Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +3465 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3465 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 489,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.006134969325153374,
|
| 13 |
+
"grad_norm": 4.404277324676514,
|
| 14 |
+
"learning_rate": 4.0816326530612243e-07,
|
| 15 |
+
"loss": 1.0324,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.012269938650306749,
|
| 20 |
+
"grad_norm": 4.627435684204102,
|
| 21 |
+
"learning_rate": 8.163265306122449e-07,
|
| 22 |
+
"loss": 1.0989,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.018404907975460124,
|
| 27 |
+
"grad_norm": 4.390231132507324,
|
| 28 |
+
"learning_rate": 1.2244897959183673e-06,
|
| 29 |
+
"loss": 0.9683,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.024539877300613498,
|
| 34 |
+
"grad_norm": 5.326767921447754,
|
| 35 |
+
"learning_rate": 1.6326530612244897e-06,
|
| 36 |
+
"loss": 1.0904,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03067484662576687,
|
| 41 |
+
"grad_norm": 6.357185363769531,
|
| 42 |
+
"learning_rate": 2.0408163265306125e-06,
|
| 43 |
+
"loss": 1.0923,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03680981595092025,
|
| 48 |
+
"grad_norm": 4.6469502449035645,
|
| 49 |
+
"learning_rate": 2.4489795918367347e-06,
|
| 50 |
+
"loss": 1.0713,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04294478527607362,
|
| 55 |
+
"grad_norm": 6.18074893951416,
|
| 56 |
+
"learning_rate": 2.8571428571428573e-06,
|
| 57 |
+
"loss": 1.0621,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.049079754601226995,
|
| 62 |
+
"grad_norm": 4.181659698486328,
|
| 63 |
+
"learning_rate": 3.2653061224489794e-06,
|
| 64 |
+
"loss": 0.9701,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.05521472392638037,
|
| 69 |
+
"grad_norm": 5.808837890625,
|
| 70 |
+
"learning_rate": 3.6734693877551024e-06,
|
| 71 |
+
"loss": 0.9578,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.06134969325153374,
|
| 76 |
+
"grad_norm": 4.635038375854492,
|
| 77 |
+
"learning_rate": 4.081632653061225e-06,
|
| 78 |
+
"loss": 1.0176,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.06748466257668712,
|
| 83 |
+
"grad_norm": 4.72430419921875,
|
| 84 |
+
"learning_rate": 4.489795918367348e-06,
|
| 85 |
+
"loss": 0.7826,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.0736196319018405,
|
| 90 |
+
"grad_norm": 3.9675962924957275,
|
| 91 |
+
"learning_rate": 4.897959183673469e-06,
|
| 92 |
+
"loss": 0.9632,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07975460122699386,
|
| 97 |
+
"grad_norm": 3.53987717628479,
|
| 98 |
+
"learning_rate": 5.306122448979593e-06,
|
| 99 |
+
"loss": 0.8969,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.08588957055214724,
|
| 104 |
+
"grad_norm": 4.441963195800781,
|
| 105 |
+
"learning_rate": 5.7142857142857145e-06,
|
| 106 |
+
"loss": 0.8889,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.09202453987730061,
|
| 111 |
+
"grad_norm": 2.9366307258605957,
|
| 112 |
+
"learning_rate": 6.122448979591837e-06,
|
| 113 |
+
"loss": 0.897,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09815950920245399,
|
| 118 |
+
"grad_norm": 2.6488900184631348,
|
| 119 |
+
"learning_rate": 6.530612244897959e-06,
|
| 120 |
+
"loss": 0.9497,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.10429447852760736,
|
| 125 |
+
"grad_norm": 3.0431630611419678,
|
| 126 |
+
"learning_rate": 6.938775510204082e-06,
|
| 127 |
+
"loss": 0.8834,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.11042944785276074,
|
| 132 |
+
"grad_norm": 3.631333351135254,
|
| 133 |
+
"learning_rate": 7.346938775510205e-06,
|
| 134 |
+
"loss": 0.8728,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1165644171779141,
|
| 139 |
+
"grad_norm": 2.983065605163574,
|
| 140 |
+
"learning_rate": 7.755102040816327e-06,
|
| 141 |
+
"loss": 0.9526,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.12269938650306748,
|
| 146 |
+
"grad_norm": 2.8151237964630127,
|
| 147 |
+
"learning_rate": 8.16326530612245e-06,
|
| 148 |
+
"loss": 0.7929,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.12883435582822086,
|
| 153 |
+
"grad_norm": 2.7403059005737305,
|
| 154 |
+
"learning_rate": 8.571428571428571e-06,
|
| 155 |
+
"loss": 0.9033,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.13496932515337423,
|
| 160 |
+
"grad_norm": 2.607174873352051,
|
| 161 |
+
"learning_rate": 8.979591836734695e-06,
|
| 162 |
+
"loss": 0.8278,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1411042944785276,
|
| 167 |
+
"grad_norm": 3.1742658615112305,
|
| 168 |
+
"learning_rate": 9.387755102040818e-06,
|
| 169 |
+
"loss": 0.7851,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.147239263803681,
|
| 174 |
+
"grad_norm": 2.7534286975860596,
|
| 175 |
+
"learning_rate": 9.795918367346939e-06,
|
| 176 |
+
"loss": 0.8584,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.15337423312883436,
|
| 181 |
+
"grad_norm": 3.070441484451294,
|
| 182 |
+
"learning_rate": 1.0204081632653063e-05,
|
| 183 |
+
"loss": 0.8879,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.15950920245398773,
|
| 188 |
+
"grad_norm": 2.5512771606445312,
|
| 189 |
+
"learning_rate": 1.0612244897959186e-05,
|
| 190 |
+
"loss": 0.7313,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.1656441717791411,
|
| 195 |
+
"grad_norm": 2.626723289489746,
|
| 196 |
+
"learning_rate": 1.1020408163265306e-05,
|
| 197 |
+
"loss": 0.8822,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.17177914110429449,
|
| 202 |
+
"grad_norm": 2.6062428951263428,
|
| 203 |
+
"learning_rate": 1.1428571428571429e-05,
|
| 204 |
+
"loss": 0.9127,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.17791411042944785,
|
| 209 |
+
"grad_norm": 2.6317055225372314,
|
| 210 |
+
"learning_rate": 1.1836734693877552e-05,
|
| 211 |
+
"loss": 0.8449,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.18404907975460122,
|
| 216 |
+
"grad_norm": 3.30899715423584,
|
| 217 |
+
"learning_rate": 1.2244897959183674e-05,
|
| 218 |
+
"loss": 0.903,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.1901840490797546,
|
| 223 |
+
"grad_norm": 3.433353900909424,
|
| 224 |
+
"learning_rate": 1.2653061224489798e-05,
|
| 225 |
+
"loss": 0.9024,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.19631901840490798,
|
| 230 |
+
"grad_norm": 3.001012086868286,
|
| 231 |
+
"learning_rate": 1.3061224489795918e-05,
|
| 232 |
+
"loss": 0.8884,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.20245398773006135,
|
| 237 |
+
"grad_norm": 2.610701322555542,
|
| 238 |
+
"learning_rate": 1.3469387755102042e-05,
|
| 239 |
+
"loss": 0.892,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.2085889570552147,
|
| 244 |
+
"grad_norm": 2.6325395107269287,
|
| 245 |
+
"learning_rate": 1.3877551020408165e-05,
|
| 246 |
+
"loss": 0.8424,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.2147239263803681,
|
| 251 |
+
"grad_norm": 3.036200523376465,
|
| 252 |
+
"learning_rate": 1.4285714285714287e-05,
|
| 253 |
+
"loss": 0.9034,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.22085889570552147,
|
| 258 |
+
"grad_norm": 2.929934024810791,
|
| 259 |
+
"learning_rate": 1.469387755102041e-05,
|
| 260 |
+
"loss": 0.893,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.22699386503067484,
|
| 265 |
+
"grad_norm": 2.6639227867126465,
|
| 266 |
+
"learning_rate": 1.510204081632653e-05,
|
| 267 |
+
"loss": 0.9461,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.2331288343558282,
|
| 272 |
+
"grad_norm": 2.712769031524658,
|
| 273 |
+
"learning_rate": 1.5510204081632655e-05,
|
| 274 |
+
"loss": 0.9379,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.2392638036809816,
|
| 279 |
+
"grad_norm": 2.3838276863098145,
|
| 280 |
+
"learning_rate": 1.5918367346938776e-05,
|
| 281 |
+
"loss": 0.7864,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.24539877300613497,
|
| 286 |
+
"grad_norm": 2.565786600112915,
|
| 287 |
+
"learning_rate": 1.63265306122449e-05,
|
| 288 |
+
"loss": 0.8587,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.25153374233128833,
|
| 293 |
+
"grad_norm": 2.628082752227783,
|
| 294 |
+
"learning_rate": 1.673469387755102e-05,
|
| 295 |
+
"loss": 0.8506,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.25766871165644173,
|
| 300 |
+
"grad_norm": 2.246677875518799,
|
| 301 |
+
"learning_rate": 1.7142857142857142e-05,
|
| 302 |
+
"loss": 0.8663,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.26380368098159507,
|
| 307 |
+
"grad_norm": 2.5994741916656494,
|
| 308 |
+
"learning_rate": 1.7551020408163266e-05,
|
| 309 |
+
"loss": 0.7208,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.26993865030674846,
|
| 314 |
+
"grad_norm": 2.5801467895507812,
|
| 315 |
+
"learning_rate": 1.795918367346939e-05,
|
| 316 |
+
"loss": 0.823,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.27607361963190186,
|
| 321 |
+
"grad_norm": 2.2797186374664307,
|
| 322 |
+
"learning_rate": 1.836734693877551e-05,
|
| 323 |
+
"loss": 0.6901,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.2822085889570552,
|
| 328 |
+
"grad_norm": 2.310608386993408,
|
| 329 |
+
"learning_rate": 1.8775510204081636e-05,
|
| 330 |
+
"loss": 0.8623,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.2883435582822086,
|
| 335 |
+
"grad_norm": 3.0622611045837402,
|
| 336 |
+
"learning_rate": 1.9183673469387756e-05,
|
| 337 |
+
"loss": 0.9739,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.294478527607362,
|
| 342 |
+
"grad_norm": 2.8359763622283936,
|
| 343 |
+
"learning_rate": 1.9591836734693877e-05,
|
| 344 |
+
"loss": 0.8541,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3006134969325153,
|
| 349 |
+
"grad_norm": 2.4857027530670166,
|
| 350 |
+
"learning_rate": 2e-05,
|
| 351 |
+
"loss": 0.8031,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.3067484662576687,
|
| 356 |
+
"grad_norm": 2.9872865676879883,
|
| 357 |
+
"learning_rate": 1.9999745104274995e-05,
|
| 358 |
+
"loss": 0.9584,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.3128834355828221,
|
| 363 |
+
"grad_norm": 2.7176148891448975,
|
| 364 |
+
"learning_rate": 1.9998980430094333e-05,
|
| 365 |
+
"loss": 0.9524,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.31901840490797545,
|
| 370 |
+
"grad_norm": 2.694809675216675,
|
| 371 |
+
"learning_rate": 1.9997706016440462e-05,
|
| 372 |
+
"loss": 0.8608,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.32515337423312884,
|
| 377 |
+
"grad_norm": 2.1502082347869873,
|
| 378 |
+
"learning_rate": 1.9995921928281893e-05,
|
| 379 |
+
"loss": 0.8371,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.3312883435582822,
|
| 384 |
+
"grad_norm": 3.4349052906036377,
|
| 385 |
+
"learning_rate": 1.999362825656992e-05,
|
| 386 |
+
"loss": 0.9361,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.3374233128834356,
|
| 391 |
+
"grad_norm": 2.6445701122283936,
|
| 392 |
+
"learning_rate": 1.9990825118233958e-05,
|
| 393 |
+
"loss": 0.9325,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.34355828220858897,
|
| 398 |
+
"grad_norm": 2.9627561569213867,
|
| 399 |
+
"learning_rate": 1.9987512656175612e-05,
|
| 400 |
+
"loss": 0.9348,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.3496932515337423,
|
| 405 |
+
"grad_norm": 2.6833505630493164,
|
| 406 |
+
"learning_rate": 1.9983691039261358e-05,
|
| 407 |
+
"loss": 0.9207,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.3558282208588957,
|
| 412 |
+
"grad_norm": 2.7319250106811523,
|
| 413 |
+
"learning_rate": 1.9979360462313965e-05,
|
| 414 |
+
"loss": 0.8873,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.3619631901840491,
|
| 419 |
+
"grad_norm": 2.852752447128296,
|
| 420 |
+
"learning_rate": 1.9974521146102535e-05,
|
| 421 |
+
"loss": 0.833,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.36809815950920244,
|
| 426 |
+
"grad_norm": 2.443208694458008,
|
| 427 |
+
"learning_rate": 1.9969173337331283e-05,
|
| 428 |
+
"loss": 0.8861,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.37423312883435583,
|
| 433 |
+
"grad_norm": 2.418011426925659,
|
| 434 |
+
"learning_rate": 1.9963317308626916e-05,
|
| 435 |
+
"loss": 0.9211,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.3803680981595092,
|
| 440 |
+
"grad_norm": 2.879551887512207,
|
| 441 |
+
"learning_rate": 1.9956953358524774e-05,
|
| 442 |
+
"loss": 0.8202,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.38650306748466257,
|
| 447 |
+
"grad_norm": 2.1719465255737305,
|
| 448 |
+
"learning_rate": 1.9950081811453598e-05,
|
| 449 |
+
"loss": 0.8951,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.39263803680981596,
|
| 454 |
+
"grad_norm": 3.3376317024230957,
|
| 455 |
+
"learning_rate": 1.9942703017718977e-05,
|
| 456 |
+
"loss": 1.0127,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.3987730061349693,
|
| 461 |
+
"grad_norm": 2.1772871017456055,
|
| 462 |
+
"learning_rate": 1.99348173534855e-05,
|
| 463 |
+
"loss": 0.8279,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.4049079754601227,
|
| 468 |
+
"grad_norm": 2.4154765605926514,
|
| 469 |
+
"learning_rate": 1.9926425220757607e-05,
|
| 470 |
+
"loss": 0.8938,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.4110429447852761,
|
| 475 |
+
"grad_norm": 2.4074478149414062,
|
| 476 |
+
"learning_rate": 1.991752704735903e-05,
|
| 477 |
+
"loss": 0.838,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.4171779141104294,
|
| 482 |
+
"grad_norm": 2.972881555557251,
|
| 483 |
+
"learning_rate": 1.990812328691105e-05,
|
| 484 |
+
"loss": 0.9311,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.4233128834355828,
|
| 489 |
+
"grad_norm": 2.783582925796509,
|
| 490 |
+
"learning_rate": 1.989821441880933e-05,
|
| 491 |
+
"loss": 0.8724,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.4294478527607362,
|
| 496 |
+
"grad_norm": 2.5357515811920166,
|
| 497 |
+
"learning_rate": 1.9887800948199496e-05,
|
| 498 |
+
"loss": 0.855,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.43558282208588955,
|
| 503 |
+
"grad_norm": 2.7631185054779053,
|
| 504 |
+
"learning_rate": 1.9876883405951378e-05,
|
| 505 |
+
"loss": 0.8601,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.44171779141104295,
|
| 510 |
+
"grad_norm": 2.3920769691467285,
|
| 511 |
+
"learning_rate": 1.9865462348631945e-05,
|
| 512 |
+
"loss": 0.7703,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.44785276073619634,
|
| 517 |
+
"grad_norm": 2.4372503757476807,
|
| 518 |
+
"learning_rate": 1.9853538358476933e-05,
|
| 519 |
+
"loss": 1.0231,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.4539877300613497,
|
| 524 |
+
"grad_norm": 2.5493886470794678,
|
| 525 |
+
"learning_rate": 1.984111204336116e-05,
|
| 526 |
+
"loss": 0.9518,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.4601226993865031,
|
| 531 |
+
"grad_norm": 3.0651488304138184,
|
| 532 |
+
"learning_rate": 1.9828184036767556e-05,
|
| 533 |
+
"loss": 0.8844,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4662576687116564,
|
| 538 |
+
"grad_norm": 2.3102948665618896,
|
| 539 |
+
"learning_rate": 1.981475499775484e-05,
|
| 540 |
+
"loss": 0.8772,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.4723926380368098,
|
| 545 |
+
"grad_norm": 2.400557279586792,
|
| 546 |
+
"learning_rate": 1.9800825610923937e-05,
|
| 547 |
+
"loss": 0.8965,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.4785276073619632,
|
| 552 |
+
"grad_norm": 3.689486503601074,
|
| 553 |
+
"learning_rate": 1.9786396586383078e-05,
|
| 554 |
+
"loss": 0.9911,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.48466257668711654,
|
| 559 |
+
"grad_norm": 2.7310731410980225,
|
| 560 |
+
"learning_rate": 1.9771468659711595e-05,
|
| 561 |
+
"loss": 0.9617,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.49079754601226994,
|
| 566 |
+
"grad_norm": 1.9721884727478027,
|
| 567 |
+
"learning_rate": 1.9756042591922436e-05,
|
| 568 |
+
"loss": 0.818,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.49693251533742333,
|
| 573 |
+
"grad_norm": 2.744872570037842,
|
| 574 |
+
"learning_rate": 1.9740119169423337e-05,
|
| 575 |
+
"loss": 0.9164,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.5030674846625767,
|
| 580 |
+
"grad_norm": 2.2941346168518066,
|
| 581 |
+
"learning_rate": 1.9723699203976768e-05,
|
| 582 |
+
"loss": 0.7954,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.50920245398773,
|
| 587 |
+
"grad_norm": 2.417813777923584,
|
| 588 |
+
"learning_rate": 1.9706783532658528e-05,
|
| 589 |
+
"loss": 0.8837,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.5153374233128835,
|
| 594 |
+
"grad_norm": 2.3277711868286133,
|
| 595 |
+
"learning_rate": 1.9689373017815076e-05,
|
| 596 |
+
"loss": 0.8001,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.5214723926380368,
|
| 601 |
+
"grad_norm": 2.7406232357025146,
|
| 602 |
+
"learning_rate": 1.9671468547019575e-05,
|
| 603 |
+
"loss": 0.8329,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.5276073619631901,
|
| 608 |
+
"grad_norm": 2.6227080821990967,
|
| 609 |
+
"learning_rate": 1.9653071033026635e-05,
|
| 610 |
+
"loss": 0.8513,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.5337423312883436,
|
| 615 |
+
"grad_norm": 2.5143091678619385,
|
| 616 |
+
"learning_rate": 1.963418141372579e-05,
|
| 617 |
+
"loss": 0.9202,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.5398773006134969,
|
| 622 |
+
"grad_norm": 2.4019291400909424,
|
| 623 |
+
"learning_rate": 1.9614800652093685e-05,
|
| 624 |
+
"loss": 0.8449,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.5460122699386503,
|
| 629 |
+
"grad_norm": 2.5892465114593506,
|
| 630 |
+
"learning_rate": 1.9594929736144978e-05,
|
| 631 |
+
"loss": 0.9929,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.5521472392638037,
|
| 636 |
+
"grad_norm": 2.1739256381988525,
|
| 637 |
+
"learning_rate": 1.9574569678881965e-05,
|
| 638 |
+
"loss": 0.9703,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.558282208588957,
|
| 643 |
+
"grad_norm": 3.1401398181915283,
|
| 644 |
+
"learning_rate": 1.955372151824297e-05,
|
| 645 |
+
"loss": 0.889,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.5644171779141104,
|
| 650 |
+
"grad_norm": 2.510582208633423,
|
| 651 |
+
"learning_rate": 1.9532386317049387e-05,
|
| 652 |
+
"loss": 0.9673,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.5705521472392638,
|
| 657 |
+
"grad_norm": 3.579469680786133,
|
| 658 |
+
"learning_rate": 1.9510565162951538e-05,
|
| 659 |
+
"loss": 0.719,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.5766871165644172,
|
| 664 |
+
"grad_norm": 2.3586325645446777,
|
| 665 |
+
"learning_rate": 1.9488259168373198e-05,
|
| 666 |
+
"loss": 0.8154,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.5828220858895705,
|
| 671 |
+
"grad_norm": 2.8746159076690674,
|
| 672 |
+
"learning_rate": 1.94654694704549e-05,
|
| 673 |
+
"loss": 0.8888,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.588957055214724,
|
| 678 |
+
"grad_norm": 2.3770508766174316,
|
| 679 |
+
"learning_rate": 1.944219723099596e-05,
|
| 680 |
+
"loss": 0.9593,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.5950920245398773,
|
| 685 |
+
"grad_norm": 2.695714235305786,
|
| 686 |
+
"learning_rate": 1.941844363639525e-05,
|
| 687 |
+
"loss": 0.9273,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.6012269938650306,
|
| 692 |
+
"grad_norm": 1.9606412649154663,
|
| 693 |
+
"learning_rate": 1.9394209897590707e-05,
|
| 694 |
+
"loss": 0.8049,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.6073619631901841,
|
| 699 |
+
"grad_norm": 2.397045373916626,
|
| 700 |
+
"learning_rate": 1.936949724999762e-05,
|
| 701 |
+
"loss": 0.9284,
|
| 702 |
+
"step": 99
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.6134969325153374,
|
| 706 |
+
"grad_norm": 2.3692755699157715,
|
| 707 |
+
"learning_rate": 1.9344306953445632e-05,
|
| 708 |
+
"loss": 0.9154,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.6196319018404908,
|
| 713 |
+
"grad_norm": 2.067070722579956,
|
| 714 |
+
"learning_rate": 1.9318640292114526e-05,
|
| 715 |
+
"loss": 0.8195,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.6257668711656442,
|
| 720 |
+
"grad_norm": 2.9347798824310303,
|
| 721 |
+
"learning_rate": 1.929249857446875e-05,
|
| 722 |
+
"loss": 0.8749,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.6319018404907976,
|
| 727 |
+
"grad_norm": 2.1538584232330322,
|
| 728 |
+
"learning_rate": 1.9265883133190715e-05,
|
| 729 |
+
"loss": 0.7886,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.6380368098159509,
|
| 734 |
+
"grad_norm": 2.57013201713562,
|
| 735 |
+
"learning_rate": 1.9238795325112867e-05,
|
| 736 |
+
"loss": 0.9187,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.6441717791411042,
|
| 741 |
+
"grad_norm": 2.269765615463257,
|
| 742 |
+
"learning_rate": 1.92112365311485e-05,
|
| 743 |
+
"loss": 0.8611,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.6503067484662577,
|
| 748 |
+
"grad_norm": 2.1213631629943848,
|
| 749 |
+
"learning_rate": 1.918320815622137e-05,
|
| 750 |
+
"loss": 0.8781,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.656441717791411,
|
| 755 |
+
"grad_norm": 2.270547389984131,
|
| 756 |
+
"learning_rate": 1.9154711629194062e-05,
|
| 757 |
+
"loss": 0.8632,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.6625766871165644,
|
| 762 |
+
"grad_norm": 2.3239433765411377,
|
| 763 |
+
"learning_rate": 1.912574840279516e-05,
|
| 764 |
+
"loss": 0.852,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.6687116564417178,
|
| 769 |
+
"grad_norm": 2.4857397079467773,
|
| 770 |
+
"learning_rate": 1.9096319953545186e-05,
|
| 771 |
+
"loss": 0.9736,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.6748466257668712,
|
| 776 |
+
"grad_norm": 2.2042980194091797,
|
| 777 |
+
"learning_rate": 1.9066427781681314e-05,
|
| 778 |
+
"loss": 0.8445,
|
| 779 |
+
"step": 110
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.6809815950920245,
|
| 783 |
+
"grad_norm": 2.358870029449463,
|
| 784 |
+
"learning_rate": 1.9036073411080917e-05,
|
| 785 |
+
"loss": 0.9168,
|
| 786 |
+
"step": 111
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.6871165644171779,
|
| 790 |
+
"grad_norm": 2.455577850341797,
|
| 791 |
+
"learning_rate": 1.900525838918385e-05,
|
| 792 |
+
"loss": 0.8593,
|
| 793 |
+
"step": 112
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.6932515337423313,
|
| 797 |
+
"grad_norm": 2.4685747623443604,
|
| 798 |
+
"learning_rate": 1.8973984286913584e-05,
|
| 799 |
+
"loss": 0.8876,
|
| 800 |
+
"step": 113
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.6993865030674846,
|
| 804 |
+
"grad_norm": 2.4090356826782227,
|
| 805 |
+
"learning_rate": 1.8942252698597113e-05,
|
| 806 |
+
"loss": 0.9748,
|
| 807 |
+
"step": 114
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.7055214723926381,
|
| 811 |
+
"grad_norm": 2.0732617378234863,
|
| 812 |
+
"learning_rate": 1.891006524188368e-05,
|
| 813 |
+
"loss": 0.8228,
|
| 814 |
+
"step": 115
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.7116564417177914,
|
| 818 |
+
"grad_norm": 2.108085870742798,
|
| 819 |
+
"learning_rate": 1.8877423557662307e-05,
|
| 820 |
+
"loss": 0.9245,
|
| 821 |
+
"step": 116
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.7177914110429447,
|
| 825 |
+
"grad_norm": 2.048963785171509,
|
| 826 |
+
"learning_rate": 1.8844329309978146e-05,
|
| 827 |
+
"loss": 0.8165,
|
| 828 |
+
"step": 117
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.7239263803680982,
|
| 832 |
+
"grad_norm": 2.941418409347534,
|
| 833 |
+
"learning_rate": 1.8810784185947648e-05,
|
| 834 |
+
"loss": 0.8717,
|
| 835 |
+
"step": 118
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.7300613496932515,
|
| 839 |
+
"grad_norm": 1.9391118288040161,
|
| 840 |
+
"learning_rate": 1.8776789895672557e-05,
|
| 841 |
+
"loss": 0.8367,
|
| 842 |
+
"step": 119
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.7361963190184049,
|
| 846 |
+
"grad_norm": 1.9185856580734253,
|
| 847 |
+
"learning_rate": 1.8742348172152728e-05,
|
| 848 |
+
"loss": 0.8201,
|
| 849 |
+
"step": 120
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.7423312883435583,
|
| 853 |
+
"grad_norm": 3.1177749633789062,
|
| 854 |
+
"learning_rate": 1.8707460771197773e-05,
|
| 855 |
+
"loss": 0.8935,
|
| 856 |
+
"step": 121
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.7484662576687117,
|
| 860 |
+
"grad_norm": 2.598524570465088,
|
| 861 |
+
"learning_rate": 1.8672129471337568e-05,
|
| 862 |
+
"loss": 0.8738,
|
| 863 |
+
"step": 122
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.754601226993865,
|
| 867 |
+
"grad_norm": 2.9220356941223145,
|
| 868 |
+
"learning_rate": 1.863635607373157e-05,
|
| 869 |
+
"loss": 0.8156,
|
| 870 |
+
"step": 123
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.7607361963190185,
|
| 874 |
+
"grad_norm": 2.300952434539795,
|
| 875 |
+
"learning_rate": 1.8600142402077006e-05,
|
| 876 |
+
"loss": 0.8801,
|
| 877 |
+
"step": 124
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.7668711656441718,
|
| 881 |
+
"grad_norm": 1.8602063655853271,
|
| 882 |
+
"learning_rate": 1.856349030251589e-05,
|
| 883 |
+
"loss": 0.7957,
|
| 884 |
+
"step": 125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.7730061349693251,
|
| 888 |
+
"grad_norm": 2.1289467811584473,
|
| 889 |
+
"learning_rate": 1.8526401643540924e-05,
|
| 890 |
+
"loss": 0.7608,
|
| 891 |
+
"step": 126
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.7791411042944786,
|
| 895 |
+
"grad_norm": 2.544464349746704,
|
| 896 |
+
"learning_rate": 1.8488878315900228e-05,
|
| 897 |
+
"loss": 0.8282,
|
| 898 |
+
"step": 127
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.7852760736196319,
|
| 902 |
+
"grad_norm": 2.2640907764434814,
|
| 903 |
+
"learning_rate": 1.8450922232500966e-05,
|
| 904 |
+
"loss": 0.9357,
|
| 905 |
+
"step": 128
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.7914110429447853,
|
| 909 |
+
"grad_norm": 2.4785866737365723,
|
| 910 |
+
"learning_rate": 1.8412535328311813e-05,
|
| 911 |
+
"loss": 0.8431,
|
| 912 |
+
"step": 129
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.7975460122699386,
|
| 916 |
+
"grad_norm": 2.4250004291534424,
|
| 917 |
+
"learning_rate": 1.837371956026433e-05,
|
| 918 |
+
"loss": 0.8008,
|
| 919 |
+
"step": 130
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.803680981595092,
|
| 923 |
+
"grad_norm": 2.192577362060547,
|
| 924 |
+
"learning_rate": 1.8334476907153177e-05,
|
| 925 |
+
"loss": 0.8483,
|
| 926 |
+
"step": 131
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.8098159509202454,
|
| 930 |
+
"grad_norm": 2.5926589965820312,
|
| 931 |
+
"learning_rate": 1.8294809369535265e-05,
|
| 932 |
+
"loss": 0.9542,
|
| 933 |
+
"step": 132
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.8159509202453987,
|
| 937 |
+
"grad_norm": 2.1088614463806152,
|
| 938 |
+
"learning_rate": 1.825471896962774e-05,
|
| 939 |
+
"loss": 0.8832,
|
| 940 |
+
"step": 133
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.8220858895705522,
|
| 944 |
+
"grad_norm": 2.1007630825042725,
|
| 945 |
+
"learning_rate": 1.8214207751204917e-05,
|
| 946 |
+
"loss": 0.805,
|
| 947 |
+
"step": 134
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.8282208588957055,
|
| 951 |
+
"grad_norm": 2.3816447257995605,
|
| 952 |
+
"learning_rate": 1.817327777949407e-05,
|
| 953 |
+
"loss": 0.9024,
|
| 954 |
+
"step": 135
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.8343558282208589,
|
| 958 |
+
"grad_norm": 2.218423843383789,
|
| 959 |
+
"learning_rate": 1.8131931141070166e-05,
|
| 960 |
+
"loss": 0.8063,
|
| 961 |
+
"step": 136
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.8404907975460123,
|
| 965 |
+
"grad_norm": 2.4789438247680664,
|
| 966 |
+
"learning_rate": 1.8090169943749477e-05,
|
| 967 |
+
"loss": 0.972,
|
| 968 |
+
"step": 137
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.8466257668711656,
|
| 972 |
+
"grad_norm": 2.5738685131073,
|
| 973 |
+
"learning_rate": 1.8047996316482134e-05,
|
| 974 |
+
"loss": 0.9459,
|
| 975 |
+
"step": 138
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.852760736196319,
|
| 979 |
+
"grad_norm": 2.327671766281128,
|
| 980 |
+
"learning_rate": 1.8005412409243604e-05,
|
| 981 |
+
"loss": 0.8606,
|
| 982 |
+
"step": 139
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.8588957055214724,
|
| 986 |
+
"grad_norm": 2.1657285690307617,
|
| 987 |
+
"learning_rate": 1.7962420392925066e-05,
|
| 988 |
+
"loss": 0.8066,
|
| 989 |
+
"step": 140
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.8650306748466258,
|
| 993 |
+
"grad_norm": 2.4161083698272705,
|
| 994 |
+
"learning_rate": 1.7919022459222754e-05,
|
| 995 |
+
"loss": 0.8587,
|
| 996 |
+
"step": 141
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.8711656441717791,
|
| 1000 |
+
"grad_norm": 2.5296366214752197,
|
| 1001 |
+
"learning_rate": 1.787522082052622e-05,
|
| 1002 |
+
"loss": 0.9125,
|
| 1003 |
+
"step": 142
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.8773006134969326,
|
| 1007 |
+
"grad_norm": 2.3278584480285645,
|
| 1008 |
+
"learning_rate": 1.7831017709805555e-05,
|
| 1009 |
+
"loss": 0.7792,
|
| 1010 |
+
"step": 143
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.8834355828220859,
|
| 1014 |
+
"grad_norm": 2.0539119243621826,
|
| 1015 |
+
"learning_rate": 1.778641538049755e-05,
|
| 1016 |
+
"loss": 0.8548,
|
| 1017 |
+
"step": 144
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.8895705521472392,
|
| 1021 |
+
"grad_norm": 1.8124006986618042,
|
| 1022 |
+
"learning_rate": 1.7741416106390828e-05,
|
| 1023 |
+
"loss": 0.8701,
|
| 1024 |
+
"step": 145
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.8957055214723927,
|
| 1028 |
+
"grad_norm": 2.146286964416504,
|
| 1029 |
+
"learning_rate": 1.7696022181509892e-05,
|
| 1030 |
+
"loss": 0.9331,
|
| 1031 |
+
"step": 146
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.901840490797546,
|
| 1035 |
+
"grad_norm": 2.170487880706787,
|
| 1036 |
+
"learning_rate": 1.7650235919998234e-05,
|
| 1037 |
+
"loss": 0.8931,
|
| 1038 |
+
"step": 147
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.9079754601226994,
|
| 1042 |
+
"grad_norm": 2.424647331237793,
|
| 1043 |
+
"learning_rate": 1.7604059656000313e-05,
|
| 1044 |
+
"loss": 0.856,
|
| 1045 |
+
"step": 148
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.9141104294478528,
|
| 1049 |
+
"grad_norm": 2.6157801151275635,
|
| 1050 |
+
"learning_rate": 1.7557495743542586e-05,
|
| 1051 |
+
"loss": 0.9213,
|
| 1052 |
+
"step": 149
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.9202453987730062,
|
| 1056 |
+
"grad_norm": 2.3399035930633545,
|
| 1057 |
+
"learning_rate": 1.75105465564135e-05,
|
| 1058 |
+
"loss": 0.8497,
|
| 1059 |
+
"step": 150
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.9263803680981595,
|
| 1063 |
+
"grad_norm": 2.282757520675659,
|
| 1064 |
+
"learning_rate": 1.7463214488042472e-05,
|
| 1065 |
+
"loss": 0.8269,
|
| 1066 |
+
"step": 151
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.9325153374233128,
|
| 1070 |
+
"grad_norm": 2.3872690200805664,
|
| 1071 |
+
"learning_rate": 1.741550195137788e-05,
|
| 1072 |
+
"loss": 0.8679,
|
| 1073 |
+
"step": 152
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.9386503067484663,
|
| 1077 |
+
"grad_norm": 2.130729913711548,
|
| 1078 |
+
"learning_rate": 1.736741137876405e-05,
|
| 1079 |
+
"loss": 0.8752,
|
| 1080 |
+
"step": 153
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.9447852760736196,
|
| 1084 |
+
"grad_norm": 2.3011746406555176,
|
| 1085 |
+
"learning_rate": 1.7318945221817255e-05,
|
| 1086 |
+
"loss": 0.8294,
|
| 1087 |
+
"step": 154
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.950920245398773,
|
| 1091 |
+
"grad_norm": 2.315067768096924,
|
| 1092 |
+
"learning_rate": 1.727010595130074e-05,
|
| 1093 |
+
"loss": 0.7756,
|
| 1094 |
+
"step": 155
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.9570552147239264,
|
| 1098 |
+
"grad_norm": 2.5285987854003906,
|
| 1099 |
+
"learning_rate": 1.7220896056998753e-05,
|
| 1100 |
+
"loss": 0.902,
|
| 1101 |
+
"step": 156
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.9631901840490797,
|
| 1105 |
+
"grad_norm": 2.388417959213257,
|
| 1106 |
+
"learning_rate": 1.7171318047589637e-05,
|
| 1107 |
+
"loss": 0.8883,
|
| 1108 |
+
"step": 157
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.9693251533742331,
|
| 1112 |
+
"grad_norm": 2.3286430835723877,
|
| 1113 |
+
"learning_rate": 1.712137445051792e-05,
|
| 1114 |
+
"loss": 0.8306,
|
| 1115 |
+
"step": 158
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.9754601226993865,
|
| 1119 |
+
"grad_norm": 2.6338658332824707,
|
| 1120 |
+
"learning_rate": 1.7071067811865477e-05,
|
| 1121 |
+
"loss": 0.9297,
|
| 1122 |
+
"step": 159
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.9815950920245399,
|
| 1126 |
+
"grad_norm": 2.3402535915374756,
|
| 1127 |
+
"learning_rate": 1.7020400696221737e-05,
|
| 1128 |
+
"loss": 0.8577,
|
| 1129 |
+
"step": 160
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.9877300613496932,
|
| 1133 |
+
"grad_norm": 1.9372575283050537,
|
| 1134 |
+
"learning_rate": 1.696937568655294e-05,
|
| 1135 |
+
"loss": 0.8977,
|
| 1136 |
+
"step": 161
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.9938650306748467,
|
| 1140 |
+
"grad_norm": 2.309221029281616,
|
| 1141 |
+
"learning_rate": 1.691799538407044e-05,
|
| 1142 |
+
"loss": 0.8754,
|
| 1143 |
+
"step": 162
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.0,
|
| 1147 |
+
"grad_norm": 2.1249825954437256,
|
| 1148 |
+
"learning_rate": 1.6866262408098134e-05,
|
| 1149 |
+
"loss": 0.8013,
|
| 1150 |
+
"step": 163
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.0061349693251533,
|
| 1154 |
+
"grad_norm": 2.3482613563537598,
|
| 1155 |
+
"learning_rate": 1.6814179395938915e-05,
|
| 1156 |
+
"loss": 0.5222,
|
| 1157 |
+
"step": 164
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.0122699386503067,
|
| 1161 |
+
"grad_norm": 1.973677396774292,
|
| 1162 |
+
"learning_rate": 1.6761749002740195e-05,
|
| 1163 |
+
"loss": 0.4045,
|
| 1164 |
+
"step": 165
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.01840490797546,
|
| 1168 |
+
"grad_norm": 1.9813287258148193,
|
| 1169 |
+
"learning_rate": 1.6708973901358603e-05,
|
| 1170 |
+
"loss": 0.4973,
|
| 1171 |
+
"step": 166
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.0245398773006136,
|
| 1175 |
+
"grad_norm": 1.8581576347351074,
|
| 1176 |
+
"learning_rate": 1.6655856782223682e-05,
|
| 1177 |
+
"loss": 0.4646,
|
| 1178 |
+
"step": 167
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 1.030674846625767,
|
| 1182 |
+
"grad_norm": 2.344332456588745,
|
| 1183 |
+
"learning_rate": 1.660240035320075e-05,
|
| 1184 |
+
"loss": 0.5098,
|
| 1185 |
+
"step": 168
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.0368098159509203,
|
| 1189 |
+
"grad_norm": 2.2874767780303955,
|
| 1190 |
+
"learning_rate": 1.6548607339452853e-05,
|
| 1191 |
+
"loss": 0.4462,
|
| 1192 |
+
"step": 169
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 1.0429447852760736,
|
| 1196 |
+
"grad_norm": 2.477348566055298,
|
| 1197 |
+
"learning_rate": 1.6494480483301836e-05,
|
| 1198 |
+
"loss": 0.5243,
|
| 1199 |
+
"step": 170
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.049079754601227,
|
| 1203 |
+
"grad_norm": 2.170513868331909,
|
| 1204 |
+
"learning_rate": 1.6440022544088553e-05,
|
| 1205 |
+
"loss": 0.4155,
|
| 1206 |
+
"step": 171
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.0552147239263803,
|
| 1210 |
+
"grad_norm": 2.501122236251831,
|
| 1211 |
+
"learning_rate": 1.6385236298032183e-05,
|
| 1212 |
+
"loss": 0.4516,
|
| 1213 |
+
"step": 172
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.0613496932515338,
|
| 1217 |
+
"grad_norm": 2.3615634441375732,
|
| 1218 |
+
"learning_rate": 1.6330124538088705e-05,
|
| 1219 |
+
"loss": 0.5354,
|
| 1220 |
+
"step": 173
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 1.0674846625766872,
|
| 1224 |
+
"grad_norm": 1.9157105684280396,
|
| 1225 |
+
"learning_rate": 1.627469007380852e-05,
|
| 1226 |
+
"loss": 0.3558,
|
| 1227 |
+
"step": 174
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.0736196319018405,
|
| 1231 |
+
"grad_norm": 2.4090816974639893,
|
| 1232 |
+
"learning_rate": 1.6218935731193223e-05,
|
| 1233 |
+
"loss": 0.457,
|
| 1234 |
+
"step": 175
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 1.0797546012269938,
|
| 1238 |
+
"grad_norm": 2.0153863430023193,
|
| 1239 |
+
"learning_rate": 1.616286435255153e-05,
|
| 1240 |
+
"loss": 0.4172,
|
| 1241 |
+
"step": 176
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.0858895705521472,
|
| 1245 |
+
"grad_norm": 1.9163466691970825,
|
| 1246 |
+
"learning_rate": 1.6106478796354382e-05,
|
| 1247 |
+
"loss": 0.3667,
|
| 1248 |
+
"step": 177
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.0920245398773005,
|
| 1252 |
+
"grad_norm": 2.265932321548462,
|
| 1253 |
+
"learning_rate": 1.6049781937089227e-05,
|
| 1254 |
+
"loss": 0.4438,
|
| 1255 |
+
"step": 178
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.098159509202454,
|
| 1259 |
+
"grad_norm": 2.421759605407715,
|
| 1260 |
+
"learning_rate": 1.599277666511347e-05,
|
| 1261 |
+
"loss": 0.401,
|
| 1262 |
+
"step": 179
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.1042944785276074,
|
| 1266 |
+
"grad_norm": 2.2517905235290527,
|
| 1267 |
+
"learning_rate": 1.5935465886507143e-05,
|
| 1268 |
+
"loss": 0.3927,
|
| 1269 |
+
"step": 180
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.1104294478527608,
|
| 1273 |
+
"grad_norm": 2.1347973346710205,
|
| 1274 |
+
"learning_rate": 1.5877852522924733e-05,
|
| 1275 |
+
"loss": 0.4361,
|
| 1276 |
+
"step": 181
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.116564417177914,
|
| 1280 |
+
"grad_norm": 2.019383430480957,
|
| 1281 |
+
"learning_rate": 1.581993951144626e-05,
|
| 1282 |
+
"loss": 0.405,
|
| 1283 |
+
"step": 182
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.1226993865030674,
|
| 1287 |
+
"grad_norm": 1.9620651006698608,
|
| 1288 |
+
"learning_rate": 1.576172980442753e-05,
|
| 1289 |
+
"loss": 0.3337,
|
| 1290 |
+
"step": 183
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.1288343558282208,
|
| 1294 |
+
"grad_norm": 2.187609910964966,
|
| 1295 |
+
"learning_rate": 1.5703226369349642e-05,
|
| 1296 |
+
"loss": 0.4073,
|
| 1297 |
+
"step": 184
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.1349693251533743,
|
| 1301 |
+
"grad_norm": 2.4111270904541016,
|
| 1302 |
+
"learning_rate": 1.5644432188667695e-05,
|
| 1303 |
+
"loss": 0.4582,
|
| 1304 |
+
"step": 185
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.1411042944785277,
|
| 1308 |
+
"grad_norm": 2.0801773071289062,
|
| 1309 |
+
"learning_rate": 1.5585350259658748e-05,
|
| 1310 |
+
"loss": 0.4038,
|
| 1311 |
+
"step": 186
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.147239263803681,
|
| 1315 |
+
"grad_norm": 3.0988597869873047,
|
| 1316 |
+
"learning_rate": 1.5525983594269026e-05,
|
| 1317 |
+
"loss": 0.4522,
|
| 1318 |
+
"step": 187
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.1533742331288344,
|
| 1322 |
+
"grad_norm": 2.858125925064087,
|
| 1323 |
+
"learning_rate": 1.546633521896038e-05,
|
| 1324 |
+
"loss": 0.4567,
|
| 1325 |
+
"step": 188
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.1595092024539877,
|
| 1329 |
+
"grad_norm": 2.1759862899780273,
|
| 1330 |
+
"learning_rate": 1.5406408174555978e-05,
|
| 1331 |
+
"loss": 0.3126,
|
| 1332 |
+
"step": 189
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 1.165644171779141,
|
| 1336 |
+
"grad_norm": 2.0893306732177734,
|
| 1337 |
+
"learning_rate": 1.5346205516085305e-05,
|
| 1338 |
+
"loss": 0.5273,
|
| 1339 |
+
"step": 190
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.1717791411042944,
|
| 1343 |
+
"grad_norm": 2.098989725112915,
|
| 1344 |
+
"learning_rate": 1.528573031262842e-05,
|
| 1345 |
+
"loss": 0.3813,
|
| 1346 |
+
"step": 191
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.177914110429448,
|
| 1350 |
+
"grad_norm": 2.873368740081787,
|
| 1351 |
+
"learning_rate": 1.5224985647159489e-05,
|
| 1352 |
+
"loss": 0.4485,
|
| 1353 |
+
"step": 192
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.1840490797546013,
|
| 1357 |
+
"grad_norm": 1.8973687887191772,
|
| 1358 |
+
"learning_rate": 1.5163974616389621e-05,
|
| 1359 |
+
"loss": 0.4054,
|
| 1360 |
+
"step": 193
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.1901840490797546,
|
| 1364 |
+
"grad_norm": 2.1408369541168213,
|
| 1365 |
+
"learning_rate": 1.5102700330609e-05,
|
| 1366 |
+
"loss": 0.3483,
|
| 1367 |
+
"step": 194
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.196319018404908,
|
| 1371 |
+
"grad_norm": 2.0534722805023193,
|
| 1372 |
+
"learning_rate": 1.504116591352832e-05,
|
| 1373 |
+
"loss": 0.4797,
|
| 1374 |
+
"step": 195
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.2024539877300613,
|
| 1378 |
+
"grad_norm": 1.9064143896102905,
|
| 1379 |
+
"learning_rate": 1.497937450211956e-05,
|
| 1380 |
+
"loss": 0.3867,
|
| 1381 |
+
"step": 196
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.2085889570552146,
|
| 1385 |
+
"grad_norm": 2.117337226867676,
|
| 1386 |
+
"learning_rate": 1.491732924645604e-05,
|
| 1387 |
+
"loss": 0.3846,
|
| 1388 |
+
"step": 197
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.2147239263803682,
|
| 1392 |
+
"grad_norm": 2.1390268802642822,
|
| 1393 |
+
"learning_rate": 1.4855033309551842e-05,
|
| 1394 |
+
"loss": 0.435,
|
| 1395 |
+
"step": 198
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.2208588957055215,
|
| 1399 |
+
"grad_norm": 2.7396576404571533,
|
| 1400 |
+
"learning_rate": 1.479248986720057e-05,
|
| 1401 |
+
"loss": 0.5327,
|
| 1402 |
+
"step": 199
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.2269938650306749,
|
| 1406 |
+
"grad_norm": 1.960433840751648,
|
| 1407 |
+
"learning_rate": 1.4729702107813438e-05,
|
| 1408 |
+
"loss": 0.2784,
|
| 1409 |
+
"step": 200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.2331288343558282,
|
| 1413 |
+
"grad_norm": 1.788409948348999,
|
| 1414 |
+
"learning_rate": 1.4666673232256738e-05,
|
| 1415 |
+
"loss": 0.4288,
|
| 1416 |
+
"step": 201
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 1.2392638036809815,
|
| 1420 |
+
"grad_norm": 1.9805573225021362,
|
| 1421 |
+
"learning_rate": 1.4603406453688656e-05,
|
| 1422 |
+
"loss": 0.3871,
|
| 1423 |
+
"step": 202
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 1.2453987730061349,
|
| 1427 |
+
"grad_norm": 2.4482967853546143,
|
| 1428 |
+
"learning_rate": 1.4539904997395468e-05,
|
| 1429 |
+
"loss": 0.5019,
|
| 1430 |
+
"step": 203
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.2515337423312882,
|
| 1434 |
+
"grad_norm": 1.9358989000320435,
|
| 1435 |
+
"learning_rate": 1.4476172100627127e-05,
|
| 1436 |
+
"loss": 0.394,
|
| 1437 |
+
"step": 204
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 1.2576687116564418,
|
| 1441 |
+
"grad_norm": 2.250089168548584,
|
| 1442 |
+
"learning_rate": 1.4412211012432213e-05,
|
| 1443 |
+
"loss": 0.3905,
|
| 1444 |
+
"step": 205
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.2638036809815951,
|
| 1448 |
+
"grad_norm": 2.763105869293213,
|
| 1449 |
+
"learning_rate": 1.4348024993492323e-05,
|
| 1450 |
+
"loss": 0.3862,
|
| 1451 |
+
"step": 206
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 1.2699386503067485,
|
| 1455 |
+
"grad_norm": 2.2534821033477783,
|
| 1456 |
+
"learning_rate": 1.4283617315955815e-05,
|
| 1457 |
+
"loss": 0.4278,
|
| 1458 |
+
"step": 207
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.2760736196319018,
|
| 1462 |
+
"grad_norm": 2.545742988586426,
|
| 1463 |
+
"learning_rate": 1.4218991263271024e-05,
|
| 1464 |
+
"loss": 0.3977,
|
| 1465 |
+
"step": 208
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.2822085889570551,
|
| 1469 |
+
"grad_norm": 1.8165545463562012,
|
| 1470 |
+
"learning_rate": 1.4154150130018867e-05,
|
| 1471 |
+
"loss": 0.3965,
|
| 1472 |
+
"step": 209
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 1.2883435582822087,
|
| 1476 |
+
"grad_norm": 1.7550116777420044,
|
| 1477 |
+
"learning_rate": 1.408909722174487e-05,
|
| 1478 |
+
"loss": 0.3701,
|
| 1479 |
+
"step": 210
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.294478527607362,
|
| 1483 |
+
"grad_norm": 2.0871880054473877,
|
| 1484 |
+
"learning_rate": 1.4023835854790682e-05,
|
| 1485 |
+
"loss": 0.4153,
|
| 1486 |
+
"step": 211
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 1.3006134969325154,
|
| 1490 |
+
"grad_norm": 1.8455545902252197,
|
| 1491 |
+
"learning_rate": 1.3958369356124986e-05,
|
| 1492 |
+
"loss": 0.4016,
|
| 1493 |
+
"step": 212
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.3067484662576687,
|
| 1497 |
+
"grad_norm": 2.1923511028289795,
|
| 1498 |
+
"learning_rate": 1.3892701063173917e-05,
|
| 1499 |
+
"loss": 0.3367,
|
| 1500 |
+
"step": 213
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.312883435582822,
|
| 1504 |
+
"grad_norm": 1.966577172279358,
|
| 1505 |
+
"learning_rate": 1.3826834323650899e-05,
|
| 1506 |
+
"loss": 0.4166,
|
| 1507 |
+
"step": 214
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 1.3190184049079754,
|
| 1511 |
+
"grad_norm": 1.6042200326919556,
|
| 1512 |
+
"learning_rate": 1.3760772495385998e-05,
|
| 1513 |
+
"loss": 0.3382,
|
| 1514 |
+
"step": 215
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.3251533742331287,
|
| 1518 |
+
"grad_norm": 2.244147300720215,
|
| 1519 |
+
"learning_rate": 1.369451894615474e-05,
|
| 1520 |
+
"loss": 0.4485,
|
| 1521 |
+
"step": 216
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.331288343558282,
|
| 1525 |
+
"grad_norm": 2.9676880836486816,
|
| 1526 |
+
"learning_rate": 1.362807705350641e-05,
|
| 1527 |
+
"loss": 0.3401,
|
| 1528 |
+
"step": 217
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 1.3374233128834356,
|
| 1532 |
+
"grad_norm": 2.3296144008636475,
|
| 1533 |
+
"learning_rate": 1.3561450204591898e-05,
|
| 1534 |
+
"loss": 0.4417,
|
| 1535 |
+
"step": 218
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 1.343558282208589,
|
| 1539 |
+
"grad_norm": 1.8303943872451782,
|
| 1540 |
+
"learning_rate": 1.3494641795990986e-05,
|
| 1541 |
+
"loss": 0.4166,
|
| 1542 |
+
"step": 219
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.3496932515337423,
|
| 1546 |
+
"grad_norm": 2.0545692443847656,
|
| 1547 |
+
"learning_rate": 1.3427655233539227e-05,
|
| 1548 |
+
"loss": 0.4432,
|
| 1549 |
+
"step": 220
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 1.3558282208588956,
|
| 1553 |
+
"grad_norm": 2.2152485847473145,
|
| 1554 |
+
"learning_rate": 1.3360493932154301e-05,
|
| 1555 |
+
"loss": 0.3649,
|
| 1556 |
+
"step": 221
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 1.3619631901840492,
|
| 1560 |
+
"grad_norm": 2.0159170627593994,
|
| 1561 |
+
"learning_rate": 1.3293161315661934e-05,
|
| 1562 |
+
"loss": 0.3905,
|
| 1563 |
+
"step": 222
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.3680981595092025,
|
| 1567 |
+
"grad_norm": 2.4724018573760986,
|
| 1568 |
+
"learning_rate": 1.3225660816621342e-05,
|
| 1569 |
+
"loss": 0.437,
|
| 1570 |
+
"step": 223
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.3742331288343559,
|
| 1574 |
+
"grad_norm": 1.973275065422058,
|
| 1575 |
+
"learning_rate": 1.3157995876150252e-05,
|
| 1576 |
+
"loss": 0.4269,
|
| 1577 |
+
"step": 224
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 1.3803680981595092,
|
| 1581 |
+
"grad_norm": 3.2629191875457764,
|
| 1582 |
+
"learning_rate": 1.3090169943749475e-05,
|
| 1583 |
+
"loss": 0.5649,
|
| 1584 |
+
"step": 225
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 1.3865030674846626,
|
| 1588 |
+
"grad_norm": 2.2407352924346924,
|
| 1589 |
+
"learning_rate": 1.302218647712706e-05,
|
| 1590 |
+
"loss": 0.3995,
|
| 1591 |
+
"step": 226
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 1.392638036809816,
|
| 1595 |
+
"grad_norm": 2.426779270172119,
|
| 1596 |
+
"learning_rate": 1.2954048942022002e-05,
|
| 1597 |
+
"loss": 0.433,
|
| 1598 |
+
"step": 227
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 1.3987730061349692,
|
| 1602 |
+
"grad_norm": 2.6514222621917725,
|
| 1603 |
+
"learning_rate": 1.288576081202759e-05,
|
| 1604 |
+
"loss": 0.4423,
|
| 1605 |
+
"step": 228
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 1.4049079754601226,
|
| 1609 |
+
"grad_norm": 2.028599739074707,
|
| 1610 |
+
"learning_rate": 1.2817325568414299e-05,
|
| 1611 |
+
"loss": 0.3972,
|
| 1612 |
+
"step": 229
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 1.4110429447852761,
|
| 1616 |
+
"grad_norm": 2.0600152015686035,
|
| 1617 |
+
"learning_rate": 1.2748746699952338e-05,
|
| 1618 |
+
"loss": 0.3693,
|
| 1619 |
+
"step": 230
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 1.4171779141104295,
|
| 1623 |
+
"grad_norm": 2.3788952827453613,
|
| 1624 |
+
"learning_rate": 1.2680027702733791e-05,
|
| 1625 |
+
"loss": 0.4398,
|
| 1626 |
+
"step": 231
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 1.4233128834355828,
|
| 1630 |
+
"grad_norm": 2.199835777282715,
|
| 1631 |
+
"learning_rate": 1.2611172079994377e-05,
|
| 1632 |
+
"loss": 0.3991,
|
| 1633 |
+
"step": 232
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 1.4294478527607362,
|
| 1637 |
+
"grad_norm": 2.5521061420440674,
|
| 1638 |
+
"learning_rate": 1.2542183341934873e-05,
|
| 1639 |
+
"loss": 0.4346,
|
| 1640 |
+
"step": 233
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 1.4355828220858895,
|
| 1644 |
+
"grad_norm": 1.8648345470428467,
|
| 1645 |
+
"learning_rate": 1.2473065005542155e-05,
|
| 1646 |
+
"loss": 0.3417,
|
| 1647 |
+
"step": 234
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.441717791411043,
|
| 1651 |
+
"grad_norm": 3.026027202606201,
|
| 1652 |
+
"learning_rate": 1.2403820594409926e-05,
|
| 1653 |
+
"loss": 0.4322,
|
| 1654 |
+
"step": 235
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 1.4478527607361964,
|
| 1658 |
+
"grad_norm": 1.9645586013793945,
|
| 1659 |
+
"learning_rate": 1.2334453638559057e-05,
|
| 1660 |
+
"loss": 0.3501,
|
| 1661 |
+
"step": 236
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 1.4539877300613497,
|
| 1665 |
+
"grad_norm": 2.4796903133392334,
|
| 1666 |
+
"learning_rate": 1.2264967674257647e-05,
|
| 1667 |
+
"loss": 0.4666,
|
| 1668 |
+
"step": 237
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 1.460122699386503,
|
| 1672 |
+
"grad_norm": 2.030738592147827,
|
| 1673 |
+
"learning_rate": 1.2195366243840745e-05,
|
| 1674 |
+
"loss": 0.4055,
|
| 1675 |
+
"step": 238
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.4662576687116564,
|
| 1679 |
+
"grad_norm": 2.0176825523376465,
|
| 1680 |
+
"learning_rate": 1.2125652895529766e-05,
|
| 1681 |
+
"loss": 0.4318,
|
| 1682 |
+
"step": 239
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.4723926380368098,
|
| 1686 |
+
"grad_norm": 2.9265341758728027,
|
| 1687 |
+
"learning_rate": 1.2055831183251608e-05,
|
| 1688 |
+
"loss": 0.5002,
|
| 1689 |
+
"step": 240
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.478527607361963,
|
| 1693 |
+
"grad_norm": 2.699991226196289,
|
| 1694 |
+
"learning_rate": 1.1985904666457455e-05,
|
| 1695 |
+
"loss": 0.4764,
|
| 1696 |
+
"step": 241
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 1.4846625766871164,
|
| 1700 |
+
"grad_norm": 2.5206851959228516,
|
| 1701 |
+
"learning_rate": 1.1915876909941356e-05,
|
| 1702 |
+
"loss": 0.4183,
|
| 1703 |
+
"step": 242
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 1.49079754601227,
|
| 1707 |
+
"grad_norm": 2.3106908798217773,
|
| 1708 |
+
"learning_rate": 1.1845751483658454e-05,
|
| 1709 |
+
"loss": 0.3822,
|
| 1710 |
+
"step": 243
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 1.4969325153374233,
|
| 1714 |
+
"grad_norm": 2.0179994106292725,
|
| 1715 |
+
"learning_rate": 1.1775531962543036e-05,
|
| 1716 |
+
"loss": 0.4404,
|
| 1717 |
+
"step": 244
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 1.5030674846625767,
|
| 1721 |
+
"grad_norm": 2.0689287185668945,
|
| 1722 |
+
"learning_rate": 1.170522192632624e-05,
|
| 1723 |
+
"loss": 0.4801,
|
| 1724 |
+
"step": 245
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 1.50920245398773,
|
| 1728 |
+
"grad_norm": 6.782520294189453,
|
| 1729 |
+
"learning_rate": 1.1634824959353602e-05,
|
| 1730 |
+
"loss": 0.3944,
|
| 1731 |
+
"step": 246
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 1.5153374233128836,
|
| 1735 |
+
"grad_norm": 2.3424599170684814,
|
| 1736 |
+
"learning_rate": 1.156434465040231e-05,
|
| 1737 |
+
"loss": 0.4614,
|
| 1738 |
+
"step": 247
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 1.521472392638037,
|
| 1742 |
+
"grad_norm": 2.1660568714141846,
|
| 1743 |
+
"learning_rate": 1.1493784592498252e-05,
|
| 1744 |
+
"loss": 0.4038,
|
| 1745 |
+
"step": 248
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 1.5276073619631902,
|
| 1749 |
+
"grad_norm": 2.2487616539001465,
|
| 1750 |
+
"learning_rate": 1.1423148382732854e-05,
|
| 1751 |
+
"loss": 0.4838,
|
| 1752 |
+
"step": 249
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 1.5337423312883436,
|
| 1756 |
+
"grad_norm": 2.6163601875305176,
|
| 1757 |
+
"learning_rate": 1.1352439622079689e-05,
|
| 1758 |
+
"loss": 0.4007,
|
| 1759 |
+
"step": 250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 1.539877300613497,
|
| 1763 |
+
"grad_norm": 2.3755905628204346,
|
| 1764 |
+
"learning_rate": 1.1281661915210931e-05,
|
| 1765 |
+
"loss": 0.4882,
|
| 1766 |
+
"step": 251
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 1.5460122699386503,
|
| 1770 |
+
"grad_norm": 1.7966959476470947,
|
| 1771 |
+
"learning_rate": 1.1210818870313548e-05,
|
| 1772 |
+
"loss": 0.3751,
|
| 1773 |
+
"step": 252
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 1.5521472392638036,
|
| 1777 |
+
"grad_norm": 2.3187804222106934,
|
| 1778 |
+
"learning_rate": 1.1139914098905406e-05,
|
| 1779 |
+
"loss": 0.4446,
|
| 1780 |
+
"step": 253
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 1.558282208588957,
|
| 1784 |
+
"grad_norm": 2.3249504566192627,
|
| 1785 |
+
"learning_rate": 1.1068951215651132e-05,
|
| 1786 |
+
"loss": 0.4888,
|
| 1787 |
+
"step": 254
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 1.5644171779141103,
|
| 1791 |
+
"grad_norm": 2.1338675022125244,
|
| 1792 |
+
"learning_rate": 1.0997933838177828e-05,
|
| 1793 |
+
"loss": 0.409,
|
| 1794 |
+
"step": 255
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 1.5705521472392638,
|
| 1798 |
+
"grad_norm": 1.8690897226333618,
|
| 1799 |
+
"learning_rate": 1.0926865586890689e-05,
|
| 1800 |
+
"loss": 0.3262,
|
| 1801 |
+
"step": 256
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 1.5766871165644172,
|
| 1805 |
+
"grad_norm": 2.41798996925354,
|
| 1806 |
+
"learning_rate": 1.08557500847884e-05,
|
| 1807 |
+
"loss": 0.3936,
|
| 1808 |
+
"step": 257
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 1.5828220858895705,
|
| 1812 |
+
"grad_norm": 2.0561769008636475,
|
| 1813 |
+
"learning_rate": 1.0784590957278452e-05,
|
| 1814 |
+
"loss": 0.4274,
|
| 1815 |
+
"step": 258
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 1.588957055214724,
|
| 1819 |
+
"grad_norm": 2.1077218055725098,
|
| 1820 |
+
"learning_rate": 1.0713391831992324e-05,
|
| 1821 |
+
"loss": 0.3786,
|
| 1822 |
+
"step": 259
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 1.5950920245398774,
|
| 1826 |
+
"grad_norm": 1.864588975906372,
|
| 1827 |
+
"learning_rate": 1.064215633860055e-05,
|
| 1828 |
+
"loss": 0.372,
|
| 1829 |
+
"step": 260
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 1.6012269938650308,
|
| 1833 |
+
"grad_norm": 1.828682541847229,
|
| 1834 |
+
"learning_rate": 1.0570888108627682e-05,
|
| 1835 |
+
"loss": 0.34,
|
| 1836 |
+
"step": 261
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 1.607361963190184,
|
| 1840 |
+
"grad_norm": 2.2353274822235107,
|
| 1841 |
+
"learning_rate": 1.0499590775267142e-05,
|
| 1842 |
+
"loss": 0.386,
|
| 1843 |
+
"step": 262
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 1.6134969325153374,
|
| 1847 |
+
"grad_norm": 2.1544179916381836,
|
| 1848 |
+
"learning_rate": 1.0428267973196027e-05,
|
| 1849 |
+
"loss": 0.3656,
|
| 1850 |
+
"step": 263
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 1.6196319018404908,
|
| 1854 |
+
"grad_norm": 2.5289499759674072,
|
| 1855 |
+
"learning_rate": 1.0356923338389807e-05,
|
| 1856 |
+
"loss": 0.4314,
|
| 1857 |
+
"step": 264
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.6257668711656441,
|
| 1861 |
+
"grad_norm": 2.1689231395721436,
|
| 1862 |
+
"learning_rate": 1.0285560507936962e-05,
|
| 1863 |
+
"loss": 0.4035,
|
| 1864 |
+
"step": 265
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 1.6319018404907975,
|
| 1868 |
+
"grad_norm": 2.546900749206543,
|
| 1869 |
+
"learning_rate": 1.0214183119853583e-05,
|
| 1870 |
+
"loss": 0.4407,
|
| 1871 |
+
"step": 266
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 1.6380368098159508,
|
| 1875 |
+
"grad_norm": 1.9981437921524048,
|
| 1876 |
+
"learning_rate": 1.0142794812897874e-05,
|
| 1877 |
+
"loss": 0.3938,
|
| 1878 |
+
"step": 267
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 1.6441717791411041,
|
| 1882 |
+
"grad_norm": 1.9977840185165405,
|
| 1883 |
+
"learning_rate": 1.0071399226384695e-05,
|
| 1884 |
+
"loss": 0.4305,
|
| 1885 |
+
"step": 268
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 1.6503067484662577,
|
| 1889 |
+
"grad_norm": 1.8538576364517212,
|
| 1890 |
+
"learning_rate": 1e-05,
|
| 1891 |
+
"loss": 0.3431,
|
| 1892 |
+
"step": 269
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 1.656441717791411,
|
| 1896 |
+
"grad_norm": 2.5277135372161865,
|
| 1897 |
+
"learning_rate": 9.928600773615306e-06,
|
| 1898 |
+
"loss": 0.4318,
|
| 1899 |
+
"step": 270
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 1.6625766871165644,
|
| 1903 |
+
"grad_norm": 2.2893011569976807,
|
| 1904 |
+
"learning_rate": 9.85720518710213e-06,
|
| 1905 |
+
"loss": 0.3972,
|
| 1906 |
+
"step": 271
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 1.668711656441718,
|
| 1910 |
+
"grad_norm": 1.7950539588928223,
|
| 1911 |
+
"learning_rate": 9.78581688014642e-06,
|
| 1912 |
+
"loss": 0.4252,
|
| 1913 |
+
"step": 272
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 1.6748466257668713,
|
| 1917 |
+
"grad_norm": 2.052996873855591,
|
| 1918 |
+
"learning_rate": 9.71443949206304e-06,
|
| 1919 |
+
"loss": 0.4269,
|
| 1920 |
+
"step": 273
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 1.6809815950920246,
|
| 1924 |
+
"grad_norm": 2.299816131591797,
|
| 1925 |
+
"learning_rate": 9.643076661610197e-06,
|
| 1926 |
+
"loss": 0.3925,
|
| 1927 |
+
"step": 274
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 1.687116564417178,
|
| 1931 |
+
"grad_norm": 2.202944040298462,
|
| 1932 |
+
"learning_rate": 9.571732026803978e-06,
|
| 1933 |
+
"loss": 0.3236,
|
| 1934 |
+
"step": 275
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 1.6932515337423313,
|
| 1938 |
+
"grad_norm": 2.154658794403076,
|
| 1939 |
+
"learning_rate": 9.500409224732863e-06,
|
| 1940 |
+
"loss": 0.3334,
|
| 1941 |
+
"step": 276
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 1.6993865030674846,
|
| 1945 |
+
"grad_norm": 1.9260175228118896,
|
| 1946 |
+
"learning_rate": 9.42911189137232e-06,
|
| 1947 |
+
"loss": 0.4292,
|
| 1948 |
+
"step": 277
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 1.705521472392638,
|
| 1952 |
+
"grad_norm": 1.929898738861084,
|
| 1953 |
+
"learning_rate": 9.357843661399447e-06,
|
| 1954 |
+
"loss": 0.4221,
|
| 1955 |
+
"step": 278
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 1.7116564417177913,
|
| 1959 |
+
"grad_norm": 2.124497413635254,
|
| 1960 |
+
"learning_rate": 9.286608168007678e-06,
|
| 1961 |
+
"loss": 0.4089,
|
| 1962 |
+
"step": 279
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 1.7177914110429446,
|
| 1966 |
+
"grad_norm": 2.5305910110473633,
|
| 1967 |
+
"learning_rate": 9.215409042721553e-06,
|
| 1968 |
+
"loss": 0.3828,
|
| 1969 |
+
"step": 280
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 1.7239263803680982,
|
| 1973 |
+
"grad_norm": 2.3219096660614014,
|
| 1974 |
+
"learning_rate": 9.144249915211605e-06,
|
| 1975 |
+
"loss": 0.4775,
|
| 1976 |
+
"step": 281
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 1.7300613496932515,
|
| 1980 |
+
"grad_norm": 2.0693113803863525,
|
| 1981 |
+
"learning_rate": 9.073134413109313e-06,
|
| 1982 |
+
"loss": 0.4324,
|
| 1983 |
+
"step": 282
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 1.7361963190184049,
|
| 1987 |
+
"grad_norm": 1.8101072311401367,
|
| 1988 |
+
"learning_rate": 9.002066161822174e-06,
|
| 1989 |
+
"loss": 0.3996,
|
| 1990 |
+
"step": 283
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 1.7423312883435584,
|
| 1994 |
+
"grad_norm": 2.0107815265655518,
|
| 1995 |
+
"learning_rate": 8.931048784348875e-06,
|
| 1996 |
+
"loss": 0.346,
|
| 1997 |
+
"step": 284
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 1.7484662576687118,
|
| 2001 |
+
"grad_norm": 2.2621984481811523,
|
| 2002 |
+
"learning_rate": 8.860085901094595e-06,
|
| 2003 |
+
"loss": 0.4883,
|
| 2004 |
+
"step": 285
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 1.7546012269938651,
|
| 2008 |
+
"grad_norm": 1.9184457063674927,
|
| 2009 |
+
"learning_rate": 8.789181129686452e-06,
|
| 2010 |
+
"loss": 0.4191,
|
| 2011 |
+
"step": 286
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 1.7607361963190185,
|
| 2015 |
+
"grad_norm": 2.13808536529541,
|
| 2016 |
+
"learning_rate": 8.718338084789074e-06,
|
| 2017 |
+
"loss": 0.3825,
|
| 2018 |
+
"step": 287
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 1.7668711656441718,
|
| 2022 |
+
"grad_norm": 2.075453281402588,
|
| 2023 |
+
"learning_rate": 8.647560377920311e-06,
|
| 2024 |
+
"loss": 0.4024,
|
| 2025 |
+
"step": 288
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.7730061349693251,
|
| 2029 |
+
"grad_norm": 2.0388762950897217,
|
| 2030 |
+
"learning_rate": 8.576851617267151e-06,
|
| 2031 |
+
"loss": 0.4048,
|
| 2032 |
+
"step": 289
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 1.7791411042944785,
|
| 2036 |
+
"grad_norm": 2.241987466812134,
|
| 2037 |
+
"learning_rate": 8.50621540750175e-06,
|
| 2038 |
+
"loss": 0.4071,
|
| 2039 |
+
"step": 290
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 1.7852760736196318,
|
| 2043 |
+
"grad_norm": 2.190722942352295,
|
| 2044 |
+
"learning_rate": 8.43565534959769e-06,
|
| 2045 |
+
"loss": 0.4283,
|
| 2046 |
+
"step": 291
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 1.7914110429447851,
|
| 2050 |
+
"grad_norm": 2.361862897872925,
|
| 2051 |
+
"learning_rate": 8.365175040646403e-06,
|
| 2052 |
+
"loss": 0.3501,
|
| 2053 |
+
"step": 292
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 1.7975460122699385,
|
| 2057 |
+
"grad_norm": 1.8603816032409668,
|
| 2058 |
+
"learning_rate": 8.294778073673762e-06,
|
| 2059 |
+
"loss": 0.4036,
|
| 2060 |
+
"step": 293
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 1.803680981595092,
|
| 2064 |
+
"grad_norm": 1.98984956741333,
|
| 2065 |
+
"learning_rate": 8.224468037456969e-06,
|
| 2066 |
+
"loss": 0.3805,
|
| 2067 |
+
"step": 294
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 1.8098159509202454,
|
| 2071 |
+
"grad_norm": 1.7755529880523682,
|
| 2072 |
+
"learning_rate": 8.154248516341547e-06,
|
| 2073 |
+
"loss": 0.3954,
|
| 2074 |
+
"step": 295
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 1.8159509202453987,
|
| 2078 |
+
"grad_norm": 2.173508405685425,
|
| 2079 |
+
"learning_rate": 8.084123090058646e-06,
|
| 2080 |
+
"loss": 0.3598,
|
| 2081 |
+
"step": 296
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.8220858895705523,
|
| 2085 |
+
"grad_norm": 2.2882938385009766,
|
| 2086 |
+
"learning_rate": 8.014095333542548e-06,
|
| 2087 |
+
"loss": 0.3793,
|
| 2088 |
+
"step": 297
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 1.8282208588957056,
|
| 2092 |
+
"grad_norm": 2.2722980976104736,
|
| 2093 |
+
"learning_rate": 7.944168816748396e-06,
|
| 2094 |
+
"loss": 0.3825,
|
| 2095 |
+
"step": 298
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 1.834355828220859,
|
| 2099 |
+
"grad_norm": 2.5786328315734863,
|
| 2100 |
+
"learning_rate": 7.874347104470234e-06,
|
| 2101 |
+
"loss": 0.4658,
|
| 2102 |
+
"step": 299
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 1.8404907975460123,
|
| 2106 |
+
"grad_norm": 2.3640172481536865,
|
| 2107 |
+
"learning_rate": 7.804633756159258e-06,
|
| 2108 |
+
"loss": 0.4318,
|
| 2109 |
+
"step": 300
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 1.8466257668711656,
|
| 2113 |
+
"grad_norm": 2.341289520263672,
|
| 2114 |
+
"learning_rate": 7.735032325742355e-06,
|
| 2115 |
+
"loss": 0.3372,
|
| 2116 |
+
"step": 301
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 1.852760736196319,
|
| 2120 |
+
"grad_norm": 2.271535873413086,
|
| 2121 |
+
"learning_rate": 7.66554636144095e-06,
|
| 2122 |
+
"loss": 0.3924,
|
| 2123 |
+
"step": 302
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.8588957055214723,
|
| 2127 |
+
"grad_norm": 2.16184139251709,
|
| 2128 |
+
"learning_rate": 7.596179405590076e-06,
|
| 2129 |
+
"loss": 0.4049,
|
| 2130 |
+
"step": 303
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 1.8650306748466257,
|
| 2134 |
+
"grad_norm": 1.9375874996185303,
|
| 2135 |
+
"learning_rate": 7.5269349944578454e-06,
|
| 2136 |
+
"loss": 0.3658,
|
| 2137 |
+
"step": 304
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 1.871165644171779,
|
| 2141 |
+
"grad_norm": 2.561995267868042,
|
| 2142 |
+
"learning_rate": 7.4578166580651335e-06,
|
| 2143 |
+
"loss": 0.4406,
|
| 2144 |
+
"step": 305
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 1.8773006134969326,
|
| 2148 |
+
"grad_norm": 2.128387689590454,
|
| 2149 |
+
"learning_rate": 7.388827920005628e-06,
|
| 2150 |
+
"loss": 0.3813,
|
| 2151 |
+
"step": 306
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 1.883435582822086,
|
| 2155 |
+
"grad_norm": 1.9198050498962402,
|
| 2156 |
+
"learning_rate": 7.319972297266215e-06,
|
| 2157 |
+
"loss": 0.3631,
|
| 2158 |
+
"step": 307
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 1.8895705521472392,
|
| 2162 |
+
"grad_norm": 2.7552289962768555,
|
| 2163 |
+
"learning_rate": 7.2512533000476625e-06,
|
| 2164 |
+
"loss": 0.3814,
|
| 2165 |
+
"step": 308
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.8957055214723928,
|
| 2169 |
+
"grad_norm": 2.614821672439575,
|
| 2170 |
+
"learning_rate": 7.182674431585703e-06,
|
| 2171 |
+
"loss": 0.4417,
|
| 2172 |
+
"step": 309
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 1.9018404907975461,
|
| 2176 |
+
"grad_norm": 1.8384186029434204,
|
| 2177 |
+
"learning_rate": 7.114239187972416e-06,
|
| 2178 |
+
"loss": 0.323,
|
| 2179 |
+
"step": 310
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 1.9079754601226995,
|
| 2183 |
+
"grad_norm": 1.896560549736023,
|
| 2184 |
+
"learning_rate": 7.045951057978001e-06,
|
| 2185 |
+
"loss": 0.3794,
|
| 2186 |
+
"step": 311
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 1.9141104294478528,
|
| 2190 |
+
"grad_norm": 2.288796901702881,
|
| 2191 |
+
"learning_rate": 6.977813522872943e-06,
|
| 2192 |
+
"loss": 0.4797,
|
| 2193 |
+
"step": 312
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.9202453987730062,
|
| 2197 |
+
"grad_norm": 2.0273969173431396,
|
| 2198 |
+
"learning_rate": 6.909830056250527e-06,
|
| 2199 |
+
"loss": 0.3846,
|
| 2200 |
+
"step": 313
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 1.9263803680981595,
|
| 2204 |
+
"grad_norm": 2.602076768875122,
|
| 2205 |
+
"learning_rate": 6.8420041238497525e-06,
|
| 2206 |
+
"loss": 0.4334,
|
| 2207 |
+
"step": 314
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 1.9325153374233128,
|
| 2211 |
+
"grad_norm": 2.0493180751800537,
|
| 2212 |
+
"learning_rate": 6.774339183378663e-06,
|
| 2213 |
+
"loss": 0.4496,
|
| 2214 |
+
"step": 315
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 1.9386503067484662,
|
| 2218 |
+
"grad_norm": 2.136305809020996,
|
| 2219 |
+
"learning_rate": 6.7068386843380695e-06,
|
| 2220 |
+
"loss": 0.4302,
|
| 2221 |
+
"step": 316
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.9447852760736195,
|
| 2225 |
+
"grad_norm": 2.9002561569213867,
|
| 2226 |
+
"learning_rate": 6.639506067845698e-06,
|
| 2227 |
+
"loss": 0.3764,
|
| 2228 |
+
"step": 317
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 1.9509202453987728,
|
| 2232 |
+
"grad_norm": 2.2220897674560547,
|
| 2233 |
+
"learning_rate": 6.572344766460776e-06,
|
| 2234 |
+
"loss": 0.3945,
|
| 2235 |
+
"step": 318
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 1.9570552147239264,
|
| 2239 |
+
"grad_norm": 1.7134284973144531,
|
| 2240 |
+
"learning_rate": 6.505358204009018e-06,
|
| 2241 |
+
"loss": 0.3844,
|
| 2242 |
+
"step": 319
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 1.9631901840490797,
|
| 2246 |
+
"grad_norm": 2.6499557495117188,
|
| 2247 |
+
"learning_rate": 6.438549795408107e-06,
|
| 2248 |
+
"loss": 0.4452,
|
| 2249 |
+
"step": 320
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 1.969325153374233,
|
| 2253 |
+
"grad_norm": 1.9206284284591675,
|
| 2254 |
+
"learning_rate": 6.3719229464935915e-06,
|
| 2255 |
+
"loss": 0.4319,
|
| 2256 |
+
"step": 321
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 1.9754601226993866,
|
| 2260 |
+
"grad_norm": 2.1525487899780273,
|
| 2261 |
+
"learning_rate": 6.305481053845262e-06,
|
| 2262 |
+
"loss": 0.3956,
|
| 2263 |
+
"step": 322
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.98159509202454,
|
| 2267 |
+
"grad_norm": 2.3156306743621826,
|
| 2268 |
+
"learning_rate": 6.239227504614004e-06,
|
| 2269 |
+
"loss": 0.4429,
|
| 2270 |
+
"step": 323
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 1.9877300613496933,
|
| 2274 |
+
"grad_norm": 2.249112129211426,
|
| 2275 |
+
"learning_rate": 6.173165676349103e-06,
|
| 2276 |
+
"loss": 0.4072,
|
| 2277 |
+
"step": 324
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 1.9938650306748467,
|
| 2281 |
+
"grad_norm": 2.2930808067321777,
|
| 2282 |
+
"learning_rate": 6.107298936826086e-06,
|
| 2283 |
+
"loss": 0.403,
|
| 2284 |
+
"step": 325
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 2.0,
|
| 2288 |
+
"grad_norm": 2.4880642890930176,
|
| 2289 |
+
"learning_rate": 6.041630643875018e-06,
|
| 2290 |
+
"loss": 0.3494,
|
| 2291 |
+
"step": 326
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 2.0061349693251533,
|
| 2295 |
+
"grad_norm": 1.532535195350647,
|
| 2296 |
+
"learning_rate": 5.9761641452093225e-06,
|
| 2297 |
+
"loss": 0.1558,
|
| 2298 |
+
"step": 327
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 2.0122699386503067,
|
| 2302 |
+
"grad_norm": 1.839203953742981,
|
| 2303 |
+
"learning_rate": 5.910902778255134e-06,
|
| 2304 |
+
"loss": 0.1526,
|
| 2305 |
+
"step": 328
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 2.01840490797546,
|
| 2309 |
+
"grad_norm": 2.041046142578125,
|
| 2310 |
+
"learning_rate": 5.845849869981137e-06,
|
| 2311 |
+
"loss": 0.2933,
|
| 2312 |
+
"step": 329
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 2.0245398773006134,
|
| 2316 |
+
"grad_norm": 1.4488948583602905,
|
| 2317 |
+
"learning_rate": 5.781008736728975e-06,
|
| 2318 |
+
"loss": 0.1941,
|
| 2319 |
+
"step": 330
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 2.0306748466257667,
|
| 2323 |
+
"grad_norm": 1.6453825235366821,
|
| 2324 |
+
"learning_rate": 5.716382684044191e-06,
|
| 2325 |
+
"loss": 0.1529,
|
| 2326 |
+
"step": 331
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 2.03680981595092,
|
| 2330 |
+
"grad_norm": 1.4784505367279053,
|
| 2331 |
+
"learning_rate": 5.6519750065076815e-06,
|
| 2332 |
+
"loss": 0.1819,
|
| 2333 |
+
"step": 332
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 2.042944785276074,
|
| 2337 |
+
"grad_norm": 1.4041457176208496,
|
| 2338 |
+
"learning_rate": 5.587788987567785e-06,
|
| 2339 |
+
"loss": 0.1751,
|
| 2340 |
+
"step": 333
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 2.049079754601227,
|
| 2344 |
+
"grad_norm": 1.6044899225234985,
|
| 2345 |
+
"learning_rate": 5.523827899372876e-06,
|
| 2346 |
+
"loss": 0.1705,
|
| 2347 |
+
"step": 334
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 2.0552147239263805,
|
| 2351 |
+
"grad_norm": 1.6499574184417725,
|
| 2352 |
+
"learning_rate": 5.460095002604533e-06,
|
| 2353 |
+
"loss": 0.1752,
|
| 2354 |
+
"step": 335
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 2.061349693251534,
|
| 2358 |
+
"grad_norm": 1.4871731996536255,
|
| 2359 |
+
"learning_rate": 5.396593546311346e-06,
|
| 2360 |
+
"loss": 0.1488,
|
| 2361 |
+
"step": 336
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 2.067484662576687,
|
| 2365 |
+
"grad_norm": 1.7872259616851807,
|
| 2366 |
+
"learning_rate": 5.333326767743263e-06,
|
| 2367 |
+
"loss": 0.142,
|
| 2368 |
+
"step": 337
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 2.0736196319018405,
|
| 2372 |
+
"grad_norm": 1.8322925567626953,
|
| 2373 |
+
"learning_rate": 5.270297892186563e-06,
|
| 2374 |
+
"loss": 0.1642,
|
| 2375 |
+
"step": 338
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 2.079754601226994,
|
| 2379 |
+
"grad_norm": 1.7445456981658936,
|
| 2380 |
+
"learning_rate": 5.207510132799436e-06,
|
| 2381 |
+
"loss": 0.1468,
|
| 2382 |
+
"step": 339
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 2.085889570552147,
|
| 2386 |
+
"grad_norm": 2.1466004848480225,
|
| 2387 |
+
"learning_rate": 5.144966690448159e-06,
|
| 2388 |
+
"loss": 0.1516,
|
| 2389 |
+
"step": 340
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 2.0920245398773005,
|
| 2393 |
+
"grad_norm": 2.1911582946777344,
|
| 2394 |
+
"learning_rate": 5.082670753543961e-06,
|
| 2395 |
+
"loss": 0.1763,
|
| 2396 |
+
"step": 341
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 2.098159509202454,
|
| 2400 |
+
"grad_norm": 1.8972338438034058,
|
| 2401 |
+
"learning_rate": 5.020625497880444e-06,
|
| 2402 |
+
"loss": 0.165,
|
| 2403 |
+
"step": 342
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 2.104294478527607,
|
| 2407 |
+
"grad_norm": 1.7345763444900513,
|
| 2408 |
+
"learning_rate": 4.958834086471683e-06,
|
| 2409 |
+
"loss": 0.162,
|
| 2410 |
+
"step": 343
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 2.1104294478527605,
|
| 2414 |
+
"grad_norm": 6.892634868621826,
|
| 2415 |
+
"learning_rate": 4.897299669391006e-06,
|
| 2416 |
+
"loss": 0.1596,
|
| 2417 |
+
"step": 344
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 2.116564417177914,
|
| 2421 |
+
"grad_norm": 1.7842110395431519,
|
| 2422 |
+
"learning_rate": 4.836025383610382e-06,
|
| 2423 |
+
"loss": 0.1289,
|
| 2424 |
+
"step": 345
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 2.1226993865030677,
|
| 2428 |
+
"grad_norm": 1.6559737920761108,
|
| 2429 |
+
"learning_rate": 4.775014352840512e-06,
|
| 2430 |
+
"loss": 0.1119,
|
| 2431 |
+
"step": 346
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 2.128834355828221,
|
| 2435 |
+
"grad_norm": 2.125093936920166,
|
| 2436 |
+
"learning_rate": 4.714269687371581e-06,
|
| 2437 |
+
"loss": 0.1375,
|
| 2438 |
+
"step": 347
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 2.1349693251533743,
|
| 2442 |
+
"grad_norm": 2.37017560005188,
|
| 2443 |
+
"learning_rate": 4.653794483914696e-06,
|
| 2444 |
+
"loss": 0.1666,
|
| 2445 |
+
"step": 348
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 2.1411042944785277,
|
| 2449 |
+
"grad_norm": 1.5750046968460083,
|
| 2450 |
+
"learning_rate": 4.593591825444028e-06,
|
| 2451 |
+
"loss": 0.1433,
|
| 2452 |
+
"step": 349
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 2.147239263803681,
|
| 2456 |
+
"grad_norm": 1.7769267559051514,
|
| 2457 |
+
"learning_rate": 4.533664781039622e-06,
|
| 2458 |
+
"loss": 0.1635,
|
| 2459 |
+
"step": 350
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 2.1533742331288344,
|
| 2463 |
+
"grad_norm": 1.677880883216858,
|
| 2464 |
+
"learning_rate": 4.474016405730973e-06,
|
| 2465 |
+
"loss": 0.1212,
|
| 2466 |
+
"step": 351
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 2.1595092024539877,
|
| 2470 |
+
"grad_norm": 2.1796493530273438,
|
| 2471 |
+
"learning_rate": 4.414649740341258e-06,
|
| 2472 |
+
"loss": 0.1555,
|
| 2473 |
+
"step": 352
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 2.165644171779141,
|
| 2477 |
+
"grad_norm": 1.721708059310913,
|
| 2478 |
+
"learning_rate": 4.355567811332311e-06,
|
| 2479 |
+
"loss": 0.1686,
|
| 2480 |
+
"step": 353
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 2.1717791411042944,
|
| 2484 |
+
"grad_norm": 3.571573257446289,
|
| 2485 |
+
"learning_rate": 4.296773630650358e-06,
|
| 2486 |
+
"loss": 0.1399,
|
| 2487 |
+
"step": 354
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 2.1779141104294477,
|
| 2491 |
+
"grad_norm": 1.773418664932251,
|
| 2492 |
+
"learning_rate": 4.2382701955724724e-06,
|
| 2493 |
+
"loss": 0.153,
|
| 2494 |
+
"step": 355
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 2.184049079754601,
|
| 2498 |
+
"grad_norm": 1.7861087322235107,
|
| 2499 |
+
"learning_rate": 4.180060488553743e-06,
|
| 2500 |
+
"loss": 0.1484,
|
| 2501 |
+
"step": 356
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 2.190184049079755,
|
| 2505 |
+
"grad_norm": 1.708827018737793,
|
| 2506 |
+
"learning_rate": 4.12214747707527e-06,
|
| 2507 |
+
"loss": 0.1596,
|
| 2508 |
+
"step": 357
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 2.196319018404908,
|
| 2512 |
+
"grad_norm": 1.4300062656402588,
|
| 2513 |
+
"learning_rate": 4.064534113492861e-06,
|
| 2514 |
+
"loss": 0.1319,
|
| 2515 |
+
"step": 358
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 2.2024539877300615,
|
| 2519 |
+
"grad_norm": 1.7988046407699585,
|
| 2520 |
+
"learning_rate": 4.007223334886531e-06,
|
| 2521 |
+
"loss": 0.1669,
|
| 2522 |
+
"step": 359
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 2.208588957055215,
|
| 2526 |
+
"grad_norm": 1.4519153833389282,
|
| 2527 |
+
"learning_rate": 3.950218062910776e-06,
|
| 2528 |
+
"loss": 0.1446,
|
| 2529 |
+
"step": 360
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 2.214723926380368,
|
| 2533 |
+
"grad_norm": 1.5383284091949463,
|
| 2534 |
+
"learning_rate": 3.893521203645618e-06,
|
| 2535 |
+
"loss": 0.1417,
|
| 2536 |
+
"step": 361
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 2.2208588957055215,
|
| 2540 |
+
"grad_norm": 1.837110161781311,
|
| 2541 |
+
"learning_rate": 3.837135647448475e-06,
|
| 2542 |
+
"loss": 0.1512,
|
| 2543 |
+
"step": 362
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 2.226993865030675,
|
| 2547 |
+
"grad_norm": 1.5508497953414917,
|
| 2548 |
+
"learning_rate": 3.78106426880678e-06,
|
| 2549 |
+
"loss": 0.1329,
|
| 2550 |
+
"step": 363
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 2.233128834355828,
|
| 2554 |
+
"grad_norm": 1.658305048942566,
|
| 2555 |
+
"learning_rate": 3.7253099261914794e-06,
|
| 2556 |
+
"loss": 0.1289,
|
| 2557 |
+
"step": 364
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 2.2392638036809815,
|
| 2561 |
+
"grad_norm": 1.5167778730392456,
|
| 2562 |
+
"learning_rate": 3.6698754619112974e-06,
|
| 2563 |
+
"loss": 0.121,
|
| 2564 |
+
"step": 365
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 2.245398773006135,
|
| 2568 |
+
"grad_norm": 1.7372796535491943,
|
| 2569 |
+
"learning_rate": 3.6147637019678195e-06,
|
| 2570 |
+
"loss": 0.1338,
|
| 2571 |
+
"step": 366
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 2.2515337423312882,
|
| 2575 |
+
"grad_norm": 1.4506868124008179,
|
| 2576 |
+
"learning_rate": 3.5599774559114475e-06,
|
| 2577 |
+
"loss": 0.141,
|
| 2578 |
+
"step": 367
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 2.2576687116564416,
|
| 2582 |
+
"grad_norm": 1.7319529056549072,
|
| 2583 |
+
"learning_rate": 3.505519516698165e-06,
|
| 2584 |
+
"loss": 0.1283,
|
| 2585 |
+
"step": 368
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 2.263803680981595,
|
| 2589 |
+
"grad_norm": 1.6819450855255127,
|
| 2590 |
+
"learning_rate": 3.4513926605471504e-06,
|
| 2591 |
+
"loss": 0.1635,
|
| 2592 |
+
"step": 369
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 2.2699386503067487,
|
| 2596 |
+
"grad_norm": 1.5419731140136719,
|
| 2597 |
+
"learning_rate": 3.3975996467992557e-06,
|
| 2598 |
+
"loss": 0.1515,
|
| 2599 |
+
"step": 370
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 2.276073619631902,
|
| 2603 |
+
"grad_norm": 1.622673749923706,
|
| 2604 |
+
"learning_rate": 3.344143217776319e-06,
|
| 2605 |
+
"loss": 0.1432,
|
| 2606 |
+
"step": 371
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 2.2822085889570554,
|
| 2610 |
+
"grad_norm": 1.9862250089645386,
|
| 2611 |
+
"learning_rate": 3.291026098641398e-06,
|
| 2612 |
+
"loss": 0.1551,
|
| 2613 |
+
"step": 372
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 2.2883435582822087,
|
| 2617 |
+
"grad_norm": 1.975870132446289,
|
| 2618 |
+
"learning_rate": 3.2382509972598087e-06,
|
| 2619 |
+
"loss": 0.1256,
|
| 2620 |
+
"step": 373
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 2.294478527607362,
|
| 2624 |
+
"grad_norm": 1.3615440130233765,
|
| 2625 |
+
"learning_rate": 3.1858206040610883e-06,
|
| 2626 |
+
"loss": 0.1114,
|
| 2627 |
+
"step": 374
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 2.3006134969325154,
|
| 2631 |
+
"grad_norm": 2.0243375301361084,
|
| 2632 |
+
"learning_rate": 3.133737591901864e-06,
|
| 2633 |
+
"loss": 0.1477,
|
| 2634 |
+
"step": 375
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 2.3067484662576687,
|
| 2638 |
+
"grad_norm": 2.148789644241333,
|
| 2639 |
+
"learning_rate": 3.0820046159295647e-06,
|
| 2640 |
+
"loss": 0.1828,
|
| 2641 |
+
"step": 376
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 2.312883435582822,
|
| 2645 |
+
"grad_norm": 1.5784991979599,
|
| 2646 |
+
"learning_rate": 3.0306243134470668e-06,
|
| 2647 |
+
"loss": 0.1466,
|
| 2648 |
+
"step": 377
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 2.3190184049079754,
|
| 2652 |
+
"grad_norm": 1.8044228553771973,
|
| 2653 |
+
"learning_rate": 2.9795993037782657e-06,
|
| 2654 |
+
"loss": 0.1136,
|
| 2655 |
+
"step": 378
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 2.3251533742331287,
|
| 2659 |
+
"grad_norm": 1.5507012605667114,
|
| 2660 |
+
"learning_rate": 2.9289321881345257e-06,
|
| 2661 |
+
"loss": 0.1112,
|
| 2662 |
+
"step": 379
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 2.331288343558282,
|
| 2666 |
+
"grad_norm": 1.4628651142120361,
|
| 2667 |
+
"learning_rate": 2.878625549482084e-06,
|
| 2668 |
+
"loss": 0.1335,
|
| 2669 |
+
"step": 380
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 2.3374233128834354,
|
| 2673 |
+
"grad_norm": 2.275484561920166,
|
| 2674 |
+
"learning_rate": 2.8286819524103657e-06,
|
| 2675 |
+
"loss": 0.1987,
|
| 2676 |
+
"step": 381
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 2.3435582822085887,
|
| 2680 |
+
"grad_norm": 1.8239965438842773,
|
| 2681 |
+
"learning_rate": 2.779103943001248e-06,
|
| 2682 |
+
"loss": 0.1327,
|
| 2683 |
+
"step": 382
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 2.3496932515337425,
|
| 2687 |
+
"grad_norm": 1.621533989906311,
|
| 2688 |
+
"learning_rate": 2.7298940486992654e-06,
|
| 2689 |
+
"loss": 0.1225,
|
| 2690 |
+
"step": 383
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 2.355828220858896,
|
| 2694 |
+
"grad_norm": 1.5327485799789429,
|
| 2695 |
+
"learning_rate": 2.681054778182748e-06,
|
| 2696 |
+
"loss": 0.1254,
|
| 2697 |
+
"step": 384
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 2.361963190184049,
|
| 2701 |
+
"grad_norm": 2.2347185611724854,
|
| 2702 |
+
"learning_rate": 2.6325886212359496e-06,
|
| 2703 |
+
"loss": 0.1375,
|
| 2704 |
+
"step": 385
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 2.3680981595092025,
|
| 2708 |
+
"grad_norm": 1.4357990026474,
|
| 2709 |
+
"learning_rate": 2.5844980486221225e-06,
|
| 2710 |
+
"loss": 0.1208,
|
| 2711 |
+
"step": 386
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 2.374233128834356,
|
| 2715 |
+
"grad_norm": 1.740544319152832,
|
| 2716 |
+
"learning_rate": 2.5367855119575314e-06,
|
| 2717 |
+
"loss": 0.1392,
|
| 2718 |
+
"step": 387
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 2.3803680981595092,
|
| 2722 |
+
"grad_norm": 2.388603925704956,
|
| 2723 |
+
"learning_rate": 2.4894534435865015e-06,
|
| 2724 |
+
"loss": 0.1776,
|
| 2725 |
+
"step": 388
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 2.3865030674846626,
|
| 2729 |
+
"grad_norm": 1.74617600440979,
|
| 2730 |
+
"learning_rate": 2.4425042564574186e-06,
|
| 2731 |
+
"loss": 0.1345,
|
| 2732 |
+
"step": 389
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 2.392638036809816,
|
| 2736 |
+
"grad_norm": 1.3598535060882568,
|
| 2737 |
+
"learning_rate": 2.395940343999691e-06,
|
| 2738 |
+
"loss": 0.1174,
|
| 2739 |
+
"step": 390
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 2.3987730061349692,
|
| 2743 |
+
"grad_norm": 1.9664957523345947,
|
| 2744 |
+
"learning_rate": 2.3497640800017687e-06,
|
| 2745 |
+
"loss": 0.1528,
|
| 2746 |
+
"step": 391
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 2.4049079754601226,
|
| 2750 |
+
"grad_norm": 1.5915489196777344,
|
| 2751 |
+
"learning_rate": 2.3039778184901086e-06,
|
| 2752 |
+
"loss": 0.1424,
|
| 2753 |
+
"step": 392
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 2.411042944785276,
|
| 2757 |
+
"grad_norm": 1.9922267198562622,
|
| 2758 |
+
"learning_rate": 2.2585838936091753e-06,
|
| 2759 |
+
"loss": 0.1728,
|
| 2760 |
+
"step": 393
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 2.4171779141104293,
|
| 2764 |
+
"grad_norm": 1.340556025505066,
|
| 2765 |
+
"learning_rate": 2.213584619502451e-06,
|
| 2766 |
+
"loss": 0.1117,
|
| 2767 |
+
"step": 394
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 2.4233128834355826,
|
| 2771 |
+
"grad_norm": 1.5564244985580444,
|
| 2772 |
+
"learning_rate": 2.1689822901944456e-06,
|
| 2773 |
+
"loss": 0.1568,
|
| 2774 |
+
"step": 395
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 2.4294478527607364,
|
| 2778 |
+
"grad_norm": 2.1123807430267334,
|
| 2779 |
+
"learning_rate": 2.1247791794737827e-06,
|
| 2780 |
+
"loss": 0.1414,
|
| 2781 |
+
"step": 396
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 2.4355828220858897,
|
| 2785 |
+
"grad_norm": 2.1588234901428223,
|
| 2786 |
+
"learning_rate": 2.0809775407772505e-06,
|
| 2787 |
+
"loss": 0.1897,
|
| 2788 |
+
"step": 397
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 2.441717791411043,
|
| 2792 |
+
"grad_norm": 2.129268169403076,
|
| 2793 |
+
"learning_rate": 2.0375796070749366e-06,
|
| 2794 |
+
"loss": 0.1569,
|
| 2795 |
+
"step": 398
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 2.4478527607361964,
|
| 2799 |
+
"grad_norm": 1.6751271486282349,
|
| 2800 |
+
"learning_rate": 1.994587590756397e-06,
|
| 2801 |
+
"loss": 0.1418,
|
| 2802 |
+
"step": 399
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 2.4539877300613497,
|
| 2806 |
+
"grad_norm": 1.5498850345611572,
|
| 2807 |
+
"learning_rate": 1.9520036835178667e-06,
|
| 2808 |
+
"loss": 0.142,
|
| 2809 |
+
"step": 400
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 2.460122699386503,
|
| 2813 |
+
"grad_norm": 1.716620683670044,
|
| 2814 |
+
"learning_rate": 1.9098300562505266e-06,
|
| 2815 |
+
"loss": 0.1502,
|
| 2816 |
+
"step": 401
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 2.4662576687116564,
|
| 2820 |
+
"grad_norm": 2.2099783420562744,
|
| 2821 |
+
"learning_rate": 1.8680688589298368e-06,
|
| 2822 |
+
"loss": 0.1714,
|
| 2823 |
+
"step": 402
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 2.4723926380368098,
|
| 2827 |
+
"grad_norm": 2.4377899169921875,
|
| 2828 |
+
"learning_rate": 1.826722220505931e-06,
|
| 2829 |
+
"loss": 0.188,
|
| 2830 |
+
"step": 403
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 2.478527607361963,
|
| 2834 |
+
"grad_norm": 1.4439836740493774,
|
| 2835 |
+
"learning_rate": 1.7857922487950873e-06,
|
| 2836 |
+
"loss": 0.1296,
|
| 2837 |
+
"step": 404
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 2.4846625766871164,
|
| 2841 |
+
"grad_norm": 1.636144757270813,
|
| 2842 |
+
"learning_rate": 1.74528103037226e-06,
|
| 2843 |
+
"loss": 0.1608,
|
| 2844 |
+
"step": 405
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 2.4907975460122698,
|
| 2848 |
+
"grad_norm": 1.9918489456176758,
|
| 2849 |
+
"learning_rate": 1.705190630464737e-06,
|
| 2850 |
+
"loss": 0.1859,
|
| 2851 |
+
"step": 406
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 2.4969325153374236,
|
| 2855 |
+
"grad_norm": 2.012042284011841,
|
| 2856 |
+
"learning_rate": 1.6655230928468257e-06,
|
| 2857 |
+
"loss": 0.1484,
|
| 2858 |
+
"step": 407
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 2.5030674846625764,
|
| 2862 |
+
"grad_norm": 1.5652939081192017,
|
| 2863 |
+
"learning_rate": 1.6262804397356747e-06,
|
| 2864 |
+
"loss": 0.1425,
|
| 2865 |
+
"step": 408
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 2.5092024539877302,
|
| 2869 |
+
"grad_norm": 2.306347370147705,
|
| 2870 |
+
"learning_rate": 1.587464671688187e-06,
|
| 2871 |
+
"loss": 0.1676,
|
| 2872 |
+
"step": 409
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 2.5153374233128836,
|
| 2876 |
+
"grad_norm": 1.5338711738586426,
|
| 2877 |
+
"learning_rate": 1.5490777674990376e-06,
|
| 2878 |
+
"loss": 0.1462,
|
| 2879 |
+
"step": 410
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 2.521472392638037,
|
| 2883 |
+
"grad_norm": 1.6177219152450562,
|
| 2884 |
+
"learning_rate": 1.5111216840997745e-06,
|
| 2885 |
+
"loss": 0.1046,
|
| 2886 |
+
"step": 411
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 2.5276073619631902,
|
| 2890 |
+
"grad_norm": 1.720763087272644,
|
| 2891 |
+
"learning_rate": 1.4735983564590784e-06,
|
| 2892 |
+
"loss": 0.1485,
|
| 2893 |
+
"step": 412
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 2.5337423312883436,
|
| 2897 |
+
"grad_norm": 1.984955072402954,
|
| 2898 |
+
"learning_rate": 1.436509697484111e-06,
|
| 2899 |
+
"loss": 0.1585,
|
| 2900 |
+
"step": 413
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 2.539877300613497,
|
| 2904 |
+
"grad_norm": 2.275015354156494,
|
| 2905 |
+
"learning_rate": 1.3998575979229944e-06,
|
| 2906 |
+
"loss": 0.1793,
|
| 2907 |
+
"step": 414
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 2.5460122699386503,
|
| 2911 |
+
"grad_norm": 1.4398918151855469,
|
| 2912 |
+
"learning_rate": 1.3636439262684299e-06,
|
| 2913 |
+
"loss": 0.1418,
|
| 2914 |
+
"step": 415
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 2.5521472392638036,
|
| 2918 |
+
"grad_norm": 1.6065205335617065,
|
| 2919 |
+
"learning_rate": 1.3278705286624328e-06,
|
| 2920 |
+
"loss": 0.1549,
|
| 2921 |
+
"step": 416
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 2.558282208588957,
|
| 2925 |
+
"grad_norm": 1.4881666898727417,
|
| 2926 |
+
"learning_rate": 1.2925392288022299e-06,
|
| 2927 |
+
"loss": 0.1432,
|
| 2928 |
+
"step": 417
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 2.5644171779141103,
|
| 2932 |
+
"grad_norm": 2.278841972351074,
|
| 2933 |
+
"learning_rate": 1.257651827847276e-06,
|
| 2934 |
+
"loss": 0.1348,
|
| 2935 |
+
"step": 418
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 2.5705521472392636,
|
| 2939 |
+
"grad_norm": 2.0989537239074707,
|
| 2940 |
+
"learning_rate": 1.2232101043274437e-06,
|
| 2941 |
+
"loss": 0.1687,
|
| 2942 |
+
"step": 419
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 2.5766871165644174,
|
| 2946 |
+
"grad_norm": 1.488337516784668,
|
| 2947 |
+
"learning_rate": 1.1892158140523546e-06,
|
| 2948 |
+
"loss": 0.1263,
|
| 2949 |
+
"step": 420
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 2.5828220858895703,
|
| 2953 |
+
"grad_norm": 1.9117529392242432,
|
| 2954 |
+
"learning_rate": 1.1556706900218572e-06,
|
| 2955 |
+
"loss": 0.1273,
|
| 2956 |
+
"step": 421
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 2.588957055214724,
|
| 2960 |
+
"grad_norm": 2.0531487464904785,
|
| 2961 |
+
"learning_rate": 1.122576442337696e-06,
|
| 2962 |
+
"loss": 0.1348,
|
| 2963 |
+
"step": 422
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 2.5950920245398774,
|
| 2967 |
+
"grad_norm": 1.714740514755249,
|
| 2968 |
+
"learning_rate": 1.0899347581163222e-06,
|
| 2969 |
+
"loss": 0.1276,
|
| 2970 |
+
"step": 423
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 2.6012269938650308,
|
| 2974 |
+
"grad_norm": 2.042924642562866,
|
| 2975 |
+
"learning_rate": 1.0577473014028872e-06,
|
| 2976 |
+
"loss": 0.1325,
|
| 2977 |
+
"step": 424
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 2.607361963190184,
|
| 2981 |
+
"grad_norm": 1.9139305353164673,
|
| 2982 |
+
"learning_rate": 1.0260157130864178e-06,
|
| 2983 |
+
"loss": 0.1278,
|
| 2984 |
+
"step": 425
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 2.6134969325153374,
|
| 2988 |
+
"grad_norm": 1.556216835975647,
|
| 2989 |
+
"learning_rate": 9.9474161081615e-07,
|
| 2990 |
+
"loss": 0.1428,
|
| 2991 |
+
"step": 426
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 2.6196319018404908,
|
| 2995 |
+
"grad_norm": 1.4173800945281982,
|
| 2996 |
+
"learning_rate": 9.63926588919083e-07,
|
| 2997 |
+
"loss": 0.1339,
|
| 2998 |
+
"step": 427
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 2.625766871165644,
|
| 3002 |
+
"grad_norm": 1.4731110334396362,
|
| 3003 |
+
"learning_rate": 9.335722183186868e-07,
|
| 3004 |
+
"loss": 0.147,
|
| 3005 |
+
"step": 428
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 2.6319018404907975,
|
| 3009 |
+
"grad_norm": 1.5274585485458374,
|
| 3010 |
+
"learning_rate": 9.036800464548157e-07,
|
| 3011 |
+
"loss": 0.1275,
|
| 3012 |
+
"step": 429
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 2.638036809815951,
|
| 3016 |
+
"grad_norm": 1.4603732824325562,
|
| 3017 |
+
"learning_rate": 8.742515972048404e-07,
|
| 3018 |
+
"loss": 0.1355,
|
| 3019 |
+
"step": 430
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 2.644171779141104,
|
| 3023 |
+
"grad_norm": 1.878303050994873,
|
| 3024 |
+
"learning_rate": 8.4528837080594e-07,
|
| 3025 |
+
"loss": 0.1383,
|
| 3026 |
+
"step": 431
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 2.6503067484662575,
|
| 3030 |
+
"grad_norm": 1.9829754829406738,
|
| 3031 |
+
"learning_rate": 8.167918437786316e-07,
|
| 3032 |
+
"loss": 0.1342,
|
| 3033 |
+
"step": 432
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 2.6564417177914113,
|
| 3037 |
+
"grad_norm": 2.8250081539154053,
|
| 3038 |
+
"learning_rate": 7.887634688515e-07,
|
| 3039 |
+
"loss": 0.1975,
|
| 3040 |
+
"step": 433
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 2.662576687116564,
|
| 3044 |
+
"grad_norm": 1.4426681995391846,
|
| 3045 |
+
"learning_rate": 7.612046748871327e-07,
|
| 3046 |
+
"loss": 0.1164,
|
| 3047 |
+
"step": 434
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 2.668711656441718,
|
| 3051 |
+
"grad_norm": 2.128920793533325,
|
| 3052 |
+
"learning_rate": 7.341168668092857e-07,
|
| 3053 |
+
"loss": 0.1311,
|
| 3054 |
+
"step": 435
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 2.6748466257668713,
|
| 3058 |
+
"grad_norm": 1.6792587041854858,
|
| 3059 |
+
"learning_rate": 7.07501425531254e-07,
|
| 3060 |
+
"loss": 0.1636,
|
| 3061 |
+
"step": 436
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 2.6809815950920246,
|
| 3065 |
+
"grad_norm": 2.3107171058654785,
|
| 3066 |
+
"learning_rate": 6.813597078854772e-07,
|
| 3067 |
+
"loss": 0.1661,
|
| 3068 |
+
"step": 437
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 2.687116564417178,
|
| 3072 |
+
"grad_norm": 1.7251126766204834,
|
| 3073 |
+
"learning_rate": 6.556930465543709e-07,
|
| 3074 |
+
"loss": 0.1407,
|
| 3075 |
+
"step": 438
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 2.6932515337423313,
|
| 3079 |
+
"grad_norm": 1.7014403343200684,
|
| 3080 |
+
"learning_rate": 6.305027500023841e-07,
|
| 3081 |
+
"loss": 0.1208,
|
| 3082 |
+
"step": 439
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 2.6993865030674846,
|
| 3086 |
+
"grad_norm": 1.8357864618301392,
|
| 3087 |
+
"learning_rate": 6.057901024092949e-07,
|
| 3088 |
+
"loss": 0.129,
|
| 3089 |
+
"step": 440
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 2.705521472392638,
|
| 3093 |
+
"grad_norm": 1.6881440877914429,
|
| 3094 |
+
"learning_rate": 5.815563636047539e-07,
|
| 3095 |
+
"loss": 0.1125,
|
| 3096 |
+
"step": 441
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 2.7116564417177913,
|
| 3100 |
+
"grad_norm": 1.9784103631973267,
|
| 3101 |
+
"learning_rate": 5.578027690040411e-07,
|
| 3102 |
+
"loss": 0.152,
|
| 3103 |
+
"step": 442
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 2.7177914110429446,
|
| 3107 |
+
"grad_norm": 1.8745219707489014,
|
| 3108 |
+
"learning_rate": 5.345305295450997e-07,
|
| 3109 |
+
"loss": 0.1385,
|
| 3110 |
+
"step": 443
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 2.7239263803680984,
|
| 3114 |
+
"grad_norm": 1.5030583143234253,
|
| 3115 |
+
"learning_rate": 5.117408316268047e-07,
|
| 3116 |
+
"loss": 0.128,
|
| 3117 |
+
"step": 444
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 2.7300613496932513,
|
| 3121 |
+
"grad_norm": 1.5685703754425049,
|
| 3122 |
+
"learning_rate": 4.894348370484648e-07,
|
| 3123 |
+
"loss": 0.1518,
|
| 3124 |
+
"step": 445
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 2.736196319018405,
|
| 3128 |
+
"grad_norm": 2.1138014793395996,
|
| 3129 |
+
"learning_rate": 4.67613682950615e-07,
|
| 3130 |
+
"loss": 0.1616,
|
| 3131 |
+
"step": 446
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 2.7423312883435584,
|
| 3135 |
+
"grad_norm": 1.76698637008667,
|
| 3136 |
+
"learning_rate": 4.4627848175703315e-07,
|
| 3137 |
+
"loss": 0.1379,
|
| 3138 |
+
"step": 447
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 2.7484662576687118,
|
| 3142 |
+
"grad_norm": 1.3497811555862427,
|
| 3143 |
+
"learning_rate": 4.254303211180355e-07,
|
| 3144 |
+
"loss": 0.1185,
|
| 3145 |
+
"step": 448
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 2.754601226993865,
|
| 3149 |
+
"grad_norm": 1.360832691192627,
|
| 3150 |
+
"learning_rate": 4.0507026385502747e-07,
|
| 3151 |
+
"loss": 0.1039,
|
| 3152 |
+
"step": 449
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 2.7607361963190185,
|
| 3156 |
+
"grad_norm": 1.7732263803482056,
|
| 3157 |
+
"learning_rate": 3.851993479063154e-07,
|
| 3158 |
+
"loss": 0.1477,
|
| 3159 |
+
"step": 450
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 2.766871165644172,
|
| 3163 |
+
"grad_norm": 1.3179359436035156,
|
| 3164 |
+
"learning_rate": 3.658185862742103e-07,
|
| 3165 |
+
"loss": 0.1136,
|
| 3166 |
+
"step": 451
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 2.773006134969325,
|
| 3170 |
+
"grad_norm": 1.9042986631393433,
|
| 3171 |
+
"learning_rate": 3.4692896697336887e-07,
|
| 3172 |
+
"loss": 0.1453,
|
| 3173 |
+
"step": 452
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 2.7791411042944785,
|
| 3177 |
+
"grad_norm": 1.8464301824569702,
|
| 3178 |
+
"learning_rate": 3.2853145298042954e-07,
|
| 3179 |
+
"loss": 0.1613,
|
| 3180 |
+
"step": 453
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 2.785276073619632,
|
| 3184 |
+
"grad_norm": 1.6313396692276,
|
| 3185 |
+
"learning_rate": 3.106269821849273e-07,
|
| 3186 |
+
"loss": 0.1627,
|
| 3187 |
+
"step": 454
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 2.791411042944785,
|
| 3191 |
+
"grad_norm": 1.4320224523544312,
|
| 3192 |
+
"learning_rate": 2.93216467341475e-07,
|
| 3193 |
+
"loss": 0.1081,
|
| 3194 |
+
"step": 455
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 2.7975460122699385,
|
| 3198 |
+
"grad_norm": 2.1733744144439697,
|
| 3199 |
+
"learning_rate": 2.7630079602323447e-07,
|
| 3200 |
+
"loss": 0.1579,
|
| 3201 |
+
"step": 456
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 2.8036809815950923,
|
| 3205 |
+
"grad_norm": 1.6012330055236816,
|
| 3206 |
+
"learning_rate": 2.5988083057666534e-07,
|
| 3207 |
+
"loss": 0.1521,
|
| 3208 |
+
"step": 457
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 2.809815950920245,
|
| 3212 |
+
"grad_norm": 1.710110068321228,
|
| 3213 |
+
"learning_rate": 2.439574080775675e-07,
|
| 3214 |
+
"loss": 0.1387,
|
| 3215 |
+
"step": 458
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 2.815950920245399,
|
| 3219 |
+
"grad_norm": 1.8426260948181152,
|
| 3220 |
+
"learning_rate": 2.2853134028840594e-07,
|
| 3221 |
+
"loss": 0.1462,
|
| 3222 |
+
"step": 459
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 2.8220858895705523,
|
| 3226 |
+
"grad_norm": 1.8103870153427124,
|
| 3227 |
+
"learning_rate": 2.1360341361692517e-07,
|
| 3228 |
+
"loss": 0.1337,
|
| 3229 |
+
"step": 460
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 2.8282208588957056,
|
| 3233 |
+
"grad_norm": 2.107290029525757,
|
| 3234 |
+
"learning_rate": 1.9917438907606556e-07,
|
| 3235 |
+
"loss": 0.1504,
|
| 3236 |
+
"step": 461
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 2.834355828220859,
|
| 3240 |
+
"grad_norm": 1.6594619750976562,
|
| 3241 |
+
"learning_rate": 1.852450022451624e-07,
|
| 3242 |
+
"loss": 0.1388,
|
| 3243 |
+
"step": 462
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 2.8404907975460123,
|
| 3247 |
+
"grad_norm": 2.51185941696167,
|
| 3248 |
+
"learning_rate": 1.7181596323244453e-07,
|
| 3249 |
+
"loss": 0.184,
|
| 3250 |
+
"step": 463
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 2.8466257668711656,
|
| 3254 |
+
"grad_norm": 1.7496821880340576,
|
| 3255 |
+
"learning_rate": 1.5888795663883904e-07,
|
| 3256 |
+
"loss": 0.1038,
|
| 3257 |
+
"step": 464
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 2.852760736196319,
|
| 3261 |
+
"grad_norm": 1.680633306503296,
|
| 3262 |
+
"learning_rate": 1.464616415230702e-07,
|
| 3263 |
+
"loss": 0.1167,
|
| 3264 |
+
"step": 465
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 2.8588957055214723,
|
| 3268 |
+
"grad_norm": 2.0072343349456787,
|
| 3269 |
+
"learning_rate": 1.345376513680574e-07,
|
| 3270 |
+
"loss": 0.1789,
|
| 3271 |
+
"step": 466
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 2.8650306748466257,
|
| 3275 |
+
"grad_norm": 1.5459482669830322,
|
| 3276 |
+
"learning_rate": 1.231165940486234e-07,
|
| 3277 |
+
"loss": 0.1357,
|
| 3278 |
+
"step": 467
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 2.871165644171779,
|
| 3282 |
+
"grad_norm": 1.8858439922332764,
|
| 3283 |
+
"learning_rate": 1.121990518005045e-07,
|
| 3284 |
+
"loss": 0.1457,
|
| 3285 |
+
"step": 468
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 2.8773006134969323,
|
| 3289 |
+
"grad_norm": 1.7070627212524414,
|
| 3290 |
+
"learning_rate": 1.0178558119067316e-07,
|
| 3291 |
+
"loss": 0.1466,
|
| 3292 |
+
"step": 469
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 2.883435582822086,
|
| 3296 |
+
"grad_norm": 1.4789223670959473,
|
| 3297 |
+
"learning_rate": 9.187671308895418e-08,
|
| 3298 |
+
"loss": 0.1072,
|
| 3299 |
+
"step": 470
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 2.889570552147239,
|
| 3303 |
+
"grad_norm": 1.750195026397705,
|
| 3304 |
+
"learning_rate": 8.247295264097288e-08,
|
| 3305 |
+
"loss": 0.181,
|
| 3306 |
+
"step": 471
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 2.895705521472393,
|
| 3310 |
+
"grad_norm": 2.0875391960144043,
|
| 3311 |
+
"learning_rate": 7.357477924239797e-08,
|
| 3312 |
+
"loss": 0.1121,
|
| 3313 |
+
"step": 472
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 2.901840490797546,
|
| 3317 |
+
"grad_norm": 1.7946627140045166,
|
| 3318 |
+
"learning_rate": 6.51826465144978e-08,
|
| 3319 |
+
"loss": 0.1424,
|
| 3320 |
+
"step": 473
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 2.9079754601226995,
|
| 3324 |
+
"grad_norm": 1.535167932510376,
|
| 3325 |
+
"learning_rate": 5.7296982281026534e-08,
|
| 3326 |
+
"loss": 0.1268,
|
| 3327 |
+
"step": 474
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 2.914110429447853,
|
| 3331 |
+
"grad_norm": 1.6847984790802002,
|
| 3332 |
+
"learning_rate": 4.991818854640396e-08,
|
| 3333 |
+
"loss": 0.1522,
|
| 3334 |
+
"step": 475
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 2.920245398773006,
|
| 3338 |
+
"grad_norm": 2.0417425632476807,
|
| 3339 |
+
"learning_rate": 4.304664147522619e-08,
|
| 3340 |
+
"loss": 0.1904,
|
| 3341 |
+
"step": 476
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 2.9263803680981595,
|
| 3345 |
+
"grad_norm": 2.0232009887695312,
|
| 3346 |
+
"learning_rate": 3.668269137308666e-08,
|
| 3347 |
+
"loss": 0.1306,
|
| 3348 |
+
"step": 477
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 2.932515337423313,
|
| 3352 |
+
"grad_norm": 2.1321542263031006,
|
| 3353 |
+
"learning_rate": 3.082666266872036e-08,
|
| 3354 |
+
"loss": 0.1814,
|
| 3355 |
+
"step": 478
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 2.938650306748466,
|
| 3359 |
+
"grad_norm": 1.9118684530258179,
|
| 3360 |
+
"learning_rate": 2.547885389746485e-08,
|
| 3361 |
+
"loss": 0.1534,
|
| 3362 |
+
"step": 479
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 2.9447852760736195,
|
| 3366 |
+
"grad_norm": 1.84584379196167,
|
| 3367 |
+
"learning_rate": 2.063953768603799e-08,
|
| 3368 |
+
"loss": 0.1425,
|
| 3369 |
+
"step": 480
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 2.950920245398773,
|
| 3373 |
+
"grad_norm": 1.933309555053711,
|
| 3374 |
+
"learning_rate": 1.630896073864352e-08,
|
| 3375 |
+
"loss": 0.1726,
|
| 3376 |
+
"step": 481
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 2.957055214723926,
|
| 3380 |
+
"grad_norm": 2.155355930328369,
|
| 3381 |
+
"learning_rate": 1.2487343824389986e-08,
|
| 3382 |
+
"loss": 0.1706,
|
| 3383 |
+
"step": 482
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 2.96319018404908,
|
| 3387 |
+
"grad_norm": 1.5894520282745361,
|
| 3388 |
+
"learning_rate": 9.174881766043086e-09,
|
| 3389 |
+
"loss": 0.1207,
|
| 3390 |
+
"step": 483
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 2.969325153374233,
|
| 3394 |
+
"grad_norm": 1.6635428667068481,
|
| 3395 |
+
"learning_rate": 6.371743430082511e-09,
|
| 3396 |
+
"loss": 0.1287,
|
| 3397 |
+
"step": 484
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 2.9754601226993866,
|
| 3401 |
+
"grad_norm": 1.982111930847168,
|
| 3402 |
+
"learning_rate": 4.0780717181077015e-09,
|
| 3403 |
+
"loss": 0.1455,
|
| 3404 |
+
"step": 485
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 2.98159509202454,
|
| 3408 |
+
"grad_norm": 1.4887462854385376,
|
| 3409 |
+
"learning_rate": 2.2939835595392477e-09,
|
| 3410 |
+
"loss": 0.1335,
|
| 3411 |
+
"step": 486
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 2.9877300613496933,
|
| 3415 |
+
"grad_norm": 2.119296073913574,
|
| 3416 |
+
"learning_rate": 1.019569905666984e-09,
|
| 3417 |
+
"loss": 0.1516,
|
| 3418 |
+
"step": 487
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 2.9938650306748467,
|
| 3422 |
+
"grad_norm": 1.4451674222946167,
|
| 3423 |
+
"learning_rate": 2.5489572500814184e-10,
|
| 3424 |
+
"loss": 0.1292,
|
| 3425 |
+
"step": 488
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 3.0,
|
| 3429 |
+
"grad_norm": 1.6885939836502075,
|
| 3430 |
+
"learning_rate": 0.0,
|
| 3431 |
+
"loss": 0.1343,
|
| 3432 |
+
"step": 489
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 3.0,
|
| 3436 |
+
"step": 489,
|
| 3437 |
+
"total_flos": 2.2107558569731686e+17,
|
| 3438 |
+
"train_loss": 0.48103629675806175,
|
| 3439 |
+
"train_runtime": 1582.8231,
|
| 3440 |
+
"train_samples_per_second": 9.856,
|
| 3441 |
+
"train_steps_per_second": 0.309
|
| 3442 |
+
}
|
| 3443 |
+
],
|
| 3444 |
+
"logging_steps": 1,
|
| 3445 |
+
"max_steps": 489,
|
| 3446 |
+
"num_input_tokens_seen": 0,
|
| 3447 |
+
"num_train_epochs": 3,
|
| 3448 |
+
"save_steps": 0,
|
| 3449 |
+
"stateful_callbacks": {
|
| 3450 |
+
"TrainerControl": {
|
| 3451 |
+
"args": {
|
| 3452 |
+
"should_epoch_stop": false,
|
| 3453 |
+
"should_evaluate": false,
|
| 3454 |
+
"should_log": false,
|
| 3455 |
+
"should_save": true,
|
| 3456 |
+
"should_training_stop": true
|
| 3457 |
+
},
|
| 3458 |
+
"attributes": {}
|
| 3459 |
+
}
|
| 3460 |
+
},
|
| 3461 |
+
"total_flos": 2.2107558569731686e+17,
|
| 3462 |
+
"train_batch_size": 8,
|
| 3463 |
+
"trial_name": null,
|
| 3464 |
+
"trial_params": null
|
| 3465 |
+
}
|