Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +1467 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1467 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.998691442030882,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 477,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010468463752944255,
|
| 13 |
+
"grad_norm": 3.760205659470345,
|
| 14 |
+
"learning_rate": 5.208333333333333e-08,
|
| 15 |
+
"logits/chosen": -2.9944963455200195,
|
| 16 |
+
"logits/rejected": -2.9324326515197754,
|
| 17 |
+
"logps/chosen": -1.0260854959487915,
|
| 18 |
+
"logps/rejected": -1.189737319946289,
|
| 19 |
+
"loss": -0.0244,
|
| 20 |
+
"rewards/accuracies": 0.612500011920929,
|
| 21 |
+
"rewards/chosen": -1.0260854959487915,
|
| 22 |
+
"rewards/margins": 0.1636517494916916,
|
| 23 |
+
"rewards/rejected": -1.189737319946289,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.02093692750588851,
|
| 28 |
+
"grad_norm": 4.14164094224387,
|
| 29 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 30 |
+
"logits/chosen": -3.0354819297790527,
|
| 31 |
+
"logits/rejected": -2.9778807163238525,
|
| 32 |
+
"logps/chosen": -1.1017096042633057,
|
| 33 |
+
"logps/rejected": -1.1320788860321045,
|
| 34 |
+
"loss": -0.0273,
|
| 35 |
+
"rewards/accuracies": 0.512499988079071,
|
| 36 |
+
"rewards/chosen": -1.1017096042633057,
|
| 37 |
+
"rewards/margins": 0.030369168147444725,
|
| 38 |
+
"rewards/rejected": -1.1320788860321045,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.031405391258832765,
|
| 43 |
+
"grad_norm": 4.393103002918313,
|
| 44 |
+
"learning_rate": 1.5624999999999999e-07,
|
| 45 |
+
"logits/chosen": -3.0005249977111816,
|
| 46 |
+
"logits/rejected": -2.995767116546631,
|
| 47 |
+
"logps/chosen": -1.0664865970611572,
|
| 48 |
+
"logps/rejected": -1.0592315196990967,
|
| 49 |
+
"loss": -0.0311,
|
| 50 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 51 |
+
"rewards/chosen": -1.0664865970611572,
|
| 52 |
+
"rewards/margins": -0.007255014963448048,
|
| 53 |
+
"rewards/rejected": -1.0592315196990967,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.04187385501177702,
|
| 58 |
+
"grad_norm": 4.089316161372408,
|
| 59 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 60 |
+
"logits/chosen": -3.015296697616577,
|
| 61 |
+
"logits/rejected": -2.9769978523254395,
|
| 62 |
+
"logps/chosen": -1.0041254758834839,
|
| 63 |
+
"logps/rejected": -1.0410144329071045,
|
| 64 |
+
"loss": -0.0327,
|
| 65 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 66 |
+
"rewards/chosen": -1.0041254758834839,
|
| 67 |
+
"rewards/margins": 0.03688892349600792,
|
| 68 |
+
"rewards/rejected": -1.0410144329071045,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05234231876472128,
|
| 73 |
+
"grad_norm": 4.216563165990877,
|
| 74 |
+
"learning_rate": 2.604166666666667e-07,
|
| 75 |
+
"logits/chosen": -3.0237412452697754,
|
| 76 |
+
"logits/rejected": -2.973569631576538,
|
| 77 |
+
"logps/chosen": -0.9541431665420532,
|
| 78 |
+
"logps/rejected": -1.1942131519317627,
|
| 79 |
+
"loss": -0.0413,
|
| 80 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 81 |
+
"rewards/chosen": -0.9541431665420532,
|
| 82 |
+
"rewards/margins": 0.2400699108839035,
|
| 83 |
+
"rewards/rejected": -1.1942131519317627,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06281078251766553,
|
| 88 |
+
"grad_norm": 5.658050284469291,
|
| 89 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 90 |
+
"logits/chosen": -3.030153274536133,
|
| 91 |
+
"logits/rejected": -2.9625256061553955,
|
| 92 |
+
"logps/chosen": -0.8760479092597961,
|
| 93 |
+
"logps/rejected": -0.9085603952407837,
|
| 94 |
+
"loss": -0.0368,
|
| 95 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 96 |
+
"rewards/chosen": -0.8760479092597961,
|
| 97 |
+
"rewards/margins": 0.032512493431568146,
|
| 98 |
+
"rewards/rejected": -0.9085603952407837,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07327924627060979,
|
| 103 |
+
"grad_norm": 5.044506012240391,
|
| 104 |
+
"learning_rate": 3.645833333333333e-07,
|
| 105 |
+
"logits/chosen": -3.0141396522521973,
|
| 106 |
+
"logits/rejected": -2.995772361755371,
|
| 107 |
+
"logps/chosen": -0.8659467697143555,
|
| 108 |
+
"logps/rejected": -1.1117627620697021,
|
| 109 |
+
"loss": -0.0476,
|
| 110 |
+
"rewards/accuracies": 0.612500011920929,
|
| 111 |
+
"rewards/chosen": -0.8659467697143555,
|
| 112 |
+
"rewards/margins": 0.24581606686115265,
|
| 113 |
+
"rewards/rejected": -1.1117627620697021,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08374771002355404,
|
| 118 |
+
"grad_norm": 4.594342778483098,
|
| 119 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 120 |
+
"logits/chosen": -2.9778103828430176,
|
| 121 |
+
"logits/rejected": -2.9474072456359863,
|
| 122 |
+
"logps/chosen": -0.8786975741386414,
|
| 123 |
+
"logps/rejected": -0.9444503784179688,
|
| 124 |
+
"loss": -0.0484,
|
| 125 |
+
"rewards/accuracies": 0.550000011920929,
|
| 126 |
+
"rewards/chosen": -0.8786975741386414,
|
| 127 |
+
"rewards/margins": 0.06575289368629456,
|
| 128 |
+
"rewards/rejected": -0.9444503784179688,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0942161737764983,
|
| 133 |
+
"grad_norm": 4.865607139601026,
|
| 134 |
+
"learning_rate": 4.6874999999999996e-07,
|
| 135 |
+
"logits/chosen": -3.025959014892578,
|
| 136 |
+
"logits/rejected": -3.019007444381714,
|
| 137 |
+
"logps/chosen": -0.8325697183609009,
|
| 138 |
+
"logps/rejected": -0.9673300981521606,
|
| 139 |
+
"loss": -0.0615,
|
| 140 |
+
"rewards/accuracies": 0.612500011920929,
|
| 141 |
+
"rewards/chosen": -0.8325697183609009,
|
| 142 |
+
"rewards/margins": 0.134760320186615,
|
| 143 |
+
"rewards/rejected": -0.9673300981521606,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.10468463752944256,
|
| 148 |
+
"grad_norm": 5.424026197216979,
|
| 149 |
+
"learning_rate": 4.999731868769026e-07,
|
| 150 |
+
"logits/chosen": -2.8837554454803467,
|
| 151 |
+
"logits/rejected": -2.9218649864196777,
|
| 152 |
+
"logps/chosen": -0.9015443921089172,
|
| 153 |
+
"logps/rejected": -1.0318125486373901,
|
| 154 |
+
"loss": -0.0525,
|
| 155 |
+
"rewards/accuracies": 0.550000011920929,
|
| 156 |
+
"rewards/chosen": -0.9015443921089172,
|
| 157 |
+
"rewards/margins": 0.13026830554008484,
|
| 158 |
+
"rewards/rejected": -1.0318125486373901,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.11515310128238682,
|
| 163 |
+
"grad_norm": 7.570587115119674,
|
| 164 |
+
"learning_rate": 4.996716052911017e-07,
|
| 165 |
+
"logits/chosen": -3.0432026386260986,
|
| 166 |
+
"logits/rejected": -3.038170576095581,
|
| 167 |
+
"logps/chosen": -0.9203222393989563,
|
| 168 |
+
"logps/rejected": -1.0882291793823242,
|
| 169 |
+
"loss": -0.0483,
|
| 170 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 171 |
+
"rewards/chosen": -0.9203222393989563,
|
| 172 |
+
"rewards/margins": 0.1679070144891739,
|
| 173 |
+
"rewards/rejected": -1.0882291793823242,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.12562156503533106,
|
| 178 |
+
"grad_norm": 8.483337894809958,
|
| 179 |
+
"learning_rate": 4.990353313429303e-07,
|
| 180 |
+
"logits/chosen": -2.9748623371124268,
|
| 181 |
+
"logits/rejected": -2.9038889408111572,
|
| 182 |
+
"logps/chosen": -0.9680806398391724,
|
| 183 |
+
"logps/rejected": -1.1494948863983154,
|
| 184 |
+
"loss": -0.0682,
|
| 185 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 186 |
+
"rewards/chosen": -0.9680806398391724,
|
| 187 |
+
"rewards/margins": 0.18141421675682068,
|
| 188 |
+
"rewards/rejected": -1.1494948863983154,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.1360900287882753,
|
| 193 |
+
"grad_norm": 6.597482381521931,
|
| 194 |
+
"learning_rate": 4.980652179769217e-07,
|
| 195 |
+
"logits/chosen": -2.9652066230773926,
|
| 196 |
+
"logits/rejected": -2.988760232925415,
|
| 197 |
+
"logps/chosen": -0.9523311853408813,
|
| 198 |
+
"logps/rejected": -1.1883153915405273,
|
| 199 |
+
"loss": -0.0605,
|
| 200 |
+
"rewards/accuracies": 0.574999988079071,
|
| 201 |
+
"rewards/chosen": -0.9523311853408813,
|
| 202 |
+
"rewards/margins": 0.23598428070545197,
|
| 203 |
+
"rewards/rejected": -1.1883153915405273,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.14655849254121958,
|
| 208 |
+
"grad_norm": 7.227660542967968,
|
| 209 |
+
"learning_rate": 4.967625656594781e-07,
|
| 210 |
+
"logits/chosen": -3.0324559211730957,
|
| 211 |
+
"logits/rejected": -2.9957118034362793,
|
| 212 |
+
"logps/chosen": -1.03068208694458,
|
| 213 |
+
"logps/rejected": -1.2021524906158447,
|
| 214 |
+
"loss": -0.0806,
|
| 215 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 216 |
+
"rewards/chosen": -1.03068208694458,
|
| 217 |
+
"rewards/margins": 0.1714704930782318,
|
| 218 |
+
"rewards/rejected": -1.2021524906158447,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.15702695629416383,
|
| 223 |
+
"grad_norm": 8.630957672516889,
|
| 224 |
+
"learning_rate": 4.951291206355559e-07,
|
| 225 |
+
"logits/chosen": -2.99220609664917,
|
| 226 |
+
"logits/rejected": -2.9557840824127197,
|
| 227 |
+
"logps/chosen": -1.0696258544921875,
|
| 228 |
+
"logps/rejected": -1.317363977432251,
|
| 229 |
+
"loss": -0.0631,
|
| 230 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 231 |
+
"rewards/chosen": -1.0696258544921875,
|
| 232 |
+
"rewards/margins": 0.2477382868528366,
|
| 233 |
+
"rewards/rejected": -1.317363977432251,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.16749542004710807,
|
| 238 |
+
"grad_norm": 6.399531931157647,
|
| 239 |
+
"learning_rate": 4.93167072587771e-07,
|
| 240 |
+
"logits/chosen": -2.994518280029297,
|
| 241 |
+
"logits/rejected": -2.961357831954956,
|
| 242 |
+
"logps/chosen": -1.0338717699050903,
|
| 243 |
+
"logps/rejected": -1.4787652492523193,
|
| 244 |
+
"loss": -0.0783,
|
| 245 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 246 |
+
"rewards/chosen": -1.0338717699050903,
|
| 247 |
+
"rewards/margins": 0.44489365816116333,
|
| 248 |
+
"rewards/rejected": -1.4787652492523193,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.17796388380005235,
|
| 253 |
+
"grad_norm": 7.358699467281965,
|
| 254 |
+
"learning_rate": 4.908790517010636e-07,
|
| 255 |
+
"logits/chosen": -2.9917774200439453,
|
| 256 |
+
"logits/rejected": -2.9551658630371094,
|
| 257 |
+
"logps/chosen": -0.9378916025161743,
|
| 258 |
+
"logps/rejected": -1.334387183189392,
|
| 259 |
+
"loss": -0.0833,
|
| 260 |
+
"rewards/accuracies": 0.625,
|
| 261 |
+
"rewards/chosen": -0.9378916025161743,
|
| 262 |
+
"rewards/margins": 0.396495521068573,
|
| 263 |
+
"rewards/rejected": -1.334387183189392,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.1884323475529966,
|
| 268 |
+
"grad_norm": 5.13038386551715,
|
| 269 |
+
"learning_rate": 4.882681251368548e-07,
|
| 270 |
+
"logits/chosen": -2.9607720375061035,
|
| 271 |
+
"logits/rejected": -2.9549806118011475,
|
| 272 |
+
"logps/chosen": -1.0484223365783691,
|
| 273 |
+
"logps/rejected": -1.3985503911972046,
|
| 274 |
+
"loss": -0.0862,
|
| 275 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 276 |
+
"rewards/chosen": -1.0484223365783691,
|
| 277 |
+
"rewards/margins": 0.3501282036304474,
|
| 278 |
+
"rewards/rejected": -1.3985503911972046,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.19890081130594087,
|
| 283 |
+
"grad_norm": 6.228634016037295,
|
| 284 |
+
"learning_rate": 4.853377929214243e-07,
|
| 285 |
+
"logits/chosen": -3.039447784423828,
|
| 286 |
+
"logits/rejected": -3.0229854583740234,
|
| 287 |
+
"logps/chosen": -1.0390279293060303,
|
| 288 |
+
"logps/rejected": -1.2659882307052612,
|
| 289 |
+
"loss": -0.0748,
|
| 290 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 291 |
+
"rewards/chosen": -1.0390279293060303,
|
| 292 |
+
"rewards/margins": 0.22696030139923096,
|
| 293 |
+
"rewards/rejected": -1.2659882307052612,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.2093692750588851,
|
| 298 |
+
"grad_norm": 13.46485741296935,
|
| 299 |
+
"learning_rate": 4.820919832540181e-07,
|
| 300 |
+
"logits/chosen": -3.0565896034240723,
|
| 301 |
+
"logits/rejected": -3.0682952404022217,
|
| 302 |
+
"logps/chosen": -1.0358636379241943,
|
| 303 |
+
"logps/rejected": -1.3846577405929565,
|
| 304 |
+
"loss": -0.0825,
|
| 305 |
+
"rewards/accuracies": 0.512499988079071,
|
| 306 |
+
"rewards/chosen": -1.0358636379241943,
|
| 307 |
+
"rewards/margins": 0.34879425168037415,
|
| 308 |
+
"rewards/rejected": -1.3846577405929565,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.21983773881182936,
|
| 313 |
+
"grad_norm": 13.308269094275541,
|
| 314 |
+
"learning_rate": 4.785350472409791e-07,
|
| 315 |
+
"logits/chosen": -3.013173818588257,
|
| 316 |
+
"logits/rejected": -3.020575761795044,
|
| 317 |
+
"logps/chosen": -1.2395133972167969,
|
| 318 |
+
"logps/rejected": -1.6027743816375732,
|
| 319 |
+
"loss": -0.1058,
|
| 320 |
+
"rewards/accuracies": 0.675000011920929,
|
| 321 |
+
"rewards/chosen": -1.2395133972167969,
|
| 322 |
+
"rewards/margins": 0.363260954618454,
|
| 323 |
+
"rewards/rejected": -1.6027743816375732,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.23030620256477363,
|
| 328 |
+
"grad_norm": 14.80943809386291,
|
| 329 |
+
"learning_rate": 4.7467175306295647e-07,
|
| 330 |
+
"logits/chosen": -2.907844066619873,
|
| 331 |
+
"logits/rejected": -2.9418551921844482,
|
| 332 |
+
"logps/chosen": -1.0689195394515991,
|
| 333 |
+
"logps/rejected": -1.62472403049469,
|
| 334 |
+
"loss": -0.1078,
|
| 335 |
+
"rewards/accuracies": 0.625,
|
| 336 |
+
"rewards/chosen": -1.0689195394515991,
|
| 337 |
+
"rewards/margins": 0.5558044910430908,
|
| 338 |
+
"rewards/rejected": -1.62472403049469,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.24077466631771788,
|
| 343 |
+
"grad_norm": 7.302061093053783,
|
| 344 |
+
"learning_rate": 4.70507279583015e-07,
|
| 345 |
+
"logits/chosen": -3.064298152923584,
|
| 346 |
+
"logits/rejected": -3.0547451972961426,
|
| 347 |
+
"logps/chosen": -1.0759824514389038,
|
| 348 |
+
"logps/rejected": -1.38835608959198,
|
| 349 |
+
"loss": -0.089,
|
| 350 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 351 |
+
"rewards/chosen": -1.0759824514389038,
|
| 352 |
+
"rewards/margins": 0.3123735785484314,
|
| 353 |
+
"rewards/rejected": -1.38835608959198,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.2512431300706621,
|
| 358 |
+
"grad_norm": 6.576904824772524,
|
| 359 |
+
"learning_rate": 4.6604720940421207e-07,
|
| 360 |
+
"logits/chosen": -2.995882272720337,
|
| 361 |
+
"logits/rejected": -2.9872708320617676,
|
| 362 |
+
"logps/chosen": -1.010962724685669,
|
| 363 |
+
"logps/rejected": -1.5580971240997314,
|
| 364 |
+
"loss": -0.0761,
|
| 365 |
+
"rewards/accuracies": 0.625,
|
| 366 |
+
"rewards/chosen": -1.010962724685669,
|
| 367 |
+
"rewards/margins": 0.547134518623352,
|
| 368 |
+
"rewards/rejected": -1.5580971240997314,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.26171159382360637,
|
| 373 |
+
"grad_norm": 10.915099417857844,
|
| 374 |
+
"learning_rate": 4.612975213859487e-07,
|
| 375 |
+
"logits/chosen": -2.8915393352508545,
|
| 376 |
+
"logits/rejected": -2.841787338256836,
|
| 377 |
+
"logps/chosen": -1.0743449926376343,
|
| 378 |
+
"logps/rejected": -1.400960922241211,
|
| 379 |
+
"loss": -0.0844,
|
| 380 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 381 |
+
"rewards/chosen": -1.0743449926376343,
|
| 382 |
+
"rewards/margins": 0.3266158401966095,
|
| 383 |
+
"rewards/rejected": -1.400960922241211,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.2721800575765506,
|
| 388 |
+
"grad_norm": 7.5813693492052066,
|
| 389 |
+
"learning_rate": 4.5626458262912735e-07,
|
| 390 |
+
"logits/chosen": -2.9785983562469482,
|
| 391 |
+
"logits/rejected": -3.023461103439331,
|
| 392 |
+
"logps/chosen": -1.0497643947601318,
|
| 393 |
+
"logps/rejected": -1.5954610109329224,
|
| 394 |
+
"loss": -0.0988,
|
| 395 |
+
"rewards/accuracies": 0.675000011920929,
|
| 396 |
+
"rewards/chosen": -1.0497643947601318,
|
| 397 |
+
"rewards/margins": 0.5456965565681458,
|
| 398 |
+
"rewards/rejected": -1.5954610109329224,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.2826485213294949,
|
| 403 |
+
"grad_norm": 7.880456151156036,
|
| 404 |
+
"learning_rate": 4.5095513994085974e-07,
|
| 405 |
+
"logits/chosen": -2.9451870918273926,
|
| 406 |
+
"logits/rejected": -2.9251372814178467,
|
| 407 |
+
"logps/chosen": -1.1994966268539429,
|
| 408 |
+
"logps/rejected": -1.6415138244628906,
|
| 409 |
+
"loss": -0.0959,
|
| 410 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 411 |
+
"rewards/chosen": -1.1994966268539429,
|
| 412 |
+
"rewards/margins": 0.442017138004303,
|
| 413 |
+
"rewards/rejected": -1.6415138244628906,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.29311698508243916,
|
| 418 |
+
"grad_norm": 11.389751848981007,
|
| 419 |
+
"learning_rate": 4.453763107901675e-07,
|
| 420 |
+
"logits/chosen": -2.9757840633392334,
|
| 421 |
+
"logits/rejected": -2.9875569343566895,
|
| 422 |
+
"logps/chosen": -1.3223577737808228,
|
| 423 |
+
"logps/rejected": -1.7084649801254272,
|
| 424 |
+
"loss": -0.0795,
|
| 425 |
+
"rewards/accuracies": 0.5625,
|
| 426 |
+
"rewards/chosen": -1.3223577737808228,
|
| 427 |
+
"rewards/margins": 0.38610729575157166,
|
| 428 |
+
"rewards/rejected": -1.7084649801254272,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.3035854488353834,
|
| 433 |
+
"grad_norm": 12.240889169213316,
|
| 434 |
+
"learning_rate": 4.395355737667985e-07,
|
| 435 |
+
"logits/chosen": -2.884559392929077,
|
| 436 |
+
"logits/rejected": -2.9181649684906006,
|
| 437 |
+
"logps/chosen": -1.162393569946289,
|
| 438 |
+
"logps/rejected": -1.6213098764419556,
|
| 439 |
+
"loss": -0.0999,
|
| 440 |
+
"rewards/accuracies": 0.625,
|
| 441 |
+
"rewards/chosen": -1.162393569946289,
|
| 442 |
+
"rewards/margins": 0.4589163362979889,
|
| 443 |
+
"rewards/rejected": -1.6213098764419556,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.31405391258832765,
|
| 448 |
+
"grad_norm": 13.394133386548184,
|
| 449 |
+
"learning_rate": 4.3344075855595097e-07,
|
| 450 |
+
"logits/chosen": -2.988391160964966,
|
| 451 |
+
"logits/rejected": -2.9592318534851074,
|
| 452 |
+
"logps/chosen": -1.1906192302703857,
|
| 453 |
+
"logps/rejected": -2.017512559890747,
|
| 454 |
+
"loss": -0.0867,
|
| 455 |
+
"rewards/accuracies": 0.6875,
|
| 456 |
+
"rewards/chosen": -1.1906192302703857,
|
| 457 |
+
"rewards/margins": 0.8268932104110718,
|
| 458 |
+
"rewards/rejected": -2.017512559890747,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.3245223763412719,
|
| 463 |
+
"grad_norm": 10.341335190859326,
|
| 464 |
+
"learning_rate": 4.271000354423425e-07,
|
| 465 |
+
"logits/chosen": -2.871455192565918,
|
| 466 |
+
"logits/rejected": -2.853935718536377,
|
| 467 |
+
"logps/chosen": -1.3897101879119873,
|
| 468 |
+
"logps/rejected": -1.8974939584732056,
|
| 469 |
+
"loss": -0.1116,
|
| 470 |
+
"rewards/accuracies": 0.699999988079071,
|
| 471 |
+
"rewards/chosen": -1.3897101879119873,
|
| 472 |
+
"rewards/margins": 0.5077837705612183,
|
| 473 |
+
"rewards/rejected": -1.8974939584732056,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.33499084009421615,
|
| 478 |
+
"grad_norm": 14.583251944991732,
|
| 479 |
+
"learning_rate": 4.2052190435769554e-07,
|
| 480 |
+
"logits/chosen": -2.824887752532959,
|
| 481 |
+
"logits/rejected": -2.8534679412841797,
|
| 482 |
+
"logps/chosen": -1.3409488201141357,
|
| 483 |
+
"logps/rejected": -2.2745461463928223,
|
| 484 |
+
"loss": -0.1046,
|
| 485 |
+
"rewards/accuracies": 0.612500011920929,
|
| 486 |
+
"rewards/chosen": -1.3409488201141357,
|
| 487 |
+
"rewards/margins": 0.933597207069397,
|
| 488 |
+
"rewards/rejected": -2.2745461463928223,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.34545930384716045,
|
| 493 |
+
"grad_norm": 13.977236800680524,
|
| 494 |
+
"learning_rate": 4.137151834863213e-07,
|
| 495 |
+
"logits/chosen": -2.9754512310028076,
|
| 496 |
+
"logits/rejected": -2.960723400115967,
|
| 497 |
+
"logps/chosen": -1.255544662475586,
|
| 498 |
+
"logps/rejected": -1.7897508144378662,
|
| 499 |
+
"loss": -0.1166,
|
| 500 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 501 |
+
"rewards/chosen": -1.255544662475586,
|
| 502 |
+
"rewards/margins": 0.534206211566925,
|
| 503 |
+
"rewards/rejected": -1.7897508144378662,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.3559277676001047,
|
| 508 |
+
"grad_norm": 21.73429928886912,
|
| 509 |
+
"learning_rate": 4.0668899744407567e-07,
|
| 510 |
+
"logits/chosen": -2.92985200881958,
|
| 511 |
+
"logits/rejected": -2.9375405311584473,
|
| 512 |
+
"logps/chosen": -1.1773474216461182,
|
| 513 |
+
"logps/rejected": -1.8664484024047852,
|
| 514 |
+
"loss": -0.1043,
|
| 515 |
+
"rewards/accuracies": 0.675000011920929,
|
| 516 |
+
"rewards/chosen": -1.1773474216461182,
|
| 517 |
+
"rewards/margins": 0.6891010999679565,
|
| 518 |
+
"rewards/rejected": -1.8664484024047852,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.36639623135304894,
|
| 523 |
+
"grad_norm": 65.37359700720873,
|
| 524 |
+
"learning_rate": 3.994527650465352e-07,
|
| 525 |
+
"logits/chosen": -2.851064682006836,
|
| 526 |
+
"logits/rejected": -2.87498140335083,
|
| 527 |
+
"logps/chosen": -1.0109388828277588,
|
| 528 |
+
"logps/rejected": -1.6845824718475342,
|
| 529 |
+
"loss": -0.1099,
|
| 530 |
+
"rewards/accuracies": 0.699999988079071,
|
| 531 |
+
"rewards/chosen": -1.0109388828277588,
|
| 532 |
+
"rewards/margins": 0.6736436486244202,
|
| 533 |
+
"rewards/rejected": -1.6845824718475342,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.3768646951059932,
|
| 538 |
+
"grad_norm": 18.528930954008196,
|
| 539 |
+
"learning_rate": 3.920161866827889e-07,
|
| 540 |
+
"logits/chosen": -2.8931515216827393,
|
| 541 |
+
"logits/rejected": -2.873427629470825,
|
| 542 |
+
"logps/chosen": -1.1016333103179932,
|
| 543 |
+
"logps/rejected": -1.9188123941421509,
|
| 544 |
+
"loss": -0.1115,
|
| 545 |
+
"rewards/accuracies": 0.625,
|
| 546 |
+
"rewards/chosen": -1.1016333103179932,
|
| 547 |
+
"rewards/margins": 0.8171789050102234,
|
| 548 |
+
"rewards/rejected": -1.9188123941421509,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.38733315885893743,
|
| 553 |
+
"grad_norm": 7.80304044080231,
|
| 554 |
+
"learning_rate": 3.8438923131177237e-07,
|
| 555 |
+
"logits/chosen": -2.8918251991271973,
|
| 556 |
+
"logits/rejected": -2.9344098567962646,
|
| 557 |
+
"logps/chosen": -1.0134468078613281,
|
| 558 |
+
"logps/rejected": -1.6890709400177002,
|
| 559 |
+
"loss": -0.1128,
|
| 560 |
+
"rewards/accuracies": 0.6875,
|
| 561 |
+
"rewards/chosen": -1.0134468078613281,
|
| 562 |
+
"rewards/margins": 0.6756240129470825,
|
| 563 |
+
"rewards/rejected": -1.6890709400177002,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.39780162261188173,
|
| 568 |
+
"grad_norm": 7.7047015056100046,
|
| 569 |
+
"learning_rate": 3.765821230985757e-07,
|
| 570 |
+
"logits/chosen": -2.8743629455566406,
|
| 571 |
+
"logits/rejected": -2.911881685256958,
|
| 572 |
+
"logps/chosen": -1.0063416957855225,
|
| 573 |
+
"logps/rejected": -1.913240671157837,
|
| 574 |
+
"loss": -0.1187,
|
| 575 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 576 |
+
"rewards/chosen": -1.0063416957855225,
|
| 577 |
+
"rewards/margins": 0.9068989753723145,
|
| 578 |
+
"rewards/rejected": -1.913240671157837,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.408270086364826,
|
| 583 |
+
"grad_norm": 7.811346817640016,
|
| 584 |
+
"learning_rate": 3.6860532770864005e-07,
|
| 585 |
+
"logits/chosen": -2.9131577014923096,
|
| 586 |
+
"logits/rejected": -2.89136004447937,
|
| 587 |
+
"logps/chosen": -1.1495698690414429,
|
| 588 |
+
"logps/rejected": -2.071730375289917,
|
| 589 |
+
"loss": -0.0945,
|
| 590 |
+
"rewards/accuracies": 0.699999988079071,
|
| 591 |
+
"rewards/chosen": -1.1495698690414429,
|
| 592 |
+
"rewards/margins": 0.9221606254577637,
|
| 593 |
+
"rewards/rejected": -2.071730375289917,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4187385501177702,
|
| 598 |
+
"grad_norm": 11.061536990800207,
|
| 599 |
+
"learning_rate": 3.604695382782159e-07,
|
| 600 |
+
"logits/chosen": -2.813307523727417,
|
| 601 |
+
"logits/rejected": -2.7539114952087402,
|
| 602 |
+
"logps/chosen": -1.268898367881775,
|
| 603 |
+
"logps/rejected": -1.69308602809906,
|
| 604 |
+
"loss": -0.1163,
|
| 605 |
+
"rewards/accuracies": 0.5625,
|
| 606 |
+
"rewards/chosen": -1.268898367881775,
|
| 607 |
+
"rewards/margins": 0.4241875112056732,
|
| 608 |
+
"rewards/rejected": -1.69308602809906,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.42920701387071447,
|
| 613 |
+
"grad_norm": 15.040971267646988,
|
| 614 |
+
"learning_rate": 3.5218566107988867e-07,
|
| 615 |
+
"logits/chosen": -2.8270034790039062,
|
| 616 |
+
"logits/rejected": -2.855402946472168,
|
| 617 |
+
"logps/chosen": -1.5387921333312988,
|
| 618 |
+
"logps/rejected": -2.168403148651123,
|
| 619 |
+
"loss": -0.1198,
|
| 620 |
+
"rewards/accuracies": 0.625,
|
| 621 |
+
"rewards/chosen": -1.5387921333312988,
|
| 622 |
+
"rewards/margins": 0.6296108961105347,
|
| 623 |
+
"rewards/rejected": -2.168403148651123,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.4396754776236587,
|
| 628 |
+
"grad_norm": 18.3545573133183,
|
| 629 |
+
"learning_rate": 3.4376480090239047e-07,
|
| 630 |
+
"logits/chosen": -2.818117618560791,
|
| 631 |
+
"logits/rejected": -2.7833092212677,
|
| 632 |
+
"logps/chosen": -1.293579339981079,
|
| 633 |
+
"logps/rejected": -1.8654571771621704,
|
| 634 |
+
"loss": -0.1002,
|
| 635 |
+
"rewards/accuracies": 0.637499988079071,
|
| 636 |
+
"rewards/chosen": -1.293579339981079,
|
| 637 |
+
"rewards/margins": 0.5718778371810913,
|
| 638 |
+
"rewards/rejected": -1.8654571771621704,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.45014394137660296,
|
| 643 |
+
"grad_norm": 8.458661335807625,
|
| 644 |
+
"learning_rate": 3.3521824616429284e-07,
|
| 645 |
+
"logits/chosen": -2.7233188152313232,
|
| 646 |
+
"logits/rejected": -2.727560043334961,
|
| 647 |
+
"logps/chosen": -1.3122286796569824,
|
| 648 |
+
"logps/rejected": -2.4676425457000732,
|
| 649 |
+
"loss": -0.1053,
|
| 650 |
+
"rewards/accuracies": 0.762499988079071,
|
| 651 |
+
"rewards/chosen": -1.3122286796569824,
|
| 652 |
+
"rewards/margins": 1.1554138660430908,
|
| 653 |
+
"rewards/rejected": -2.4676425457000732,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.46061240512954726,
|
| 658 |
+
"grad_norm": 8.86883909656063,
|
| 659 |
+
"learning_rate": 3.265574537815398e-07,
|
| 660 |
+
"logits/chosen": -2.719998359680176,
|
| 661 |
+
"logits/rejected": -2.7265241146087646,
|
| 662 |
+
"logps/chosen": -1.129214882850647,
|
| 663 |
+
"logps/rejected": -1.7588284015655518,
|
| 664 |
+
"loss": -0.1146,
|
| 665 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 666 |
+
"rewards/chosen": -1.129214882850647,
|
| 667 |
+
"rewards/margins": 0.6296136975288391,
|
| 668 |
+
"rewards/rejected": -1.7588284015655518,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.4710808688824915,
|
| 673 |
+
"grad_norm": 10.016548371714508,
|
| 674 |
+
"learning_rate": 3.1779403380910425e-07,
|
| 675 |
+
"logits/chosen": -2.811687707901001,
|
| 676 |
+
"logits/rejected": -2.7085366249084473,
|
| 677 |
+
"logps/chosen": -1.3638594150543213,
|
| 678 |
+
"logps/rejected": -2.3513691425323486,
|
| 679 |
+
"loss": -0.1184,
|
| 680 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 681 |
+
"rewards/chosen": -1.3638594150543213,
|
| 682 |
+
"rewards/margins": 0.9875099062919617,
|
| 683 |
+
"rewards/rejected": -2.3513691425323486,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.48154933263543576,
|
| 688 |
+
"grad_norm": 21.998434755048503,
|
| 689 |
+
"learning_rate": 3.0893973387735683e-07,
|
| 690 |
+
"logits/chosen": -2.6830623149871826,
|
| 691 |
+
"logits/rejected": -2.6364707946777344,
|
| 692 |
+
"logps/chosen": -1.3768701553344727,
|
| 693 |
+
"logps/rejected": -2.424539089202881,
|
| 694 |
+
"loss": -0.1267,
|
| 695 |
+
"rewards/accuracies": 0.625,
|
| 696 |
+
"rewards/chosen": -1.3768701553344727,
|
| 697 |
+
"rewards/margins": 1.0476689338684082,
|
| 698 |
+
"rewards/rejected": -2.424539089202881,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.49201779638838,
|
| 703 |
+
"grad_norm": 15.145805829963615,
|
| 704 |
+
"learning_rate": 3.000064234440111e-07,
|
| 705 |
+
"logits/chosen": -2.4627432823181152,
|
| 706 |
+
"logits/rejected": -2.4076457023620605,
|
| 707 |
+
"logps/chosen": -1.2042968273162842,
|
| 708 |
+
"logps/rejected": -2.078500270843506,
|
| 709 |
+
"loss": -0.1185,
|
| 710 |
+
"rewards/accuracies": 0.612500011920929,
|
| 711 |
+
"rewards/chosen": -1.2042968273162842,
|
| 712 |
+
"rewards/margins": 0.8742035031318665,
|
| 713 |
+
"rewards/rejected": -2.078500270843506,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5024862601413242,
|
| 718 |
+
"grad_norm": 10.783814010063464,
|
| 719 |
+
"learning_rate": 2.910060778827554e-07,
|
| 720 |
+
"logits/chosen": -2.5319347381591797,
|
| 721 |
+
"logits/rejected": -2.5071659088134766,
|
| 722 |
+
"logps/chosen": -1.5030553340911865,
|
| 723 |
+
"logps/rejected": -2.248628616333008,
|
| 724 |
+
"loss": -0.1178,
|
| 725 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 726 |
+
"rewards/chosen": -1.5030553340911865,
|
| 727 |
+
"rewards/margins": 0.7455734014511108,
|
| 728 |
+
"rewards/rejected": -2.248628616333008,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5129547238942685,
|
| 733 |
+
"grad_norm": 17.35265824839872,
|
| 734 |
+
"learning_rate": 2.8195076242990116e-07,
|
| 735 |
+
"logits/chosen": -2.629790782928467,
|
| 736 |
+
"logits/rejected": -2.5811970233917236,
|
| 737 |
+
"logps/chosen": -1.1765772104263306,
|
| 738 |
+
"logps/rejected": -2.45619535446167,
|
| 739 |
+
"loss": -0.1283,
|
| 740 |
+
"rewards/accuracies": 0.675000011920929,
|
| 741 |
+
"rewards/chosen": -1.1765772104263306,
|
| 742 |
+
"rewards/margins": 1.2796180248260498,
|
| 743 |
+
"rewards/rejected": -2.45619535446167,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.5234231876472127,
|
| 748 |
+
"grad_norm": 21.631571252123923,
|
| 749 |
+
"learning_rate": 2.7285261601056697e-07,
|
| 750 |
+
"logits/chosen": -2.559217929840088,
|
| 751 |
+
"logits/rejected": -2.4746146202087402,
|
| 752 |
+
"logps/chosen": -1.4234548807144165,
|
| 753 |
+
"logps/rejected": -2.4785306453704834,
|
| 754 |
+
"loss": -0.1032,
|
| 755 |
+
"rewards/accuracies": 0.637499988079071,
|
| 756 |
+
"rewards/chosen": -1.4234548807144165,
|
| 757 |
+
"rewards/margins": 1.055075764656067,
|
| 758 |
+
"rewards/rejected": -2.4785306453704834,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.533891651400157,
|
| 763 |
+
"grad_norm": 10.665899447598967,
|
| 764 |
+
"learning_rate": 2.6372383496608186e-07,
|
| 765 |
+
"logits/chosen": -2.662745237350464,
|
| 766 |
+
"logits/rejected": -2.6075425148010254,
|
| 767 |
+
"logps/chosen": -1.1387856006622314,
|
| 768 |
+
"logps/rejected": -2.129990577697754,
|
| 769 |
+
"loss": -0.1171,
|
| 770 |
+
"rewards/accuracies": 0.574999988079071,
|
| 771 |
+
"rewards/chosen": -1.1387856006622314,
|
| 772 |
+
"rewards/margins": 0.9912050366401672,
|
| 773 |
+
"rewards/rejected": -2.129990577697754,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5443601151531012,
|
| 778 |
+
"grad_norm": 30.24510546028332,
|
| 779 |
+
"learning_rate": 2.5457665670441937e-07,
|
| 780 |
+
"logits/chosen": -2.5815982818603516,
|
| 781 |
+
"logits/rejected": -2.5034749507904053,
|
| 782 |
+
"logps/chosen": -1.3083416223526,
|
| 783 |
+
"logps/rejected": -2.3620128631591797,
|
| 784 |
+
"loss": -0.1357,
|
| 785 |
+
"rewards/accuracies": 0.737500011920929,
|
| 786 |
+
"rewards/chosen": -1.3083416223526,
|
| 787 |
+
"rewards/margins": 1.0536713600158691,
|
| 788 |
+
"rewards/rejected": -2.3620128631591797,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5548285789060455,
|
| 793 |
+
"grad_norm": 36.43768779189584,
|
| 794 |
+
"learning_rate": 2.454233432955807e-07,
|
| 795 |
+
"logits/chosen": -2.4912540912628174,
|
| 796 |
+
"logits/rejected": -2.503256320953369,
|
| 797 |
+
"logps/chosen": -1.527940034866333,
|
| 798 |
+
"logps/rejected": -2.379206895828247,
|
| 799 |
+
"loss": -0.1328,
|
| 800 |
+
"rewards/accuracies": 0.675000011920929,
|
| 801 |
+
"rewards/chosen": -1.527940034866333,
|
| 802 |
+
"rewards/margins": 0.8512667417526245,
|
| 803 |
+
"rewards/rejected": -2.379206895828247,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.5652970426589898,
|
| 808 |
+
"grad_norm": 21.411949693790206,
|
| 809 |
+
"learning_rate": 2.3627616503391812e-07,
|
| 810 |
+
"logits/chosen": -2.429790735244751,
|
| 811 |
+
"logits/rejected": -2.451469898223877,
|
| 812 |
+
"logps/chosen": -1.8636459112167358,
|
| 813 |
+
"logps/rejected": -2.370575428009033,
|
| 814 |
+
"loss": -0.1231,
|
| 815 |
+
"rewards/accuracies": 0.612500011920929,
|
| 816 |
+
"rewards/chosen": -1.8636459112167358,
|
| 817 |
+
"rewards/margins": 0.5069290399551392,
|
| 818 |
+
"rewards/rejected": -2.370575428009033,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.575765506411934,
|
| 823 |
+
"grad_norm": 26.28588094272757,
|
| 824 |
+
"learning_rate": 2.2714738398943308e-07,
|
| 825 |
+
"logits/chosen": -2.484380006790161,
|
| 826 |
+
"logits/rejected": -2.3554978370666504,
|
| 827 |
+
"logps/chosen": -1.374985933303833,
|
| 828 |
+
"logps/rejected": -3.2982125282287598,
|
| 829 |
+
"loss": -0.1605,
|
| 830 |
+
"rewards/accuracies": 0.737500011920929,
|
| 831 |
+
"rewards/chosen": -1.374985933303833,
|
| 832 |
+
"rewards/margins": 1.9232263565063477,
|
| 833 |
+
"rewards/rejected": -3.2982125282287598,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.5862339701648783,
|
| 838 |
+
"grad_norm": 15.64313913527933,
|
| 839 |
+
"learning_rate": 2.1804923757009882e-07,
|
| 840 |
+
"logits/chosen": -2.5084352493286133,
|
| 841 |
+
"logits/rejected": -2.379485607147217,
|
| 842 |
+
"logps/chosen": -1.2756072282791138,
|
| 843 |
+
"logps/rejected": -2.862361431121826,
|
| 844 |
+
"loss": -0.1283,
|
| 845 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 846 |
+
"rewards/chosen": -1.2756072282791138,
|
| 847 |
+
"rewards/margins": 1.586754560470581,
|
| 848 |
+
"rewards/rejected": -2.862361431121826,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.5967024339178225,
|
| 853 |
+
"grad_norm": 31.160163808849045,
|
| 854 |
+
"learning_rate": 2.089939221172446e-07,
|
| 855 |
+
"logits/chosen": -2.464437484741211,
|
| 856 |
+
"logits/rejected": -2.4234752655029297,
|
| 857 |
+
"logps/chosen": -1.7923996448516846,
|
| 858 |
+
"logps/rejected": -3.0464653968811035,
|
| 859 |
+
"loss": -0.1107,
|
| 860 |
+
"rewards/accuracies": 0.699999988079071,
|
| 861 |
+
"rewards/chosen": -1.7923996448516846,
|
| 862 |
+
"rewards/margins": 1.2540655136108398,
|
| 863 |
+
"rewards/rejected": -3.0464653968811035,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6071708976707668,
|
| 868 |
+
"grad_norm": 14.525714466074616,
|
| 869 |
+
"learning_rate": 1.9999357655598891e-07,
|
| 870 |
+
"logits/chosen": -2.3748879432678223,
|
| 871 |
+
"logits/rejected": -2.277477741241455,
|
| 872 |
+
"logps/chosen": -1.8306829929351807,
|
| 873 |
+
"logps/rejected": -2.5916666984558105,
|
| 874 |
+
"loss": -0.1171,
|
| 875 |
+
"rewards/accuracies": 0.675000011920929,
|
| 876 |
+
"rewards/chosen": -1.8306829929351807,
|
| 877 |
+
"rewards/margins": 0.7609835863113403,
|
| 878 |
+
"rewards/rejected": -2.5916666984558105,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.6176393614237111,
|
| 883 |
+
"grad_norm": 12.205724246076834,
|
| 884 |
+
"learning_rate": 1.9106026612264315e-07,
|
| 885 |
+
"logits/chosen": -2.5410079956054688,
|
| 886 |
+
"logits/rejected": -2.368896484375,
|
| 887 |
+
"logps/chosen": -1.3447208404541016,
|
| 888 |
+
"logps/rejected": -2.198944091796875,
|
| 889 |
+
"loss": -0.1286,
|
| 890 |
+
"rewards/accuracies": 0.675000011920929,
|
| 891 |
+
"rewards/chosen": -1.3447208404541016,
|
| 892 |
+
"rewards/margins": 0.8542232513427734,
|
| 893 |
+
"rewards/rejected": -2.198944091796875,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6281078251766553,
|
| 898 |
+
"grad_norm": 13.048722996544631,
|
| 899 |
+
"learning_rate": 1.8220596619089573e-07,
|
| 900 |
+
"logits/chosen": -2.3360071182250977,
|
| 901 |
+
"logits/rejected": -2.238037347793579,
|
| 902 |
+
"logps/chosen": -1.3523575067520142,
|
| 903 |
+
"logps/rejected": -2.568472146987915,
|
| 904 |
+
"loss": -0.1212,
|
| 905 |
+
"rewards/accuracies": 0.737500011920929,
|
| 906 |
+
"rewards/chosen": -1.3523575067520142,
|
| 907 |
+
"rewards/margins": 1.2161142826080322,
|
| 908 |
+
"rewards/rejected": -2.568472146987915,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6385762889295996,
|
| 913 |
+
"grad_norm": 14.965881571164392,
|
| 914 |
+
"learning_rate": 1.7344254621846017e-07,
|
| 915 |
+
"logits/chosen": -2.4726943969726562,
|
| 916 |
+
"logits/rejected": -2.4096648693084717,
|
| 917 |
+
"logps/chosen": -1.4510114192962646,
|
| 918 |
+
"logps/rejected": -2.0148894786834717,
|
| 919 |
+
"loss": -0.1379,
|
| 920 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 921 |
+
"rewards/chosen": -1.4510114192962646,
|
| 922 |
+
"rewards/margins": 0.563878059387207,
|
| 923 |
+
"rewards/rejected": -2.0148894786834717,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6490447526825438,
|
| 928 |
+
"grad_norm": 14.196438929994413,
|
| 929 |
+
"learning_rate": 1.647817538357072e-07,
|
| 930 |
+
"logits/chosen": -2.434021472930908,
|
| 931 |
+
"logits/rejected": -2.2209062576293945,
|
| 932 |
+
"logps/chosen": -1.2371381521224976,
|
| 933 |
+
"logps/rejected": -2.5744917392730713,
|
| 934 |
+
"loss": -0.1344,
|
| 935 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 936 |
+
"rewards/chosen": -1.2371381521224976,
|
| 937 |
+
"rewards/margins": 1.3373535871505737,
|
| 938 |
+
"rewards/rejected": -2.5744917392730713,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6595132164354881,
|
| 943 |
+
"grad_norm": 13.34734425640694,
|
| 944 |
+
"learning_rate": 1.562351990976095e-07,
|
| 945 |
+
"logits/chosen": -2.482234477996826,
|
| 946 |
+
"logits/rejected": -2.3337721824645996,
|
| 947 |
+
"logps/chosen": -1.7002273797988892,
|
| 948 |
+
"logps/rejected": -2.32104754447937,
|
| 949 |
+
"loss": -0.1352,
|
| 950 |
+
"rewards/accuracies": 0.625,
|
| 951 |
+
"rewards/chosen": -1.7002273797988892,
|
| 952 |
+
"rewards/margins": 0.6208202838897705,
|
| 953 |
+
"rewards/rejected": -2.32104754447937,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.6699816801884323,
|
| 958 |
+
"grad_norm": 54.856648124460555,
|
| 959 |
+
"learning_rate": 1.478143389201113e-07,
|
| 960 |
+
"logits/chosen": -2.4059951305389404,
|
| 961 |
+
"logits/rejected": -2.2941792011260986,
|
| 962 |
+
"logps/chosen": -1.7276633977890015,
|
| 963 |
+
"logps/rejected": -2.791877031326294,
|
| 964 |
+
"loss": -0.1395,
|
| 965 |
+
"rewards/accuracies": 0.6875,
|
| 966 |
+
"rewards/chosen": -1.7276633977890015,
|
| 967 |
+
"rewards/margins": 1.0642132759094238,
|
| 968 |
+
"rewards/rejected": -2.791877031326294,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.6804501439413766,
|
| 973 |
+
"grad_norm": 16.338313868801983,
|
| 974 |
+
"learning_rate": 1.3953046172178413e-07,
|
| 975 |
+
"logits/chosen": -2.3474369049072266,
|
| 976 |
+
"logits/rejected": -2.2156565189361572,
|
| 977 |
+
"logps/chosen": -1.581926941871643,
|
| 978 |
+
"logps/rejected": -3.4111123085021973,
|
| 979 |
+
"loss": -0.1119,
|
| 980 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 981 |
+
"rewards/chosen": -1.581926941871643,
|
| 982 |
+
"rewards/margins": 1.8291854858398438,
|
| 983 |
+
"rewards/rejected": -3.4111123085021973,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.6909186076943209,
|
| 988 |
+
"grad_norm": 27.599650373019013,
|
| 989 |
+
"learning_rate": 1.3139467229135998e-07,
|
| 990 |
+
"logits/chosen": -2.424649238586426,
|
| 991 |
+
"logits/rejected": -2.313764810562134,
|
| 992 |
+
"logps/chosen": -1.4561196565628052,
|
| 993 |
+
"logps/rejected": -2.5569052696228027,
|
| 994 |
+
"loss": -0.1255,
|
| 995 |
+
"rewards/accuracies": 0.675000011920929,
|
| 996 |
+
"rewards/chosen": -1.4561196565628052,
|
| 997 |
+
"rewards/margins": 1.1007858514785767,
|
| 998 |
+
"rewards/rejected": -2.5569052696228027,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.7013870714472651,
|
| 1003 |
+
"grad_norm": 13.71174939782389,
|
| 1004 |
+
"learning_rate": 1.2341787690142435e-07,
|
| 1005 |
+
"logits/chosen": -2.182265520095825,
|
| 1006 |
+
"logits/rejected": -2.240220069885254,
|
| 1007 |
+
"logps/chosen": -2.133391857147217,
|
| 1008 |
+
"logps/rejected": -2.705057382583618,
|
| 1009 |
+
"loss": -0.1229,
|
| 1010 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1011 |
+
"rewards/chosen": -2.133391857147217,
|
| 1012 |
+
"rewards/margins": 0.571665346622467,
|
| 1013 |
+
"rewards/rejected": -2.705057382583618,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7118555352002094,
|
| 1018 |
+
"grad_norm": 22.094952760541535,
|
| 1019 |
+
"learning_rate": 1.1561076868822755e-07,
|
| 1020 |
+
"logits/chosen": -2.351863384246826,
|
| 1021 |
+
"logits/rejected": -2.2399449348449707,
|
| 1022 |
+
"logps/chosen": -1.3272682428359985,
|
| 1023 |
+
"logps/rejected": -3.206008195877075,
|
| 1024 |
+
"loss": -0.1347,
|
| 1025 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1026 |
+
"rewards/chosen": -1.3272682428359985,
|
| 1027 |
+
"rewards/margins": 1.8787400722503662,
|
| 1028 |
+
"rewards/rejected": -3.206008195877075,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.7223239989531536,
|
| 1033 |
+
"grad_norm": 14.410515050019907,
|
| 1034 |
+
"learning_rate": 1.0798381331721107e-07,
|
| 1035 |
+
"logits/chosen": -2.3990814685821533,
|
| 1036 |
+
"logits/rejected": -2.256584644317627,
|
| 1037 |
+
"logps/chosen": -1.4144983291625977,
|
| 1038 |
+
"logps/rejected": -2.471238851547241,
|
| 1039 |
+
"loss": -0.1205,
|
| 1040 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1041 |
+
"rewards/chosen": -1.4144983291625977,
|
| 1042 |
+
"rewards/margins": 1.056740641593933,
|
| 1043 |
+
"rewards/rejected": -2.471238851547241,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7327924627060979,
|
| 1048 |
+
"grad_norm": 17.737495980302032,
|
| 1049 |
+
"learning_rate": 1.0054723495346482e-07,
|
| 1050 |
+
"logits/chosen": -2.3040690422058105,
|
| 1051 |
+
"logits/rejected": -1.9883592128753662,
|
| 1052 |
+
"logps/chosen": -1.8120450973510742,
|
| 1053 |
+
"logps/rejected": -2.608847141265869,
|
| 1054 |
+
"loss": -0.1314,
|
| 1055 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1056 |
+
"rewards/chosen": -1.8120450973510742,
|
| 1057 |
+
"rewards/margins": 0.7968018651008606,
|
| 1058 |
+
"rewards/rejected": -2.608847141265869,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7432609264590422,
|
| 1063 |
+
"grad_norm": 147.24723390344812,
|
| 1064 |
+
"learning_rate": 9.331100255592436e-08,
|
| 1065 |
+
"logits/chosen": -2.2860684394836426,
|
| 1066 |
+
"logits/rejected": -2.1697564125061035,
|
| 1067 |
+
"logps/chosen": -1.6789791584014893,
|
| 1068 |
+
"logps/rejected": -3.0929505825042725,
|
| 1069 |
+
"loss": -0.1132,
|
| 1070 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1071 |
+
"rewards/chosen": -1.6789791584014893,
|
| 1072 |
+
"rewards/margins": 1.413971185684204,
|
| 1073 |
+
"rewards/rejected": -3.0929505825042725,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7537293902119864,
|
| 1078 |
+
"grad_norm": 25.64073118103101,
|
| 1079 |
+
"learning_rate": 8.628481651367875e-08,
|
| 1080 |
+
"logits/chosen": -2.218724012374878,
|
| 1081 |
+
"logits/rejected": -2.1296398639678955,
|
| 1082 |
+
"logps/chosen": -1.6159321069717407,
|
| 1083 |
+
"logps/rejected": -2.5099380016326904,
|
| 1084 |
+
"loss": -0.1222,
|
| 1085 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1086 |
+
"rewards/chosen": -1.6159321069717407,
|
| 1087 |
+
"rewards/margins": 0.8940057754516602,
|
| 1088 |
+
"rewards/rejected": -2.5099380016326904,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.7641978539649307,
|
| 1093 |
+
"grad_norm": 12.902928067743126,
|
| 1094 |
+
"learning_rate": 7.947809564230445e-08,
|
| 1095 |
+
"logits/chosen": -2.3579747676849365,
|
| 1096 |
+
"logits/rejected": -2.1200146675109863,
|
| 1097 |
+
"logps/chosen": -1.5285216569900513,
|
| 1098 |
+
"logps/rejected": -3.1276838779449463,
|
| 1099 |
+
"loss": -0.145,
|
| 1100 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1101 |
+
"rewards/chosen": -1.5285216569900513,
|
| 1102 |
+
"rewards/margins": 1.5991623401641846,
|
| 1103 |
+
"rewards/rejected": -3.1276838779449463,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.7746663177178749,
|
| 1108 |
+
"grad_norm": 15.785906225773893,
|
| 1109 |
+
"learning_rate": 7.289996455765748e-08,
|
| 1110 |
+
"logits/chosen": -2.407064914703369,
|
| 1111 |
+
"logits/rejected": -2.30210542678833,
|
| 1112 |
+
"logps/chosen": -1.3575342893600464,
|
| 1113 |
+
"logps/rejected": -2.2857096195220947,
|
| 1114 |
+
"loss": -0.1468,
|
| 1115 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1116 |
+
"rewards/chosen": -1.3575342893600464,
|
| 1117 |
+
"rewards/margins": 0.9281753301620483,
|
| 1118 |
+
"rewards/rejected": -2.2857096195220947,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.7851347814708192,
|
| 1123 |
+
"grad_norm": 66.73212890394835,
|
| 1124 |
+
"learning_rate": 6.655924144404906e-08,
|
| 1125 |
+
"logits/chosen": -2.2724781036376953,
|
| 1126 |
+
"logits/rejected": -2.2895565032958984,
|
| 1127 |
+
"logps/chosen": -1.549403429031372,
|
| 1128 |
+
"logps/rejected": -2.9475514888763428,
|
| 1129 |
+
"loss": -0.1203,
|
| 1130 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1131 |
+
"rewards/chosen": -1.549403429031372,
|
| 1132 |
+
"rewards/margins": 1.3981479406356812,
|
| 1133 |
+
"rewards/rejected": -2.9475514888763428,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.7956032452237635,
|
| 1138 |
+
"grad_norm": 38.70646418643775,
|
| 1139 |
+
"learning_rate": 6.046442623320145e-08,
|
| 1140 |
+
"logits/chosen": -2.399785041809082,
|
| 1141 |
+
"logits/rejected": -2.3085193634033203,
|
| 1142 |
+
"logps/chosen": -1.3037949800491333,
|
| 1143 |
+
"logps/rejected": -2.2208690643310547,
|
| 1144 |
+
"loss": -0.1115,
|
| 1145 |
+
"rewards/accuracies": 0.75,
|
| 1146 |
+
"rewards/chosen": -1.3037949800491333,
|
| 1147 |
+
"rewards/margins": 0.9170739054679871,
|
| 1148 |
+
"rewards/rejected": -2.2208690643310547,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8060717089767077,
|
| 1153 |
+
"grad_norm": 85.36661297405541,
|
| 1154 |
+
"learning_rate": 5.4623689209832484e-08,
|
| 1155 |
+
"logits/chosen": -2.350160598754883,
|
| 1156 |
+
"logits/rejected": -2.304274559020996,
|
| 1157 |
+
"logps/chosen": -1.328881859779358,
|
| 1158 |
+
"logps/rejected": -2.426609516143799,
|
| 1159 |
+
"loss": -0.1355,
|
| 1160 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1161 |
+
"rewards/chosen": -1.328881859779358,
|
| 1162 |
+
"rewards/margins": 1.0977277755737305,
|
| 1163 |
+
"rewards/rejected": -2.426609516143799,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.816540172729652,
|
| 1168 |
+
"grad_norm": 21.914030829763547,
|
| 1169 |
+
"learning_rate": 4.904486005914027e-08,
|
| 1170 |
+
"logits/chosen": -2.41226863861084,
|
| 1171 |
+
"logits/rejected": -2.258676767349243,
|
| 1172 |
+
"logps/chosen": -1.3917608261108398,
|
| 1173 |
+
"logps/rejected": -2.255201578140259,
|
| 1174 |
+
"loss": -0.1082,
|
| 1175 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1176 |
+
"rewards/chosen": -1.3917608261108398,
|
| 1177 |
+
"rewards/margins": 0.8634408712387085,
|
| 1178 |
+
"rewards/rejected": -2.255201578140259,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.8270086364825961,
|
| 1183 |
+
"grad_norm": 11.128742517489252,
|
| 1184 |
+
"learning_rate": 4.373541737087263e-08,
|
| 1185 |
+
"logits/chosen": -2.3416173458099365,
|
| 1186 |
+
"logits/rejected": -2.183840274810791,
|
| 1187 |
+
"logps/chosen": -1.4819704294204712,
|
| 1188 |
+
"logps/rejected": -2.7424230575561523,
|
| 1189 |
+
"loss": -0.1199,
|
| 1190 |
+
"rewards/accuracies": 0.6875,
|
| 1191 |
+
"rewards/chosen": -1.4819704294204712,
|
| 1192 |
+
"rewards/margins": 1.2604527473449707,
|
| 1193 |
+
"rewards/rejected": -2.7424230575561523,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8374771002355405,
|
| 1198 |
+
"grad_norm": 13.499128871751621,
|
| 1199 |
+
"learning_rate": 3.8702478614051345e-08,
|
| 1200 |
+
"logits/chosen": -2.4478988647460938,
|
| 1201 |
+
"logits/rejected": -2.2530007362365723,
|
| 1202 |
+
"logps/chosen": -1.503173589706421,
|
| 1203 |
+
"logps/rejected": -2.4205775260925293,
|
| 1204 |
+
"loss": -0.1196,
|
| 1205 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1206 |
+
"rewards/chosen": -1.503173589706421,
|
| 1207 |
+
"rewards/margins": 0.9174041748046875,
|
| 1208 |
+
"rewards/rejected": -2.4205775260925293,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8479455639884846,
|
| 1213 |
+
"grad_norm": 46.00477049565933,
|
| 1214 |
+
"learning_rate": 3.3952790595787986e-08,
|
| 1215 |
+
"logits/chosen": -2.428011894226074,
|
| 1216 |
+
"logits/rejected": -2.195507526397705,
|
| 1217 |
+
"logps/chosen": -1.488608956336975,
|
| 1218 |
+
"logps/rejected": -3.2226758003234863,
|
| 1219 |
+
"loss": -0.1329,
|
| 1220 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1221 |
+
"rewards/chosen": -1.488608956336975,
|
| 1222 |
+
"rewards/margins": 1.7340667247772217,
|
| 1223 |
+
"rewards/rejected": -3.2226758003234863,
|
| 1224 |
+
"step": 405
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 0.8584140277414289,
|
| 1228 |
+
"grad_norm": 15.147104463873568,
|
| 1229 |
+
"learning_rate": 2.9492720416985e-08,
|
| 1230 |
+
"logits/chosen": -2.3303403854370117,
|
| 1231 |
+
"logits/rejected": -2.2017440795898438,
|
| 1232 |
+
"logps/chosen": -1.4811475276947021,
|
| 1233 |
+
"logps/rejected": -3.1947004795074463,
|
| 1234 |
+
"loss": -0.1345,
|
| 1235 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1236 |
+
"rewards/chosen": -1.4811475276947021,
|
| 1237 |
+
"rewards/margins": 1.7135528326034546,
|
| 1238 |
+
"rewards/rejected": -3.1947004795074463,
|
| 1239 |
+
"step": 410
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.8688824914943732,
|
| 1243 |
+
"grad_norm": 32.54043568171796,
|
| 1244 |
+
"learning_rate": 2.5328246937043525e-08,
|
| 1245 |
+
"logits/chosen": -2.3127968311309814,
|
| 1246 |
+
"logits/rejected": -2.2588541507720947,
|
| 1247 |
+
"logps/chosen": -1.7812035083770752,
|
| 1248 |
+
"logps/rejected": -2.363581418991089,
|
| 1249 |
+
"loss": -0.1144,
|
| 1250 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1251 |
+
"rewards/chosen": -1.7812035083770752,
|
| 1252 |
+
"rewards/margins": 0.5823777318000793,
|
| 1253 |
+
"rewards/rejected": -2.363581418991089,
|
| 1254 |
+
"step": 415
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 0.8793509552473174,
|
| 1258 |
+
"grad_norm": 16.364361772449453,
|
| 1259 |
+
"learning_rate": 2.1464952759020856e-08,
|
| 1260 |
+
"logits/chosen": -2.2632575035095215,
|
| 1261 |
+
"logits/rejected": -2.1989002227783203,
|
| 1262 |
+
"logps/chosen": -1.3589812517166138,
|
| 1263 |
+
"logps/rejected": -2.665706157684326,
|
| 1264 |
+
"loss": -0.1402,
|
| 1265 |
+
"rewards/accuracies": 0.6875,
|
| 1266 |
+
"rewards/chosen": -1.3589812517166138,
|
| 1267 |
+
"rewards/margins": 1.3067247867584229,
|
| 1268 |
+
"rewards/rejected": -2.665706157684326,
|
| 1269 |
+
"step": 420
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.8898194190002617,
|
| 1273 |
+
"grad_norm": 24.459898730148318,
|
| 1274 |
+
"learning_rate": 1.7908016745981856e-08,
|
| 1275 |
+
"logits/chosen": -2.4754738807678223,
|
| 1276 |
+
"logits/rejected": -2.2798328399658203,
|
| 1277 |
+
"logps/chosen": -1.1175788640975952,
|
| 1278 |
+
"logps/rejected": -2.699981212615967,
|
| 1279 |
+
"loss": -0.1488,
|
| 1280 |
+
"rewards/accuracies": 0.6875,
|
| 1281 |
+
"rewards/chosen": -1.1175788640975952,
|
| 1282 |
+
"rewards/margins": 1.5824021100997925,
|
| 1283 |
+
"rewards/rejected": -2.699981212615967,
|
| 1284 |
+
"step": 425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.9002878827532059,
|
| 1288 |
+
"grad_norm": 20.15635024127512,
|
| 1289 |
+
"learning_rate": 1.4662207078575684e-08,
|
| 1290 |
+
"logits/chosen": -2.3345370292663574,
|
| 1291 |
+
"logits/rejected": -2.0842413902282715,
|
| 1292 |
+
"logps/chosen": -1.416595458984375,
|
| 1293 |
+
"logps/rejected": -2.772968053817749,
|
| 1294 |
+
"loss": -0.1388,
|
| 1295 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1296 |
+
"rewards/chosen": -1.416595458984375,
|
| 1297 |
+
"rewards/margins": 1.3563728332519531,
|
| 1298 |
+
"rewards/rejected": -2.772968053817749,
|
| 1299 |
+
"step": 430
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.9107563465061502,
|
| 1303 |
+
"grad_norm": 25.1930677516597,
|
| 1304 |
+
"learning_rate": 1.1731874863145142e-08,
|
| 1305 |
+
"logits/chosen": -2.2844507694244385,
|
| 1306 |
+
"logits/rejected": -2.2151055335998535,
|
| 1307 |
+
"logps/chosen": -1.5157394409179688,
|
| 1308 |
+
"logps/rejected": -2.6189606189727783,
|
| 1309 |
+
"loss": -0.1335,
|
| 1310 |
+
"rewards/accuracies": 0.625,
|
| 1311 |
+
"rewards/chosen": -1.5157394409179688,
|
| 1312 |
+
"rewards/margins": 1.1032211780548096,
|
| 1313 |
+
"rewards/rejected": -2.6189606189727783,
|
| 1314 |
+
"step": 435
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 0.9212248102590945,
|
| 1318 |
+
"grad_norm": 94.3340138658481,
|
| 1319 |
+
"learning_rate": 9.12094829893642e-09,
|
| 1320 |
+
"logits/chosen": -2.3165175914764404,
|
| 1321 |
+
"logits/rejected": -2.216142177581787,
|
| 1322 |
+
"logps/chosen": -1.6333303451538086,
|
| 1323 |
+
"logps/rejected": -2.6600871086120605,
|
| 1324 |
+
"loss": -0.1365,
|
| 1325 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1326 |
+
"rewards/chosen": -1.6333303451538086,
|
| 1327 |
+
"rewards/margins": 1.0267568826675415,
|
| 1328 |
+
"rewards/rejected": -2.6600871086120605,
|
| 1329 |
+
"step": 440
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 0.9316932740120387,
|
| 1333 |
+
"grad_norm": 37.68445074958579,
|
| 1334 |
+
"learning_rate": 6.832927412229017e-09,
|
| 1335 |
+
"logits/chosen": -2.173781394958496,
|
| 1336 |
+
"logits/rejected": -2.0770103931427,
|
| 1337 |
+
"logps/chosen": -1.7893623113632202,
|
| 1338 |
+
"logps/rejected": -2.782428026199341,
|
| 1339 |
+
"loss": -0.1362,
|
| 1340 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1341 |
+
"rewards/chosen": -1.7893623113632202,
|
| 1342 |
+
"rewards/margins": 0.9930654764175415,
|
| 1343 |
+
"rewards/rejected": -2.782428026199341,
|
| 1344 |
+
"step": 445
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 0.942161737764983,
|
| 1348 |
+
"grad_norm": 65.20343901413754,
|
| 1349 |
+
"learning_rate": 4.8708793644441086e-09,
|
| 1350 |
+
"logits/chosen": -2.4339003562927246,
|
| 1351 |
+
"logits/rejected": -2.235538959503174,
|
| 1352 |
+
"logps/chosen": -1.1971713304519653,
|
| 1353 |
+
"logps/rejected": -2.028782367706299,
|
| 1354 |
+
"loss": -0.1358,
|
| 1355 |
+
"rewards/accuracies": 0.625,
|
| 1356 |
+
"rewards/chosen": -1.1971713304519653,
|
| 1357 |
+
"rewards/margins": 0.8316110372543335,
|
| 1358 |
+
"rewards/rejected": -2.028782367706299,
|
| 1359 |
+
"step": 450
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.9526302015179272,
|
| 1363 |
+
"grad_norm": 29.43315569875579,
|
| 1364 |
+
"learning_rate": 3.2374343405217884e-09,
|
| 1365 |
+
"logits/chosen": -2.406641721725464,
|
| 1366 |
+
"logits/rejected": -2.2101848125457764,
|
| 1367 |
+
"logps/chosen": -1.7016384601593018,
|
| 1368 |
+
"logps/rejected": -2.7031149864196777,
|
| 1369 |
+
"loss": -0.1135,
|
| 1370 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1371 |
+
"rewards/chosen": -1.7016384601593018,
|
| 1372 |
+
"rewards/margins": 1.0014766454696655,
|
| 1373 |
+
"rewards/rejected": -2.7031149864196777,
|
| 1374 |
+
"step": 455
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.9630986652708715,
|
| 1378 |
+
"grad_norm": 58.96075147402077,
|
| 1379 |
+
"learning_rate": 1.9347820230782295e-09,
|
| 1380 |
+
"logits/chosen": -2.5120296478271484,
|
| 1381 |
+
"logits/rejected": -2.0890145301818848,
|
| 1382 |
+
"logps/chosen": -1.2953853607177734,
|
| 1383 |
+
"logps/rejected": -3.554316759109497,
|
| 1384 |
+
"loss": -0.1399,
|
| 1385 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1386 |
+
"rewards/chosen": -1.2953853607177734,
|
| 1387 |
+
"rewards/margins": 2.2589316368103027,
|
| 1388 |
+
"rewards/rejected": -3.554316759109497,
|
| 1389 |
+
"step": 460
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.9735671290238157,
|
| 1393 |
+
"grad_norm": 34.51709660208177,
|
| 1394 |
+
"learning_rate": 9.64668657069706e-10,
|
| 1395 |
+
"logits/chosen": -2.295931339263916,
|
| 1396 |
+
"logits/rejected": -2.02848219871521,
|
| 1397 |
+
"logps/chosen": -1.3973679542541504,
|
| 1398 |
+
"logps/rejected": -3.212092638015747,
|
| 1399 |
+
"loss": -0.145,
|
| 1400 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1401 |
+
"rewards/chosen": -1.3973679542541504,
|
| 1402 |
+
"rewards/margins": 1.814724326133728,
|
| 1403 |
+
"rewards/rejected": -3.212092638015747,
|
| 1404 |
+
"step": 465
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.98403559277676,
|
| 1408 |
+
"grad_norm": 11.228558025822675,
|
| 1409 |
+
"learning_rate": 3.2839470889836627e-10,
|
| 1410 |
+
"logits/chosen": -2.295358180999756,
|
| 1411 |
+
"logits/rejected": -2.151076316833496,
|
| 1412 |
+
"logps/chosen": -1.4753532409667969,
|
| 1413 |
+
"logps/rejected": -2.6792643070220947,
|
| 1414 |
+
"loss": -0.131,
|
| 1415 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1416 |
+
"rewards/chosen": -1.4753532409667969,
|
| 1417 |
+
"rewards/margins": 1.2039110660552979,
|
| 1418 |
+
"rewards/rejected": -2.6792643070220947,
|
| 1419 |
+
"step": 470
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.9945040565297043,
|
| 1423 |
+
"grad_norm": 15.547524058275151,
|
| 1424 |
+
"learning_rate": 2.6813123097352287e-11,
|
| 1425 |
+
"logits/chosen": -2.3587679862976074,
|
| 1426 |
+
"logits/rejected": -2.1848464012145996,
|
| 1427 |
+
"logps/chosen": -1.336055874824524,
|
| 1428 |
+
"logps/rejected": -2.380079746246338,
|
| 1429 |
+
"loss": -0.1371,
|
| 1430 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1431 |
+
"rewards/chosen": -1.336055874824524,
|
| 1432 |
+
"rewards/margins": 1.044023871421814,
|
| 1433 |
+
"rewards/rejected": -2.380079746246338,
|
| 1434 |
+
"step": 475
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 0.998691442030882,
|
| 1438 |
+
"step": 477,
|
| 1439 |
+
"total_flos": 0.0,
|
| 1440 |
+
"train_loss": -0.1067421951134,
|
| 1441 |
+
"train_runtime": 20026.0763,
|
| 1442 |
+
"train_samples_per_second": 3.053,
|
| 1443 |
+
"train_steps_per_second": 0.024
|
| 1444 |
+
}
|
| 1445 |
+
],
|
| 1446 |
+
"logging_steps": 5,
|
| 1447 |
+
"max_steps": 477,
|
| 1448 |
+
"num_input_tokens_seen": 0,
|
| 1449 |
+
"num_train_epochs": 1,
|
| 1450 |
+
"save_steps": 239,
|
| 1451 |
+
"stateful_callbacks": {
|
| 1452 |
+
"TrainerControl": {
|
| 1453 |
+
"args": {
|
| 1454 |
+
"should_epoch_stop": false,
|
| 1455 |
+
"should_evaluate": false,
|
| 1456 |
+
"should_log": false,
|
| 1457 |
+
"should_save": true,
|
| 1458 |
+
"should_training_stop": false
|
| 1459 |
+
},
|
| 1460 |
+
"attributes": {}
|
| 1461 |
+
}
|
| 1462 |
+
},
|
| 1463 |
+
"total_flos": 0.0,
|
| 1464 |
+
"train_batch_size": 1,
|
| 1465 |
+
"trial_name": null,
|
| 1466 |
+
"trial_params": null
|
| 1467 |
+
}
|