4.4747290120168316e-07,
|
| 1975 |
+
"loss": 0.2614,
|
| 1976 |
+
"step": 2810
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.5821935483870968,
|
| 1980 |
+
"grad_norm": 0.4048943417693181,
|
| 1981 |
+
"learning_rate": 4.438898447491036e-07,
|
| 1982 |
+
"loss": 0.2623,
|
| 1983 |
+
"step": 2820
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.584258064516129,
|
| 1987 |
+
"grad_norm": 0.41841621193034295,
|
| 1988 |
+
"learning_rate": 4.403097041445395e-07,
|
| 1989 |
+
"loss": 0.2626,
|
| 1990 |
+
"step": 2830
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.5863225806451613,
|
| 1994 |
+
"grad_norm": 0.41332586169270713,
|
| 1995 |
+
"learning_rate": 4.367326654353436e-07,
|
| 1996 |
+
"loss": 0.2619,
|
| 1997 |
+
"step": 2840
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.5883870967741935,
|
| 2001 |
+
"grad_norm": 0.42980292999632475,
|
| 2002 |
+
"learning_rate": 4.331589145076746e-07,
|
| 2003 |
+
"loss": 0.2662,
|
| 2004 |
+
"step": 2850
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.5904516129032258,
|
| 2008 |
+
"grad_norm": 0.4085925990454811,
|
| 2009 |
+
"learning_rate": 4.295886370768367e-07,
|
| 2010 |
+
"loss": 0.2661,
|
| 2011 |
+
"step": 2860
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.592516129032258,
|
| 2015 |
+
"grad_norm": 0.3938359040261475,
|
| 2016 |
+
"learning_rate": 4.2602201867762875e-07,
|
| 2017 |
+
"loss": 0.2549,
|
| 2018 |
+
"step": 2870
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.5945806451612903,
|
| 2022 |
+
"grad_norm": 0.4009963393291789,
|
| 2023 |
+
"learning_rate": 4.224592446547024e-07,
|
| 2024 |
+
"loss": 0.2584,
|
| 2025 |
+
"step": 2880
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.5966451612903226,
|
| 2029 |
+
"grad_norm": 0.40377828376876823,
|
| 2030 |
+
"learning_rate": 4.1890050015293036e-07,
|
| 2031 |
+
"loss": 0.2642,
|
| 2032 |
+
"step": 2890
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 0.5987096774193549,
|
| 2036 |
+
"grad_norm": 0.4084739109095333,
|
| 2037 |
+
"learning_rate": 4.153459701077856e-07,
|
| 2038 |
+
"loss": 0.2637,
|
| 2039 |
+
"step": 2900
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.6007741935483871,
|
| 2043 |
+
"grad_norm": 0.4355917374420026,
|
| 2044 |
+
"learning_rate": 4.117958392357303e-07,
|
| 2045 |
+
"loss": 0.267,
|
| 2046 |
+
"step": 2910
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 0.6028387096774194,
|
| 2050 |
+
"grad_norm": 0.4111357071539257,
|
| 2051 |
+
"learning_rate": 4.082502920246173e-07,
|
| 2052 |
+
"loss": 0.2571,
|
| 2053 |
+
"step": 2920
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 0.6049032258064516,
|
| 2057 |
+
"grad_norm": 0.4300310059373518,
|
| 2058 |
+
"learning_rate": 4.047095127241026e-07,
|
| 2059 |
+
"loss": 0.2663,
|
| 2060 |
+
"step": 2930
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.6069677419354839,
|
| 2064 |
+
"grad_norm": 0.41504350370170773,
|
| 2065 |
+
"learning_rate": 4.011736853360702e-07,
|
| 2066 |
+
"loss": 0.2668,
|
| 2067 |
+
"step": 2940
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.6090322580645161,
|
| 2071 |
+
"grad_norm": 0.39347081849396953,
|
| 2072 |
+
"learning_rate": 3.976429936050709e-07,
|
| 2073 |
+
"loss": 0.262,
|
| 2074 |
+
"step": 2950
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.6110967741935484,
|
| 2078 |
+
"grad_norm": 0.4001140349315572,
|
| 2079 |
+
"learning_rate": 3.941176210087737e-07,
|
| 2080 |
+
"loss": 0.2607,
|
| 2081 |
+
"step": 2960
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.6131612903225806,
|
| 2085 |
+
"grad_norm": 0.41999186019603213,
|
| 2086 |
+
"learning_rate": 3.905977507484304e-07,
|
| 2087 |
+
"loss": 0.2631,
|
| 2088 |
+
"step": 2970
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 0.615225806451613,
|
| 2092 |
+
"grad_norm": 0.40433153562642776,
|
| 2093 |
+
"learning_rate": 3.870835657393558e-07,
|
| 2094 |
+
"loss": 0.2674,
|
| 2095 |
+
"step": 2980
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 0.6172903225806452,
|
| 2099 |
+
"grad_norm": 0.41600570415494387,
|
| 2100 |
+
"learning_rate": 3.8357524860142205e-07,
|
| 2101 |
+
"loss": 0.2633,
|
| 2102 |
+
"step": 2990
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.6193548387096774,
|
| 2106 |
+
"grad_norm": 0.39498755788207757,
|
| 2107 |
+
"learning_rate": 3.8007298164956866e-07,
|
| 2108 |
+
"loss": 0.2677,
|
| 2109 |
+
"step": 3000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.6214193548387097,
|
| 2113 |
+
"grad_norm": 0.4222886993315553,
|
| 2114 |
+
"learning_rate": 3.765769468843284e-07,
|
| 2115 |
+
"loss": 0.2513,
|
| 2116 |
+
"step": 3010
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.6234838709677419,
|
| 2120 |
+
"grad_norm": 0.3886939892314166,
|
| 2121 |
+
"learning_rate": 3.7308732598236896e-07,
|
| 2122 |
+
"loss": 0.2589,
|
| 2123 |
+
"step": 3020
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 0.6255483870967742,
|
| 2127 |
+
"grad_norm": 0.41019034012665107,
|
| 2128 |
+
"learning_rate": 3.696043002870521e-07,
|
| 2129 |
+
"loss": 0.2602,
|
| 2130 |
+
"step": 3030
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 0.6276129032258064,
|
| 2134 |
+
"grad_norm": 0.41109492024152083,
|
| 2135 |
+
"learning_rate": 3.6612805079900954e-07,
|
| 2136 |
+
"loss": 0.2615,
|
| 2137 |
+
"step": 3040
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 0.6296774193548387,
|
| 2141 |
+
"grad_norm": 0.40387895619985564,
|
| 2142 |
+
"learning_rate": 3.6265875816673784e-07,
|
| 2143 |
+
"loss": 0.2623,
|
| 2144 |
+
"step": 3050
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.6317419354838709,
|
| 2148 |
+
"grad_norm": 0.41900812828527817,
|
| 2149 |
+
"learning_rate": 3.591966026772096e-07,
|
| 2150 |
+
"loss": 0.2588,
|
| 2151 |
+
"step": 3060
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.6338064516129033,
|
| 2155 |
+
"grad_norm": 0.40787156582167655,
|
| 2156 |
+
"learning_rate": 3.557417642465055e-07,
|
| 2157 |
+
"loss": 0.2639,
|
| 2158 |
+
"step": 3070
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 0.6358709677419355,
|
| 2162 |
+
"grad_norm": 0.3898972418936542,
|
| 2163 |
+
"learning_rate": 3.522944224104643e-07,
|
| 2164 |
+
"loss": 0.2649,
|
| 2165 |
+
"step": 3080
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 0.6379354838709678,
|
| 2169 |
+
"grad_norm": 0.4141718763979276,
|
| 2170 |
+
"learning_rate": 3.4885475631535253e-07,
|
| 2171 |
+
"loss": 0.2692,
|
| 2172 |
+
"step": 3090
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.64,
|
| 2176 |
+
"grad_norm": 0.41506013689499804,
|
| 2177 |
+
"learning_rate": 3.4542294470855626e-07,
|
| 2178 |
+
"loss": 0.2602,
|
| 2179 |
+
"step": 3100
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 0.6420645161290323,
|
| 2183 |
+
"grad_norm": 0.39476928102019726,
|
| 2184 |
+
"learning_rate": 3.4199916592929055e-07,
|
| 2185 |
+
"loss": 0.261,
|
| 2186 |
+
"step": 3110
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.6441290322580645,
|
| 2190 |
+
"grad_norm": 0.416429380761246,
|
| 2191 |
+
"learning_rate": 3.3858359789933313e-07,
|
| 2192 |
+
"loss": 0.2627,
|
| 2193 |
+
"step": 3120
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.6461935483870968,
|
| 2197 |
+
"grad_norm": 0.40310812066816015,
|
| 2198 |
+
"learning_rate": 3.351764181137771e-07,
|
| 2199 |
+
"loss": 0.2636,
|
| 2200 |
+
"step": 3130
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 0.648258064516129,
|
| 2204 |
+
"grad_norm": 0.3946905030128788,
|
| 2205 |
+
"learning_rate": 3.3177780363180923e-07,
|
| 2206 |
+
"loss": 0.2622,
|
| 2207 |
+
"step": 3140
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.6503225806451612,
|
| 2211 |
+
"grad_norm": 0.42397756082438576,
|
| 2212 |
+
"learning_rate": 3.2838793106750625e-07,
|
| 2213 |
+
"loss": 0.2603,
|
| 2214 |
+
"step": 3150
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.6523870967741936,
|
| 2218 |
+
"grad_norm": 0.4169594318618278,
|
| 2219 |
+
"learning_rate": 3.250069765806586e-07,
|
| 2220 |
+
"loss": 0.2613,
|
| 2221 |
+
"step": 3160
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.6544516129032258,
|
| 2225 |
+
"grad_norm": 0.41756086123454705,
|
| 2226 |
+
"learning_rate": 3.2163511586761566e-07,
|
| 2227 |
+
"loss": 0.2617,
|
| 2228 |
+
"step": 3170
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.6565161290322581,
|
| 2232 |
+
"grad_norm": 0.40493296338507545,
|
| 2233 |
+
"learning_rate": 3.1827252415215454e-07,
|
| 2234 |
+
"loss": 0.2656,
|
| 2235 |
+
"step": 3180
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.6585806451612903,
|
| 2239 |
+
"grad_norm": 0.4086571674620378,
|
| 2240 |
+
"learning_rate": 3.149193761763764e-07,
|
| 2241 |
+
"loss": 0.2611,
|
| 2242 |
+
"step": 3190
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.6606451612903226,
|
| 2246 |
+
"grad_norm": 0.4171128006614707,
|
| 2247 |
+
"learning_rate": 3.115758461916229e-07,
|
| 2248 |
+
"loss": 0.2597,
|
| 2249 |
+
"step": 3200
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.6627096774193548,
|
| 2253 |
+
"grad_norm": 0.3880167728055982,
|
| 2254 |
+
"learning_rate": 3.0824210794942333e-07,
|
| 2255 |
+
"loss": 0.263,
|
| 2256 |
+
"step": 3210
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.6647741935483871,
|
| 2260 |
+
"grad_norm": 0.40448031317959615,
|
| 2261 |
+
"learning_rate": 3.049183346924643e-07,
|
| 2262 |
+
"loss": 0.2632,
|
| 2263 |
+
"step": 3220
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.6668387096774193,
|
| 2267 |
+
"grad_norm": 0.41914072800271474,
|
| 2268 |
+
"learning_rate": 3.016046991455865e-07,
|
| 2269 |
+
"loss": 0.2579,
|
| 2270 |
+
"step": 3230
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.6689032258064516,
|
| 2274 |
+
"grad_norm": 0.3860018527353455,
|
| 2275 |
+
"learning_rate": 2.983013735068105e-07,
|
| 2276 |
+
"loss": 0.2647,
|
| 2277 |
+
"step": 3240
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.6709677419354839,
|
| 2281 |
+
"grad_norm": 0.40975610194833284,
|
| 2282 |
+
"learning_rate": 2.9500852943838603e-07,
|
| 2283 |
+
"loss": 0.2573,
|
| 2284 |
+
"step": 3250
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 0.6730322580645162,
|
| 2288 |
+
"grad_norm": 0.42646644618083696,
|
| 2289 |
+
"learning_rate": 2.9172633805787293e-07,
|
| 2290 |
+
"loss": 0.2653,
|
| 2291 |
+
"step": 3260
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 0.6750967741935484,
|
| 2295 |
+
"grad_norm": 0.4086698532968832,
|
| 2296 |
+
"learning_rate": 2.88454969929248e-07,
|
| 2297 |
+
"loss": 0.2574,
|
| 2298 |
+
"step": 3270
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.6771612903225807,
|
| 2302 |
+
"grad_norm": 0.4178375142437365,
|
| 2303 |
+
"learning_rate": 2.851945950540419e-07,
|
| 2304 |
+
"loss": 0.2623,
|
| 2305 |
+
"step": 3280
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.6792258064516129,
|
| 2309 |
+
"grad_norm": 0.4057288769351614,
|
| 2310 |
+
"learning_rate": 2.819453828625038e-07,
|
| 2311 |
+
"loss": 0.2651,
|
| 2312 |
+
"step": 3290
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.6812903225806451,
|
| 2316 |
+
"grad_norm": 0.4158815157531405,
|
| 2317 |
+
"learning_rate": 2.787075022047981e-07,
|
| 2318 |
+
"loss": 0.263,
|
| 2319 |
+
"step": 3300
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.6833548387096774,
|
| 2323 |
+
"grad_norm": 0.4049970316693273,
|
| 2324 |
+
"learning_rate": 2.754811213422288e-07,
|
| 2325 |
+
"loss": 0.2645,
|
| 2326 |
+
"step": 3310
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 0.6854193548387096,
|
| 2330 |
+
"grad_norm": 0.39671368245623034,
|
| 2331 |
+
"learning_rate": 2.722664079384954e-07,
|
| 2332 |
+
"loss": 0.2589,
|
| 2333 |
+
"step": 3320
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 0.6874838709677419,
|
| 2337 |
+
"grad_norm": 0.4039120828242038,
|
| 2338 |
+
"learning_rate": 2.6906352905098116e-07,
|
| 2339 |
+
"loss": 0.2614,
|
| 2340 |
+
"step": 3330
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 0.6895483870967742,
|
| 2344 |
+
"grad_norm": 0.4170702189116275,
|
| 2345 |
+
"learning_rate": 2.658726511220712e-07,
|
| 2346 |
+
"loss": 0.2657,
|
| 2347 |
+
"step": 3340
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.6916129032258065,
|
| 2351 |
+
"grad_norm": 0.4237911525616141,
|
| 2352 |
+
"learning_rate": 2.6269393997050205e-07,
|
| 2353 |
+
"loss": 0.2618,
|
| 2354 |
+
"step": 3350
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.6936774193548387,
|
| 2358 |
+
"grad_norm": 0.4111737496158447,
|
| 2359 |
+
"learning_rate": 2.595275607827457e-07,
|
| 2360 |
+
"loss": 0.2606,
|
| 2361 |
+
"step": 3360
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.695741935483871,
|
| 2365 |
+
"grad_norm": 0.38990938914606454,
|
| 2366 |
+
"learning_rate": 2.563736781044262e-07,
|
| 2367 |
+
"loss": 0.2577,
|
| 2368 |
+
"step": 3370
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.6978064516129032,
|
| 2372 |
+
"grad_norm": 0.407914899989363,
|
| 2373 |
+
"learning_rate": 2.5323245583176667e-07,
|
| 2374 |
+
"loss": 0.2609,
|
| 2375 |
+
"step": 3380
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 0.6998709677419355,
|
| 2379 |
+
"grad_norm": 0.39451690942708684,
|
| 2380 |
+
"learning_rate": 2.5010405720307355e-07,
|
| 2381 |
+
"loss": 0.2595,
|
| 2382 |
+
"step": 3390
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 0.7019354838709677,
|
| 2386 |
+
"grad_norm": 0.4092342022576978,
|
| 2387 |
+
"learning_rate": 2.469886447902541e-07,
|
| 2388 |
+
"loss": 0.2627,
|
| 2389 |
+
"step": 3400
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 0.704,
|
| 2393 |
+
"grad_norm": 0.39988371342369805,
|
| 2394 |
+
"learning_rate": 2.438863804903666e-07,
|
| 2395 |
+
"loss": 0.262,
|
| 2396 |
+
"step": 3410
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.7060645161290322,
|
| 2400 |
+
"grad_norm": 0.3964335866793338,
|
| 2401 |
+
"learning_rate": 2.4079742551720887e-07,
|
| 2402 |
+
"loss": 0.2592,
|
| 2403 |
+
"step": 3420
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.7081290322580646,
|
| 2407 |
+
"grad_norm": 0.4315689515435995,
|
| 2408 |
+
"learning_rate": 2.3772194039293896e-07,
|
| 2409 |
+
"loss": 0.2629,
|
| 2410 |
+
"step": 3430
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 0.7101935483870968,
|
| 2414 |
+
"grad_norm": 0.4070533032882783,
|
| 2415 |
+
"learning_rate": 2.3466008493973477e-07,
|
| 2416 |
+
"loss": 0.2634,
|
| 2417 |
+
"step": 3440
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.712258064516129,
|
| 2421 |
+
"grad_norm": 0.4139707970034922,
|
| 2422 |
+
"learning_rate": 2.3161201827148725e-07,
|
| 2423 |
+
"loss": 0.2668,
|
| 2424 |
+
"step": 3450
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.7143225806451613,
|
| 2428 |
+
"grad_norm": 0.4080257693965919,
|
| 2429 |
+
"learning_rate": 2.2857789878553309e-07,
|
| 2430 |
+
"loss": 0.2639,
|
| 2431 |
+
"step": 3460
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.7163870967741935,
|
| 2435 |
+
"grad_norm": 0.38944573511805763,
|
| 2436 |
+
"learning_rate": 2.2555788415442288e-07,
|
| 2437 |
+
"loss": 0.2591,
|
| 2438 |
+
"step": 3470
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.7184516129032258,
|
| 2442 |
+
"grad_norm": 0.40451831206378125,
|
| 2443 |
+
"learning_rate": 2.22552131317727e-07,
|
| 2444 |
+
"loss": 0.2608,
|
| 2445 |
+
"step": 3480
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.720516129032258,
|
| 2449 |
+
"grad_norm": 0.40945571442698814,
|
| 2450 |
+
"learning_rate": 2.1956079647388025e-07,
|
| 2451 |
+
"loss": 0.254,
|
| 2452 |
+
"step": 3490
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 0.7225806451612903,
|
| 2456 |
+
"grad_norm": 0.4120262445632641,
|
| 2457 |
+
"learning_rate": 2.165840350720655e-07,
|
| 2458 |
+
"loss": 0.2634,
|
| 2459 |
+
"step": 3500
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 0.7246451612903225,
|
| 2463 |
+
"grad_norm": 0.38102966990936765,
|
| 2464 |
+
"learning_rate": 2.1362200180413481e-07,
|
| 2465 |
+
"loss": 0.2612,
|
| 2466 |
+
"step": 3510
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 0.7267096774193549,
|
| 2470 |
+
"grad_norm": 0.40448293811396235,
|
| 2471 |
+
"learning_rate": 2.1067485059657032e-07,
|
| 2472 |
+
"loss": 0.2654,
|
| 2473 |
+
"step": 3520
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.7287741935483871,
|
| 2477 |
+
"grad_norm": 0.42379282553088654,
|
| 2478 |
+
"learning_rate": 2.0774273460248577e-07,
|
| 2479 |
+
"loss": 0.266,
|
| 2480 |
+
"step": 3530
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.7308387096774194,
|
| 2484 |
+
"grad_norm": 0.3958635594454363,
|
| 2485 |
+
"learning_rate": 2.0482580619366796e-07,
|
| 2486 |
+
"loss": 0.2599,
|
| 2487 |
+
"step": 3540
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.7329032258064516,
|
| 2491 |
+
"grad_norm": 0.3929275138734596,
|
| 2492 |
+
"learning_rate": 2.019242169526581e-07,
|
| 2493 |
+
"loss": 0.2614,
|
| 2494 |
+
"step": 3550
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.7349677419354839,
|
| 2498 |
+
"grad_norm": 0.4099908819814468,
|
| 2499 |
+
"learning_rate": 1.9903811766487426e-07,
|
| 2500 |
+
"loss": 0.2595,
|
| 2501 |
+
"step": 3560
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.7370322580645161,
|
| 2505 |
+
"grad_norm": 0.4136881756772272,
|
| 2506 |
+
"learning_rate": 1.9616765831077603e-07,
|
| 2507 |
+
"loss": 0.268,
|
| 2508 |
+
"step": 3570
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.7390967741935484,
|
| 2512 |
+
"grad_norm": 0.4103952583965791,
|
| 2513 |
+
"learning_rate": 1.9331298805807095e-07,
|
| 2514 |
+
"loss": 0.2667,
|
| 2515 |
+
"step": 3580
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.7411612903225806,
|
| 2519 |
+
"grad_norm": 0.4043875841263039,
|
| 2520 |
+
"learning_rate": 1.9047425525396161e-07,
|
| 2521 |
+
"loss": 0.2591,
|
| 2522 |
+
"step": 3590
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.743225806451613,
|
| 2526 |
+
"grad_norm": 0.40432519757481955,
|
| 2527 |
+
"learning_rate": 1.876516074174379e-07,
|
| 2528 |
+
"loss": 0.2583,
|
| 2529 |
+
"step": 3600
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.7452903225806452,
|
| 2533 |
+
"grad_norm": 0.3923980263069265,
|
| 2534 |
+
"learning_rate": 1.848451912316103e-07,
|
| 2535 |
+
"loss": 0.2622,
|
| 2536 |
+
"step": 3610
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.7473548387096774,
|
| 2540 |
+
"grad_norm": 0.41193174262995386,
|
| 2541 |
+
"learning_rate": 1.8205515253608688e-07,
|
| 2542 |
+
"loss": 0.2591,
|
| 2543 |
+
"step": 3620
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.7494193548387097,
|
| 2547 |
+
"grad_norm": 0.3997627378786646,
|
| 2548 |
+
"learning_rate": 1.792816363193952e-07,
|
| 2549 |
+
"loss": 0.2632,
|
| 2550 |
+
"step": 3630
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.7514838709677419,
|
| 2554 |
+
"grad_norm": 0.4203820131288817,
|
| 2555 |
+
"learning_rate": 1.7652478671144755e-07,
|
| 2556 |
+
"loss": 0.2603,
|
| 2557 |
+
"step": 3640
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.7535483870967742,
|
| 2561 |
+
"grad_norm": 0.41133420796621517,
|
| 2562 |
+
"learning_rate": 1.7378474697605128e-07,
|
| 2563 |
+
"loss": 0.2642,
|
| 2564 |
+
"step": 3650
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.7556129032258064,
|
| 2568 |
+
"grad_norm": 0.4037148313043383,
|
| 2569 |
+
"learning_rate": 1.7106165950346318e-07,
|
| 2570 |
+
"loss": 0.2656,
|
| 2571 |
+
"step": 3660
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.7576774193548387,
|
| 2575 |
+
"grad_norm": 0.4010420205215928,
|
| 2576 |
+
"learning_rate": 1.683556658029903e-07,
|
| 2577 |
+
"loss": 0.2601,
|
| 2578 |
+
"step": 3670
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.7597419354838709,
|
| 2582 |
+
"grad_norm": 0.4278976559359214,
|
| 2583 |
+
"learning_rate": 1.656669064956368e-07,
|
| 2584 |
+
"loss": 0.2586,
|
| 2585 |
+
"step": 3680
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 0.7618064516129033,
|
| 2589 |
+
"grad_norm": 0.41318596694086984,
|
| 2590 |
+
"learning_rate": 1.6299552130679578e-07,
|
| 2591 |
+
"loss": 0.2652,
|
| 2592 |
+
"step": 3690
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.7638709677419355,
|
| 2596 |
+
"grad_norm": 0.41756399746678385,
|
| 2597 |
+
"learning_rate": 1.6034164905898768e-07,
|
| 2598 |
+
"loss": 0.2634,
|
| 2599 |
+
"step": 3700
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 0.7659354838709678,
|
| 2603 |
+
"grad_norm": 0.45439314697770533,
|
| 2604 |
+
"learning_rate": 1.5770542766464773e-07,
|
| 2605 |
+
"loss": 0.2616,
|
| 2606 |
+
"step": 3710
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.768,
|
| 2610 |
+
"grad_norm": 0.3959443888976775,
|
| 2611 |
+
"learning_rate": 1.5508699411895738e-07,
|
| 2612 |
+
"loss": 0.2668,
|
| 2613 |
+
"step": 3720
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 0.7700645161290323,
|
| 2617 |
+
"grad_norm": 0.39850312610937977,
|
| 2618 |
+
"learning_rate": 1.524864844927266e-07,
|
| 2619 |
+
"loss": 0.2646,
|
| 2620 |
+
"step": 3730
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.7721290322580645,
|
| 2624 |
+
"grad_norm": 0.41376874719752166,
|
| 2625 |
+
"learning_rate": 1.4990403392532159e-07,
|
| 2626 |
+
"loss": 0.2577,
|
| 2627 |
+
"step": 3740
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.7741935483870968,
|
| 2631 |
+
"grad_norm": 0.4063544496998586,
|
| 2632 |
+
"learning_rate": 1.473397766176431e-07,
|
| 2633 |
+
"loss": 0.2649,
|
| 2634 |
+
"step": 3750
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.776258064516129,
|
| 2638 |
+
"grad_norm": 0.4353014336434687,
|
| 2639 |
+
"learning_rate": 1.4479384582515153e-07,
|
| 2640 |
+
"loss": 0.2643,
|
| 2641 |
+
"step": 3760
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.7783225806451612,
|
| 2645 |
+
"grad_norm": 0.4018175153121587,
|
| 2646 |
+
"learning_rate": 1.4226637385094247e-07,
|
| 2647 |
+
"loss": 0.2618,
|
| 2648 |
+
"step": 3770
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.7803870967741936,
|
| 2652 |
+
"grad_norm": 0.42157127333864497,
|
| 2653 |
+
"learning_rate": 1.3975749203887228e-07,
|
| 2654 |
+
"loss": 0.2651,
|
| 2655 |
+
"step": 3780
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 0.7824516129032258,
|
| 2659 |
+
"grad_norm": 0.43186635599469986,
|
| 2660 |
+
"learning_rate": 1.3726733076673085e-07,
|
| 2661 |
+
"loss": 0.2633,
|
| 2662 |
+
"step": 3790
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 0.7845161290322581,
|
| 2666 |
+
"grad_norm": 0.3964540005139217,
|
| 2667 |
+
"learning_rate": 1.3479601943946761e-07,
|
| 2668 |
+
"loss": 0.2566,
|
| 2669 |
+
"step": 3800
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 0.7865806451612903,
|
| 2673 |
+
"grad_norm": 0.44845859907363916,
|
| 2674 |
+
"learning_rate": 1.323436864824664e-07,
|
| 2675 |
+
"loss": 0.2635,
|
| 2676 |
+
"step": 3810
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 0.7886451612903226,
|
| 2680 |
+
"grad_norm": 0.39439621730318203,
|
| 2681 |
+
"learning_rate": 1.299104593348721e-07,
|
| 2682 |
+
"loss": 0.267,
|
| 2683 |
+
"step": 3820
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.7907096774193548,
|
| 2687 |
+
"grad_norm": 0.4258443910811124,
|
| 2688 |
+
"learning_rate": 1.2749646444296703e-07,
|
| 2689 |
+
"loss": 0.2629,
|
| 2690 |
+
"step": 3830
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.7927741935483871,
|
| 2694 |
+
"grad_norm": 0.4061749201396272,
|
| 2695 |
+
"learning_rate": 1.2510182725360086e-07,
|
| 2696 |
+
"loss": 0.256,
|
| 2697 |
+
"step": 3840
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.7948387096774193,
|
| 2701 |
+
"grad_norm": 0.4053363945100216,
|
| 2702 |
+
"learning_rate": 1.2272667220767158e-07,
|
| 2703 |
+
"loss": 0.267,
|
| 2704 |
+
"step": 3850
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 0.7969032258064516,
|
| 2708 |
+
"grad_norm": 0.39724264268663745,
|
| 2709 |
+
"learning_rate": 1.2037112273365818e-07,
|
| 2710 |
+
"loss": 0.2601,
|
| 2711 |
+
"step": 3860
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.7989677419354839,
|
| 2715 |
+
"grad_norm": 0.40962689626618026,
|
| 2716 |
+
"learning_rate": 1.1803530124120714e-07,
|
| 2717 |
+
"loss": 0.2614,
|
| 2718 |
+
"step": 3870
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 0.8010322580645162,
|
| 2722 |
+
"grad_norm": 0.39171805970284057,
|
| 2723 |
+
"learning_rate": 1.157193291147705e-07,
|
| 2724 |
+
"loss": 0.2604,
|
| 2725 |
+
"step": 3880
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 0.8030967741935484,
|
| 2729 |
+
"grad_norm": 0.38720737774209957,
|
| 2730 |
+
"learning_rate": 1.1342332670729882e-07,
|
| 2731 |
+
"loss": 0.258,
|
| 2732 |
+
"step": 3890
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.8051612903225807,
|
| 2736 |
+
"grad_norm": 0.398012624254619,
|
| 2737 |
+
"learning_rate": 1.1114741333398592e-07,
|
| 2738 |
+
"loss": 0.2621,
|
| 2739 |
+
"step": 3900
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.8072258064516129,
|
| 2743 |
+
"grad_norm": 0.40289439621605116,
|
| 2744 |
+
"learning_rate": 1.0889170726606933e-07,
|
| 2745 |
+
"loss": 0.2586,
|
| 2746 |
+
"step": 3910
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.8092903225806451,
|
| 2750 |
+
"grad_norm": 0.4232682123948614,
|
| 2751 |
+
"learning_rate": 1.066563257246838e-07,
|
| 2752 |
+
"loss": 0.2569,
|
| 2753 |
+
"step": 3920
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.8113548387096774,
|
| 2757 |
+
"grad_norm": 0.4028783048716607,
|
| 2758 |
+
"learning_rate": 1.0444138487476944e-07,
|
| 2759 |
+
"loss": 0.2609,
|
| 2760 |
+
"step": 3930
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.8134193548387096,
|
| 2764 |
+
"grad_norm": 0.4002686348680681,
|
| 2765 |
+
"learning_rate": 1.0224699981903517e-07,
|
| 2766 |
+
"loss": 0.2654,
|
| 2767 |
+
"step": 3940
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 0.8154838709677419,
|
| 2771 |
+
"grad_norm": 0.39543469139493387,
|
| 2772 |
+
"learning_rate": 1.0007328459197778e-07,
|
| 2773 |
+
"loss": 0.2628,
|
| 2774 |
+
"step": 3950
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.8175483870967742,
|
| 2778 |
+
"grad_norm": 0.4157630096664263,
|
| 2779 |
+
"learning_rate": 9.792035215395556e-08,
|
| 2780 |
+
"loss": 0.263,
|
| 2781 |
+
"step": 3960
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 0.8196129032258065,
|
| 2785 |
+
"grad_norm": 0.41685800951642293,
|
| 2786 |
+
"learning_rate": 9.578831438531776e-08,
|
| 2787 |
+
"loss": 0.2669,
|
| 2788 |
+
"step": 3970
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 0.8216774193548387,
|
| 2792 |
+
"grad_norm": 0.4061936200965336,
|
| 2793 |
+
"learning_rate": 9.3677282080591e-08,
|
| 2794 |
+
"loss": 0.2622,
|
| 2795 |
+
"step": 3980
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 0.823741935483871,
|
| 2799 |
+
"grad_norm": 0.40925273863210293,
|
| 2800 |
+
"learning_rate": 9.158736494272179e-08,
|
| 2801 |
+
"loss": 0.2612,
|
| 2802 |
+
"step": 3990
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.8258064516129032,
|
| 2806 |
+
"grad_norm": 0.40618534177947946,
|
| 2807 |
+
"learning_rate": 8.951867157737558e-08,
|
| 2808 |
+
"loss": 0.2682,
|
| 2809 |
+
"step": 4000
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 0.8278709677419355,
|
| 2813 |
+
"grad_norm": 0.3979706890222173,
|
| 2814 |
+
"learning_rate": 8.747130948729226e-08,
|
| 2815 |
+
"loss": 0.2596,
|
| 2816 |
+
"step": 4010
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.8299354838709677,
|
| 2820 |
+
"grad_norm": 0.4071872432393417,
|
| 2821 |
+
"learning_rate": 8.544538506670074e-08,
|
| 2822 |
+
"loss": 0.2587,
|
| 2823 |
+
"step": 4020
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 0.832,
|
| 2827 |
+
"grad_norm": 0.40836170061759924,
|
| 2828 |
+
"learning_rate": 8.344100359578904e-08,
|
| 2829 |
+
"loss": 0.2588,
|
| 2830 |
+
"step": 4030
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.8340645161290322,
|
| 2834 |
+
"grad_norm": 0.3895000386110786,
|
| 2835 |
+
"learning_rate": 8.145826923523358e-08,
|
| 2836 |
+
"loss": 0.2547,
|
| 2837 |
+
"step": 4040
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 0.8361290322580646,
|
| 2841 |
+
"grad_norm": 0.40154696748699115,
|
| 2842 |
+
"learning_rate": 7.949728502078668e-08,
|
| 2843 |
+
"loss": 0.2584,
|
| 2844 |
+
"step": 4050
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.8381935483870968,
|
| 2848 |
+
"grad_norm": 0.4376818236899298,
|
| 2849 |
+
"learning_rate": 7.755815285792172e-08,
|
| 2850 |
+
"loss": 0.2688,
|
| 2851 |
+
"step": 4060
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 0.840258064516129,
|
| 2855 |
+
"grad_norm": 0.4046160460755729,
|
| 2856 |
+
"learning_rate": 7.564097351653742e-08,
|
| 2857 |
+
"loss": 0.2705,
|
| 2858 |
+
"step": 4070
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.8423225806451613,
|
| 2862 |
+
"grad_norm": 0.39547101151078623,
|
| 2863 |
+
"learning_rate": 7.374584662572142e-08,
|
| 2864 |
+
"loss": 0.2659,
|
| 2865 |
+
"step": 4080
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 0.8443870967741935,
|
| 2869 |
+
"grad_norm": 0.37332809148900736,
|
| 2870 |
+
"learning_rate": 7.187287066857289e-08,
|
| 2871 |
+
"loss": 0.2574,
|
| 2872 |
+
"step": 4090
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 0.8464516129032258,
|
| 2876 |
+
"grad_norm": 0.46033640896289585,
|
| 2877 |
+
"learning_rate": 7.002214297708481e-08,
|
| 2878 |
+
"loss": 0.2569,
|
| 2879 |
+
"step": 4100
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 0.848516129032258,
|
| 2883 |
+
"grad_norm": 0.39191927118713255,
|
| 2884 |
+
"learning_rate": 6.819375972708536e-08,
|
| 2885 |
+
"loss": 0.258,
|
| 2886 |
+
"step": 4110
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 0.8505806451612903,
|
| 2890 |
+
"grad_norm": 0.41428077941004704,
|
| 2891 |
+
"learning_rate": 6.6387815933241e-08,
|
| 2892 |
+
"loss": 0.2618,
|
| 2893 |
+
"step": 4120
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 0.8526451612903225,
|
| 2897 |
+
"grad_norm": 0.4107387220224137,
|
| 2898 |
+
"learning_rate": 6.460440544411777e-08,
|
| 2899 |
+
"loss": 0.2622,
|
| 2900 |
+
"step": 4130
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 0.8547096774193549,
|
| 2904 |
+
"grad_norm": 0.39642375544117775,
|
| 2905 |
+
"learning_rate": 6.284362093730545e-08,
|
| 2906 |
+
"loss": 0.2585,
|
| 2907 |
+
"step": 4140
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 0.8567741935483871,
|
| 2911 |
+
"grad_norm": 0.37930337406912923,
|
| 2912 |
+
"learning_rate": 6.110555391460026e-08,
|
| 2913 |
+
"loss": 0.2617,
|
| 2914 |
+
"step": 4150
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 0.8588387096774194,
|
| 2918 |
+
"grad_norm": 0.377232478401539,
|
| 2919 |
+
"learning_rate": 5.9390294697251045e-08,
|
| 2920 |
+
"loss": 0.2583,
|
| 2921 |
+
"step": 4160
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 0.8609032258064516,
|
| 2925 |
+
"grad_norm": 0.3888835569603603,
|
| 2926 |
+
"learning_rate": 5.7697932421264415e-08,
|
| 2927 |
+
"loss": 0.2529,
|
| 2928 |
+
"step": 4170
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 0.8629677419354839,
|
| 2932 |
+
"grad_norm": 0.3934785671025381,
|
| 2933 |
+
"learning_rate": 5.602855503277376e-08,
|
| 2934 |
+
"loss": 0.2593,
|
| 2935 |
+
"step": 4180
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 0.8650322580645161,
|
| 2939 |
+
"grad_norm": 0.38987380628908475,
|
| 2940 |
+
"learning_rate": 5.438224928346791e-08,
|
| 2941 |
+
"loss": 0.2611,
|
| 2942 |
+
"step": 4190
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.8670967741935484,
|
| 2946 |
+
"grad_norm": 0.40191458902063176,
|
| 2947 |
+
"learning_rate": 5.275910072608408e-08,
|
| 2948 |
+
"loss": 0.2648,
|
| 2949 |
+
"step": 4200
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.8691612903225806,
|
| 2953 |
+
"grad_norm": 0.40555133430607165,
|
| 2954 |
+
"learning_rate": 5.115919370996097e-08,
|
| 2955 |
+
"loss": 0.2646,
|
| 2956 |
+
"step": 4210
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.8712258064516128,
|
| 2960 |
+
"grad_norm": 0.39189346564182975,
|
| 2961 |
+
"learning_rate": 4.9582611376655924e-08,
|
| 2962 |
+
"loss": 0.2697,
|
| 2963 |
+
"step": 4220
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 0.8732903225806452,
|
| 2967 |
+
"grad_norm": 0.3977403739928427,
|
| 2968 |
+
"learning_rate": 4.8029435655624785e-08,
|
| 2969 |
+
"loss": 0.256,
|
| 2970 |
+
"step": 4230
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.8753548387096775,
|
| 2974 |
+
"grad_norm": 0.4107418168891212,
|
| 2975 |
+
"learning_rate": 4.6499747259963254e-08,
|
| 2976 |
+
"loss": 0.2618,
|
| 2977 |
+
"step": 4240
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 0.8774193548387097,
|
| 2981 |
+
"grad_norm": 0.4152156746116232,
|
| 2982 |
+
"learning_rate": 4.499362568221327e-08,
|
| 2983 |
+
"loss": 0.2562,
|
| 2984 |
+
"step": 4250
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 0.8794838709677419,
|
| 2988 |
+
"grad_norm": 0.41269224554676137,
|
| 2989 |
+
"learning_rate": 4.351114919023197e-08,
|
| 2990 |
+
"loss": 0.2638,
|
| 2991 |
+
"step": 4260
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 0.8815483870967742,
|
| 2995 |
+
"grad_norm": 0.39019715520848103,
|
| 2996 |
+
"learning_rate": 4.205239482312445e-08,
|
| 2997 |
+
"loss": 0.2608,
|
| 2998 |
+
"step": 4270
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.8836129032258064,
|
| 3002 |
+
"grad_norm": 0.39460366251134543,
|
| 3003 |
+
"learning_rate": 4.0617438387239746e-08,
|
| 3004 |
+
"loss": 0.2644,
|
| 3005 |
+
"step": 4280
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 0.8856774193548387,
|
| 3009 |
+
"grad_norm": 0.38841049201774,
|
| 3010 |
+
"learning_rate": 3.9206354452232135e-08,
|
| 3011 |
+
"loss": 0.2586,
|
| 3012 |
+
"step": 4290
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.8877419354838709,
|
| 3016 |
+
"grad_norm": 0.3995452257322332,
|
| 3017 |
+
"learning_rate": 3.7819216347185715e-08,
|
| 3018 |
+
"loss": 0.2632,
|
| 3019 |
+
"step": 4300
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 0.8898064516129032,
|
| 3023 |
+
"grad_norm": 0.3921492973317569,
|
| 3024 |
+
"learning_rate": 3.6456096156803616e-08,
|
| 3025 |
+
"loss": 0.2591,
|
| 3026 |
+
"step": 4310
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.8918709677419355,
|
| 3030 |
+
"grad_norm": 0.37963166184003594,
|
| 3031 |
+
"learning_rate": 3.5117064717662406e-08,
|
| 3032 |
+
"loss": 0.2574,
|
| 3033 |
+
"step": 4320
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 0.8939354838709678,
|
| 3037 |
+
"grad_norm": 0.4129813226681988,
|
| 3038 |
+
"learning_rate": 3.380219161453063e-08,
|
| 3039 |
+
"loss": 0.2662,
|
| 3040 |
+
"step": 4330
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.896,
|
| 3044 |
+
"grad_norm": 0.40974368918108645,
|
| 3045 |
+
"learning_rate": 3.251154517675264e-08,
|
| 3046 |
+
"loss": 0.2604,
|
| 3047 |
+
"step": 4340
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 0.8980645161290323,
|
| 3051 |
+
"grad_norm": 0.4245296466579207,
|
| 3052 |
+
"learning_rate": 3.124519247469814e-08,
|
| 3053 |
+
"loss": 0.2607,
|
| 3054 |
+
"step": 4350
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 0.9001290322580645,
|
| 3058 |
+
"grad_norm": 0.42346220490081854,
|
| 3059 |
+
"learning_rate": 3.0003199316276486e-08,
|
| 3060 |
+
"loss": 0.2568,
|
| 3061 |
+
"step": 4360
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 0.9021935483870968,
|
| 3065 |
+
"grad_norm": 0.384390730131664,
|
| 3066 |
+
"learning_rate": 2.8785630243517156e-08,
|
| 3067 |
+
"loss": 0.2554,
|
| 3068 |
+
"step": 4370
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.904258064516129,
|
| 3072 |
+
"grad_norm": 0.4161617174420004,
|
| 3073 |
+
"learning_rate": 2.759254852921522e-08,
|
| 3074 |
+
"loss": 0.2611,
|
| 3075 |
+
"step": 4380
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 0.9063225806451612,
|
| 3079 |
+
"grad_norm": 0.4096140142802804,
|
| 3080 |
+
"learning_rate": 2.6424016173643816e-08,
|
| 3081 |
+
"loss": 0.2613,
|
| 3082 |
+
"step": 4390
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.9083870967741936,
|
| 3086 |
+
"grad_norm": 0.4036134224250512,
|
| 3087 |
+
"learning_rate": 2.5280093901331957e-08,
|
| 3088 |
+
"loss": 0.2564,
|
| 3089 |
+
"step": 4400
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.9104516129032258,
|
| 3093 |
+
"grad_norm": 0.3951213596923248,
|
| 3094 |
+
"learning_rate": 2.4160841157908894e-08,
|
| 3095 |
+
"loss": 0.2626,
|
| 3096 |
+
"step": 4410
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 0.9125161290322581,
|
| 3100 |
+
"grad_norm": 0.4255511159427451,
|
| 3101 |
+
"learning_rate": 2.3066316107014984e-08,
|
| 3102 |
+
"loss": 0.267,
|
| 3103 |
+
"step": 4420
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 0.9145806451612903,
|
| 3107 |
+
"grad_norm": 0.39941397499664477,
|
| 3108 |
+
"learning_rate": 2.199657562727919e-08,
|
| 3109 |
+
"loss": 0.2644,
|
| 3110 |
+
"step": 4430
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 0.9166451612903226,
|
| 3114 |
+
"grad_norm": 0.41202712449315076,
|
| 3115 |
+
"learning_rate": 2.0951675309363038e-08,
|
| 3116 |
+
"loss": 0.2662,
|
| 3117 |
+
"step": 4440
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 0.9187096774193548,
|
| 3121 |
+
"grad_norm": 0.4141354253620865,
|
| 3122 |
+
"learning_rate": 1.9931669453072064e-08,
|
| 3123 |
+
"loss": 0.2682,
|
| 3124 |
+
"step": 4450
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 0.9207741935483871,
|
| 3128 |
+
"grad_norm": 0.407975000927539,
|
| 3129 |
+
"learning_rate": 1.893661106453387e-08,
|
| 3130 |
+
"loss": 0.2657,
|
| 3131 |
+
"step": 4460
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 0.9228387096774193,
|
| 3135 |
+
"grad_norm": 0.4080505082932939,
|
| 3136 |
+
"learning_rate": 1.7966551853443813e-08,
|
| 3137 |
+
"loss": 0.2591,
|
| 3138 |
+
"step": 4470
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 0.9249032258064516,
|
| 3142 |
+
"grad_norm": 0.4177794728607882,
|
| 3143 |
+
"learning_rate": 1.7021542230377495e-08,
|
| 3144 |
+
"loss": 0.2602,
|
| 3145 |
+
"step": 4480
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 0.9269677419354839,
|
| 3149 |
+
"grad_norm": 0.4337602592447264,
|
| 3150 |
+
"learning_rate": 1.610163130417119e-08,
|
| 3151 |
+
"loss": 0.2629,
|
| 3152 |
+
"step": 4490
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 0.9290322580645162,
|
| 3156 |
+
"grad_norm": 0.40025965085906845,
|
| 3157 |
+
"learning_rate": 1.520686687937006e-08,
|
| 3158 |
+
"loss": 0.2601,
|
| 3159 |
+
"step": 4500
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 0.9310967741935484,
|
| 3163 |
+
"grad_norm": 0.3954536343382873,
|
| 3164 |
+
"learning_rate": 1.4337295453743848e-08,
|
| 3165 |
+
"loss": 0.2625,
|
| 3166 |
+
"step": 4510
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 0.9331612903225807,
|
| 3170 |
+
"grad_norm": 0.4030179174929269,
|
| 3171 |
+
"learning_rate": 1.3492962215870208e-08,
|
| 3172 |
+
"loss": 0.2625,
|
| 3173 |
+
"step": 4520
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 0.9352258064516129,
|
| 3177 |
+
"grad_norm": 0.40819467221659395,
|
| 3178 |
+
"learning_rate": 1.2673911042786812e-08,
|
| 3179 |
+
"loss": 0.2623,
|
| 3180 |
+
"step": 4530
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 0.9372903225806452,
|
| 3184 |
+
"grad_norm": 0.3975548605776835,
|
| 3185 |
+
"learning_rate": 1.188018449771111e-08,
|
| 3186 |
+
"loss": 0.2587,
|
| 3187 |
+
"step": 4540
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 0.9393548387096774,
|
| 3191 |
+
"grad_norm": 0.38620015604289587,
|
| 3192 |
+
"learning_rate": 1.1111823827828438e-08,
|
| 3193 |
+
"loss": 0.2629,
|
| 3194 |
+
"step": 4550
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 0.9414193548387096,
|
| 3198 |
+
"grad_norm": 0.3915952951322532,
|
| 3199 |
+
"learning_rate": 1.0368868962148446e-08,
|
| 3200 |
+
"loss": 0.2602,
|
| 3201 |
+
"step": 4560
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 0.9434838709677419,
|
| 3205 |
+
"grad_norm": 0.42822828805165253,
|
| 3206 |
+
"learning_rate": 9.651358509430385e-09,
|
| 3207 |
+
"loss": 0.2668,
|
| 3208 |
+
"step": 4570
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 0.9455483870967742,
|
| 3212 |
+
"grad_norm": 0.4110557870118209,
|
| 3213 |
+
"learning_rate": 8.959329756176359e-09,
|
| 3214 |
+
"loss": 0.2614,
|
| 3215 |
+
"step": 4580
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 0.9476129032258065,
|
| 3219 |
+
"grad_norm": 0.4051400508982645,
|
| 3220 |
+
"learning_rate": 8.292818664694223e-09,
|
| 3221 |
+
"loss": 0.2579,
|
| 3222 |
+
"step": 4590
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 0.9496774193548387,
|
| 3226 |
+
"grad_norm": 0.41096745151064146,
|
| 3227 |
+
"learning_rate": 7.651859871228072e-09,
|
| 3228 |
+
"loss": 0.2638,
|
| 3229 |
+
"step": 4600
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 0.951741935483871,
|
| 3233 |
+
"grad_norm": 0.40156811491530275,
|
| 3234 |
+
"learning_rate": 7.0364866841589045e-09,
|
| 3235 |
+
"loss": 0.2631,
|
| 3236 |
+
"step": 4610
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 0.9538064516129032,
|
| 3240 |
+
"grad_norm": 0.4336183195339135,
|
| 3241 |
+
"learning_rate": 6.446731082273449e-09,
|
| 3242 |
+
"loss": 0.2626,
|
| 3243 |
+
"step": 4620
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 0.9558709677419355,
|
| 3247 |
+
"grad_norm": 0.41525878972145586,
|
| 3248 |
+
"learning_rate": 5.8826237131022196e-09,
|
| 3249 |
+
"loss": 0.2637,
|
| 3250 |
+
"step": 4630
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 0.9579354838709677,
|
| 3254 |
+
"grad_norm": 0.39563865795758707,
|
| 3255 |
+
"learning_rate": 5.344193891327286e-09,
|
| 3256 |
+
"loss": 0.2599,
|
| 3257 |
+
"step": 4640
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 0.96,
|
| 3261 |
+
"grad_norm": 0.3902289846827882,
|
| 3262 |
+
"learning_rate": 4.831469597258331e-09,
|
| 3263 |
+
"loss": 0.2588,
|
| 3264 |
+
"step": 4650
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 0.9620645161290322,
|
| 3268 |
+
"grad_norm": 0.4265778114435462,
|
| 3269 |
+
"learning_rate": 4.344477475379027e-09,
|
| 3270 |
+
"loss": 0.263,
|
| 3271 |
+
"step": 4660
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 0.9641290322580646,
|
| 3275 |
+
"grad_norm": 0.3952736993136822,
|
| 3276 |
+
"learning_rate": 3.883242832962319e-09,
|
| 3277 |
+
"loss": 0.266,
|
| 3278 |
+
"step": 4670
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 0.9661935483870968,
|
| 3282 |
+
"grad_norm": 0.4204128593760075,
|
| 3283 |
+
"learning_rate": 3.4477896387552497e-09,
|
| 3284 |
+
"loss": 0.2632,
|
| 3285 |
+
"step": 4680
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 0.9682580645161291,
|
| 3289 |
+
"grad_norm": 0.39870257721591323,
|
| 3290 |
+
"learning_rate": 3.0381405217333457e-09,
|
| 3291 |
+
"loss": 0.2598,
|
| 3292 |
+
"step": 4690
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 0.9703225806451613,
|
| 3296 |
+
"grad_norm": 0.40504192880628953,
|
| 3297 |
+
"learning_rate": 2.6543167699247248e-09,
|
| 3298 |
+
"loss": 0.2663,
|
| 3299 |
+
"step": 4700
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 0.9723870967741935,
|
| 3303 |
+
"grad_norm": 0.4276504248620953,
|
| 3304 |
+
"learning_rate": 2.2963383293039264e-09,
|
| 3305 |
+
"loss": 0.2681,
|
| 3306 |
+
"step": 4710
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 0.9744516129032258,
|
| 3310 |
+
"grad_norm": 0.4194237088370349,
|
| 3311 |
+
"learning_rate": 1.964223802755238e-09,
|
| 3312 |
+
"loss": 0.2642,
|
| 3313 |
+
"step": 4720
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 0.976516129032258,
|
| 3317 |
+
"grad_norm": 0.41830150874097594,
|
| 3318 |
+
"learning_rate": 1.6579904491059726e-09,
|
| 3319 |
+
"loss": 0.257,
|
| 3320 |
+
"step": 4730
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 0.9785806451612903,
|
| 3324 |
+
"grad_norm": 0.381130263031969,
|
| 3325 |
+
"learning_rate": 1.3776541822297926e-09,
|
| 3326 |
+
"loss": 0.2652,
|
| 3327 |
+
"step": 4740
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 0.9806451612903225,
|
| 3331 |
+
"grad_norm": 0.4000580182685434,
|
| 3332 |
+
"learning_rate": 1.1232295702193751e-09,
|
| 3333 |
+
"loss": 0.2635,
|
| 3334 |
+
"step": 4750
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 0.9827096774193549,
|
| 3338 |
+
"grad_norm": 0.41681278877265543,
|
| 3339 |
+
"learning_rate": 8.947298346296816e-10,
|
| 3340 |
+
"loss": 0.2683,
|
| 3341 |
+
"step": 4760
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 0.9847741935483871,
|
| 3345 |
+
"grad_norm": 0.40430126615446177,
|
| 3346 |
+
"learning_rate": 6.921668497907873e-10,
|
| 3347 |
+
"loss": 0.262,
|
| 3348 |
+
"step": 4770
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 0.9868387096774194,
|
| 3352 |
+
"grad_norm": 0.41793940361219006,
|
| 3353 |
+
"learning_rate": 5.155511421906511e-10,
|
| 3354 |
+
"loss": 0.2619,
|
| 3355 |
+
"step": 4780
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 0.9889032258064516,
|
| 3359 |
+
"grad_norm": 0.4020044075973662,
|
| 3360 |
+
"learning_rate": 3.64891889928276e-10,
|
| 3361 |
+
"loss": 0.262,
|
| 3362 |
+
"step": 4790
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 0.9909677419354839,
|
| 3366 |
+
"grad_norm": 0.39705009551285064,
|
| 3367 |
+
"learning_rate": 2.4019692223675727e-10,
|
| 3368 |
+
"loss": 0.2597,
|
| 3369 |
+
"step": 4800
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 0.9930322580645161,
|
| 3373 |
+
"grad_norm": 0.41207414069949416,
|
| 3374 |
+
"learning_rate": 1.4147271907621883e-10,
|
| 3375 |
+
"loss": 0.2605,
|
| 3376 |
+
"step": 4810
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 0.9950967741935484,
|
| 3380 |
+
"grad_norm": 0.4262315482989849,
|
| 3381 |
+
"learning_rate": 6.87244107974716e-11,
|
| 3382 |
+
"loss": 0.2601,
|
| 3383 |
+
"step": 4820
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 0.9971612903225806,
|
| 3387 |
+
"grad_norm": 0.4195255805487681,
|
| 3388 |
+
"learning_rate": 2.1955777874838045e-11,
|
| 3389 |
+
"loss": 0.2636,
|
| 3390 |
+
"step": 4830
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 0.9992258064516129,
|
| 3394 |
+
"grad_norm": 0.40670072777909505,
|
| 3395 |
+
"learning_rate": 1.1692507103089334e-12,
|
| 3396 |
+
"loss": 0.2565,
|
| 3397 |
+
"step": 4840
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 0.9998451612903225,
|
| 3401 |
+
"step": 4843,
|
| 3402 |
+
"total_flos": 1061427576569856.0,
|
| 3403 |
+
"train_loss": 0.09964955267816591,
|
| 3404 |
+
"train_runtime": 121089.0768,
|
| 3405 |
+
"train_samples_per_second": 10.24,
|
| 3406 |
+
"train_steps_per_second": 0.04
|
| 3407 |
+
}
|
| 3408 |
+
],
|
| 3409 |
+
"logging_steps": 10,
|
| 3410 |
+
"max_steps": 4843,
|
| 3411 |
+
"num_input_tokens_seen": 0,
|
| 3412 |
+
"num_train_epochs": 1,
|
| 3413 |
+
"save_steps": 1000,
|
| 3414 |
+
"stateful_callbacks": {
|
| 3415 |
+
"TrainerControl": {
|
| 3416 |
+
"args": {
|
| 3417 |
+
"should_epoch_stop": false,
|
| 3418 |
+
"should_evaluate": false,
|
| 3419 |
+
"should_log": false,
|
| 3420 |
+
"should_save": true,
|
| 3421 |
+
"should_training_stop": true
|
| 3422 |
+
},
|
| 3423 |
+
"attributes": {}
|
| 3424 |
+
}
|
| 3425 |
+
},
|
| 3426 |
+
"total_flos": 1061427576569856.0,
|
| 3427 |
+
"train_batch_size": 8,
|
| 3428 |
+
"trial_name": null,
|
| 3429 |
+
"trial_params": null
|
| 3430 |
+
}
|