Upload SFT/trainer_state.json with huggingface_hub
Browse files- SFT/trainer_state.json +2978 -0
SFT/trainer_state.json
ADDED
|
@@ -0,0 +1,2978 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.19655267894268036,
|
| 3 |
+
"best_model_checkpoint": "saves/Orca/SFT/checkpoint-2000",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 250,
|
| 6 |
+
"global_step": 2062,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.004849660523763337,
|
| 13 |
+
"grad_norm": 1.569652795791626,
|
| 14 |
+
"learning_rate": 8.333333333333333e-07,
|
| 15 |
+
"loss": 1.763,
|
| 16 |
+
"step": 5
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.009699321047526674,
|
| 20 |
+
"grad_norm": 1.4889137744903564,
|
| 21 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 22 |
+
"loss": 1.7673,
|
| 23 |
+
"step": 10
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.014548981571290009,
|
| 27 |
+
"grad_norm": 1.3860485553741455,
|
| 28 |
+
"learning_rate": 2.5e-06,
|
| 29 |
+
"loss": 1.7696,
|
| 30 |
+
"step": 15
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.019398642095053348,
|
| 34 |
+
"grad_norm": 1.4874159097671509,
|
| 35 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 36 |
+
"loss": 1.7723,
|
| 37 |
+
"step": 20
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02424830261881668,
|
| 41 |
+
"grad_norm": 1.4718151092529297,
|
| 42 |
+
"learning_rate": 4.166666666666667e-06,
|
| 43 |
+
"loss": 1.7295,
|
| 44 |
+
"step": 25
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.029097963142580018,
|
| 48 |
+
"grad_norm": 1.5262246131896973,
|
| 49 |
+
"learning_rate": 5e-06,
|
| 50 |
+
"loss": 1.7863,
|
| 51 |
+
"step": 30
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03394762366634336,
|
| 55 |
+
"grad_norm": 1.6720690727233887,
|
| 56 |
+
"learning_rate": 5.833333333333334e-06,
|
| 57 |
+
"loss": 1.709,
|
| 58 |
+
"step": 35
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.038797284190106696,
|
| 62 |
+
"grad_norm": 1.462561845779419,
|
| 63 |
+
"learning_rate": 6.666666666666667e-06,
|
| 64 |
+
"loss": 1.6664,
|
| 65 |
+
"step": 40
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.04364694471387003,
|
| 69 |
+
"grad_norm": 1.6730310916900635,
|
| 70 |
+
"learning_rate": 7.5e-06,
|
| 71 |
+
"loss": 1.6421,
|
| 72 |
+
"step": 45
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.04849660523763336,
|
| 76 |
+
"grad_norm": 1.538442850112915,
|
| 77 |
+
"learning_rate": 8.333333333333334e-06,
|
| 78 |
+
"loss": 1.6169,
|
| 79 |
+
"step": 50
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.0533462657613967,
|
| 83 |
+
"grad_norm": 1.7747137546539307,
|
| 84 |
+
"learning_rate": 9.166666666666666e-06,
|
| 85 |
+
"loss": 1.5235,
|
| 86 |
+
"step": 55
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.058195926285160036,
|
| 90 |
+
"grad_norm": 1.600160837173462,
|
| 91 |
+
"learning_rate": 1e-05,
|
| 92 |
+
"loss": 1.4392,
|
| 93 |
+
"step": 60
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.06304558680892337,
|
| 97 |
+
"grad_norm": 1.3897379636764526,
|
| 98 |
+
"learning_rate": 1.0833333333333334e-05,
|
| 99 |
+
"loss": 1.3295,
|
| 100 |
+
"step": 65
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.06789524733268672,
|
| 104 |
+
"grad_norm": 1.3479865789413452,
|
| 105 |
+
"learning_rate": 1.1666666666666668e-05,
|
| 106 |
+
"loss": 1.1758,
|
| 107 |
+
"step": 70
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.07274490785645005,
|
| 111 |
+
"grad_norm": 1.0993175506591797,
|
| 112 |
+
"learning_rate": 1.25e-05,
|
| 113 |
+
"loss": 1.1012,
|
| 114 |
+
"step": 75
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.07759456838021339,
|
| 118 |
+
"grad_norm": 1.0077180862426758,
|
| 119 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 120 |
+
"loss": 0.9917,
|
| 121 |
+
"step": 80
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.08244422890397672,
|
| 125 |
+
"grad_norm": 0.8057171106338501,
|
| 126 |
+
"learning_rate": 1.4166666666666668e-05,
|
| 127 |
+
"loss": 0.8798,
|
| 128 |
+
"step": 85
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.08729388942774007,
|
| 132 |
+
"grad_norm": 0.6777159571647644,
|
| 133 |
+
"learning_rate": 1.5e-05,
|
| 134 |
+
"loss": 0.7895,
|
| 135 |
+
"step": 90
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.0921435499515034,
|
| 139 |
+
"grad_norm": 0.5001732707023621,
|
| 140 |
+
"learning_rate": 1.5833333333333333e-05,
|
| 141 |
+
"loss": 0.7364,
|
| 142 |
+
"step": 95
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.09699321047526673,
|
| 146 |
+
"grad_norm": 0.4005896747112274,
|
| 147 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 148 |
+
"loss": 0.6936,
|
| 149 |
+
"step": 100
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.10184287099903007,
|
| 153 |
+
"grad_norm": 0.3276143968105316,
|
| 154 |
+
"learning_rate": 1.75e-05,
|
| 155 |
+
"loss": 0.6595,
|
| 156 |
+
"step": 105
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.1066925315227934,
|
| 160 |
+
"grad_norm": 0.29122063517570496,
|
| 161 |
+
"learning_rate": 1.8333333333333333e-05,
|
| 162 |
+
"loss": 0.6316,
|
| 163 |
+
"step": 110
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.11154219204655674,
|
| 167 |
+
"grad_norm": 0.2869010269641876,
|
| 168 |
+
"learning_rate": 1.9166666666666667e-05,
|
| 169 |
+
"loss": 0.6033,
|
| 170 |
+
"step": 115
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.11639185257032007,
|
| 174 |
+
"grad_norm": 0.25405701994895935,
|
| 175 |
+
"learning_rate": 2e-05,
|
| 176 |
+
"loss": 0.5716,
|
| 177 |
+
"step": 120
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.12124151309408342,
|
| 181 |
+
"grad_norm": 0.24002915620803833,
|
| 182 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 183 |
+
"loss": 0.5691,
|
| 184 |
+
"step": 125
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.12609117361784675,
|
| 188 |
+
"grad_norm": 0.2358996719121933,
|
| 189 |
+
"learning_rate": 2.1666666666666667e-05,
|
| 190 |
+
"loss": 0.554,
|
| 191 |
+
"step": 130
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.1309408341416101,
|
| 195 |
+
"grad_norm": 0.21090734004974365,
|
| 196 |
+
"learning_rate": 2.25e-05,
|
| 197 |
+
"loss": 0.5194,
|
| 198 |
+
"step": 135
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.13579049466537343,
|
| 202 |
+
"grad_norm": 0.20737610757350922,
|
| 203 |
+
"learning_rate": 2.3333333333333336e-05,
|
| 204 |
+
"loss": 0.4918,
|
| 205 |
+
"step": 140
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.14064015518913675,
|
| 209 |
+
"grad_norm": 0.1901807337999344,
|
| 210 |
+
"learning_rate": 2.4166666666666667e-05,
|
| 211 |
+
"loss": 0.4705,
|
| 212 |
+
"step": 145
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.1454898157129001,
|
| 216 |
+
"grad_norm": 0.19849984347820282,
|
| 217 |
+
"learning_rate": 2.5e-05,
|
| 218 |
+
"loss": 0.4595,
|
| 219 |
+
"step": 150
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.15033947623666344,
|
| 223 |
+
"grad_norm": 0.18864648044109344,
|
| 224 |
+
"learning_rate": 2.5833333333333336e-05,
|
| 225 |
+
"loss": 0.4407,
|
| 226 |
+
"step": 155
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.15518913676042678,
|
| 230 |
+
"grad_norm": 0.17015086114406586,
|
| 231 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 232 |
+
"loss": 0.4214,
|
| 233 |
+
"step": 160
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.1600387972841901,
|
| 237 |
+
"grad_norm": 0.17595945298671722,
|
| 238 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 239 |
+
"loss": 0.4343,
|
| 240 |
+
"step": 165
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.16488845780795344,
|
| 244 |
+
"grad_norm": 0.16629475355148315,
|
| 245 |
+
"learning_rate": 2.8333333333333335e-05,
|
| 246 |
+
"loss": 0.4045,
|
| 247 |
+
"step": 170
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.1697381183317168,
|
| 251 |
+
"grad_norm": 0.16392391920089722,
|
| 252 |
+
"learning_rate": 2.916666666666667e-05,
|
| 253 |
+
"loss": 0.3803,
|
| 254 |
+
"step": 175
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.17458777885548013,
|
| 258 |
+
"grad_norm": 0.1675371527671814,
|
| 259 |
+
"learning_rate": 3e-05,
|
| 260 |
+
"loss": 0.3762,
|
| 261 |
+
"step": 180
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.17943743937924345,
|
| 265 |
+
"grad_norm": 0.15912921726703644,
|
| 266 |
+
"learning_rate": 3.0833333333333335e-05,
|
| 267 |
+
"loss": 0.3698,
|
| 268 |
+
"step": 185
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.1842870999030068,
|
| 272 |
+
"grad_norm": 0.1546890288591385,
|
| 273 |
+
"learning_rate": 3.1666666666666666e-05,
|
| 274 |
+
"loss": 0.359,
|
| 275 |
+
"step": 190
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.18913676042677013,
|
| 279 |
+
"grad_norm": 0.14289158582687378,
|
| 280 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 281 |
+
"loss": 0.349,
|
| 282 |
+
"step": 195
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.19398642095053345,
|
| 286 |
+
"grad_norm": 0.13693265616893768,
|
| 287 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 288 |
+
"loss": 0.3145,
|
| 289 |
+
"step": 200
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.1988360814742968,
|
| 293 |
+
"grad_norm": 0.15631736814975739,
|
| 294 |
+
"learning_rate": 3.4166666666666666e-05,
|
| 295 |
+
"loss": 0.3266,
|
| 296 |
+
"step": 205
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.20368574199806014,
|
| 300 |
+
"grad_norm": 0.13729828596115112,
|
| 301 |
+
"learning_rate": 3.5e-05,
|
| 302 |
+
"loss": 0.3368,
|
| 303 |
+
"step": 210
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.20853540252182348,
|
| 307 |
+
"grad_norm": 0.13479730486869812,
|
| 308 |
+
"learning_rate": 3.5833333333333335e-05,
|
| 309 |
+
"loss": 0.3025,
|
| 310 |
+
"step": 215
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.2133850630455868,
|
| 314 |
+
"grad_norm": 0.13599981367588043,
|
| 315 |
+
"learning_rate": 3.6666666666666666e-05,
|
| 316 |
+
"loss": 0.3121,
|
| 317 |
+
"step": 220
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.21823472356935014,
|
| 321 |
+
"grad_norm": 0.1292022317647934,
|
| 322 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 323 |
+
"loss": 0.2982,
|
| 324 |
+
"step": 225
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.22308438409311349,
|
| 328 |
+
"grad_norm": 0.13855475187301636,
|
| 329 |
+
"learning_rate": 3.8333333333333334e-05,
|
| 330 |
+
"loss": 0.2709,
|
| 331 |
+
"step": 230
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.22793404461687683,
|
| 335 |
+
"grad_norm": 0.12373723834753036,
|
| 336 |
+
"learning_rate": 3.9166666666666665e-05,
|
| 337 |
+
"loss": 0.2903,
|
| 338 |
+
"step": 235
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.23278370514064015,
|
| 342 |
+
"grad_norm": 0.1207897961139679,
|
| 343 |
+
"learning_rate": 4e-05,
|
| 344 |
+
"loss": 0.3064,
|
| 345 |
+
"step": 240
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.2376333656644035,
|
| 349 |
+
"grad_norm": 0.12190013378858566,
|
| 350 |
+
"learning_rate": 4.0833333333333334e-05,
|
| 351 |
+
"loss": 0.3189,
|
| 352 |
+
"step": 245
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.24248302618816683,
|
| 356 |
+
"grad_norm": 0.11036136001348495,
|
| 357 |
+
"learning_rate": 4.166666666666667e-05,
|
| 358 |
+
"loss": 0.2823,
|
| 359 |
+
"step": 250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.24248302618816683,
|
| 363 |
+
"eval_loss": 0.2783246636390686,
|
| 364 |
+
"eval_runtime": 48.1034,
|
| 365 |
+
"eval_samples_per_second": 14.011,
|
| 366 |
+
"eval_steps_per_second": 0.894,
|
| 367 |
+
"step": 250
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.24733268671193018,
|
| 371 |
+
"grad_norm": 0.11338551342487335,
|
| 372 |
+
"learning_rate": 4.25e-05,
|
| 373 |
+
"loss": 0.2865,
|
| 374 |
+
"step": 255
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.2521823472356935,
|
| 378 |
+
"grad_norm": 0.1275234967470169,
|
| 379 |
+
"learning_rate": 4.3333333333333334e-05,
|
| 380 |
+
"loss": 0.279,
|
| 381 |
+
"step": 260
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.25703200775945684,
|
| 385 |
+
"grad_norm": 0.1382416933774948,
|
| 386 |
+
"learning_rate": 4.4166666666666665e-05,
|
| 387 |
+
"loss": 0.2578,
|
| 388 |
+
"step": 265
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.2618816682832202,
|
| 392 |
+
"grad_norm": 0.13280175626277924,
|
| 393 |
+
"learning_rate": 4.5e-05,
|
| 394 |
+
"loss": 0.3027,
|
| 395 |
+
"step": 270
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.2667313288069835,
|
| 399 |
+
"grad_norm": 0.11413029581308365,
|
| 400 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 401 |
+
"loss": 0.2647,
|
| 402 |
+
"step": 275
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.27158098933074687,
|
| 406 |
+
"grad_norm": 0.12923401594161987,
|
| 407 |
+
"learning_rate": 4.666666666666667e-05,
|
| 408 |
+
"loss": 0.2984,
|
| 409 |
+
"step": 280
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.27643064985451016,
|
| 413 |
+
"grad_norm": 0.11887092143297195,
|
| 414 |
+
"learning_rate": 4.75e-05,
|
| 415 |
+
"loss": 0.2683,
|
| 416 |
+
"step": 285
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.2812803103782735,
|
| 420 |
+
"grad_norm": 0.12000833451747894,
|
| 421 |
+
"learning_rate": 4.8333333333333334e-05,
|
| 422 |
+
"loss": 0.2747,
|
| 423 |
+
"step": 290
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.28612997090203685,
|
| 427 |
+
"grad_norm": 0.11771322786808014,
|
| 428 |
+
"learning_rate": 4.9166666666666665e-05,
|
| 429 |
+
"loss": 0.233,
|
| 430 |
+
"step": 295
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.2909796314258002,
|
| 434 |
+
"grad_norm": 0.10835352540016174,
|
| 435 |
+
"learning_rate": 5e-05,
|
| 436 |
+
"loss": 0.2444,
|
| 437 |
+
"step": 300
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.29582929194956353,
|
| 441 |
+
"grad_norm": 0.12001065164804459,
|
| 442 |
+
"learning_rate": 4.999900657500545e-05,
|
| 443 |
+
"loss": 0.2234,
|
| 444 |
+
"step": 305
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.3006789524733269,
|
| 448 |
+
"grad_norm": 0.11877791583538055,
|
| 449 |
+
"learning_rate": 4.9996026378973266e-05,
|
| 450 |
+
"loss": 0.2467,
|
| 451 |
+
"step": 310
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.3055286129970902,
|
| 455 |
+
"grad_norm": 0.12458127737045288,
|
| 456 |
+
"learning_rate": 4.999105964875153e-05,
|
| 457 |
+
"loss": 0.2559,
|
| 458 |
+
"step": 315
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.31037827352085356,
|
| 462 |
+
"grad_norm": 0.11346752941608429,
|
| 463 |
+
"learning_rate": 4.9984106779066174e-05,
|
| 464 |
+
"loss": 0.2523,
|
| 465 |
+
"step": 320
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.31522793404461685,
|
| 469 |
+
"grad_norm": 0.13425442576408386,
|
| 470 |
+
"learning_rate": 4.9975168322489554e-05,
|
| 471 |
+
"loss": 0.2697,
|
| 472 |
+
"step": 325
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.3200775945683802,
|
| 476 |
+
"grad_norm": 0.11504397541284561,
|
| 477 |
+
"learning_rate": 4.996424498939656e-05,
|
| 478 |
+
"loss": 0.2553,
|
| 479 |
+
"step": 330
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.32492725509214354,
|
| 483 |
+
"grad_norm": 0.12158407270908356,
|
| 484 |
+
"learning_rate": 4.9951337647908165e-05,
|
| 485 |
+
"loss": 0.2477,
|
| 486 |
+
"step": 335
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.3297769156159069,
|
| 490 |
+
"grad_norm": 0.12656843662261963,
|
| 491 |
+
"learning_rate": 4.9936447323822424e-05,
|
| 492 |
+
"loss": 0.2328,
|
| 493 |
+
"step": 340
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.33462657613967023,
|
| 497 |
+
"grad_norm": 0.1163104996085167,
|
| 498 |
+
"learning_rate": 4.991957520053294e-05,
|
| 499 |
+
"loss": 0.2509,
|
| 500 |
+
"step": 345
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.3394762366634336,
|
| 504 |
+
"grad_norm": 0.12447489798069,
|
| 505 |
+
"learning_rate": 4.990072261893484e-05,
|
| 506 |
+
"loss": 0.2589,
|
| 507 |
+
"step": 350
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.3443258971871969,
|
| 511 |
+
"grad_norm": 0.11267515271902084,
|
| 512 |
+
"learning_rate": 4.9879891077318176e-05,
|
| 513 |
+
"loss": 0.2446,
|
| 514 |
+
"step": 355
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.34917555771096026,
|
| 518 |
+
"grad_norm": 0.13075344264507294,
|
| 519 |
+
"learning_rate": 4.985708223124888e-05,
|
| 520 |
+
"loss": 0.2587,
|
| 521 |
+
"step": 360
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.35402521823472355,
|
| 525 |
+
"grad_norm": 0.10673461854457855,
|
| 526 |
+
"learning_rate": 4.9832297893437186e-05,
|
| 527 |
+
"loss": 0.2457,
|
| 528 |
+
"step": 365
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.3588748787584869,
|
| 532 |
+
"grad_norm": 0.12786507606506348,
|
| 533 |
+
"learning_rate": 4.980554003359354e-05,
|
| 534 |
+
"loss": 0.2369,
|
| 535 |
+
"step": 370
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.36372453928225024,
|
| 539 |
+
"grad_norm": 0.1437879055738449,
|
| 540 |
+
"learning_rate": 4.9776810778272075e-05,
|
| 541 |
+
"loss": 0.2736,
|
| 542 |
+
"step": 375
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.3685741998060136,
|
| 546 |
+
"grad_norm": 0.12847022712230682,
|
| 547 |
+
"learning_rate": 4.9746112410701625e-05,
|
| 548 |
+
"loss": 0.2252,
|
| 549 |
+
"step": 380
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.3734238603297769,
|
| 553 |
+
"grad_norm": 0.11457338184118271,
|
| 554 |
+
"learning_rate": 4.9713447370604236e-05,
|
| 555 |
+
"loss": 0.2508,
|
| 556 |
+
"step": 385
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.37827352085354027,
|
| 560 |
+
"grad_norm": 0.11042597144842148,
|
| 561 |
+
"learning_rate": 4.967881825400129e-05,
|
| 562 |
+
"loss": 0.2306,
|
| 563 |
+
"step": 390
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.3831231813773036,
|
| 567 |
+
"grad_norm": 0.12592245638370514,
|
| 568 |
+
"learning_rate": 4.964222781300719e-05,
|
| 569 |
+
"loss": 0.2599,
|
| 570 |
+
"step": 395
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.3879728419010669,
|
| 574 |
+
"grad_norm": 0.11546192318201065,
|
| 575 |
+
"learning_rate": 4.960367895561063e-05,
|
| 576 |
+
"loss": 0.2225,
|
| 577 |
+
"step": 400
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.39282250242483024,
|
| 581 |
+
"grad_norm": 0.12539242208003998,
|
| 582 |
+
"learning_rate": 4.956317474544348e-05,
|
| 583 |
+
"loss": 0.2432,
|
| 584 |
+
"step": 405
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.3976721629485936,
|
| 588 |
+
"grad_norm": 0.10763729363679886,
|
| 589 |
+
"learning_rate": 4.952071840153732e-05,
|
| 590 |
+
"loss": 0.2307,
|
| 591 |
+
"step": 410
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.40252182347235693,
|
| 595 |
+
"grad_norm": 0.11994805932044983,
|
| 596 |
+
"learning_rate": 4.947631329806761e-05,
|
| 597 |
+
"loss": 0.2282,
|
| 598 |
+
"step": 415
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.4073714839961203,
|
| 602 |
+
"grad_norm": 0.13727760314941406,
|
| 603 |
+
"learning_rate": 4.9429962964085517e-05,
|
| 604 |
+
"loss": 0.2225,
|
| 605 |
+
"step": 420
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.4122211445198836,
|
| 609 |
+
"grad_norm": 0.1300915628671646,
|
| 610 |
+
"learning_rate": 4.9381671083237476e-05,
|
| 611 |
+
"loss": 0.2393,
|
| 612 |
+
"step": 425
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.41707080504364696,
|
| 616 |
+
"grad_norm": 0.12672416865825653,
|
| 617 |
+
"learning_rate": 4.9331441493472395e-05,
|
| 618 |
+
"loss": 0.2296,
|
| 619 |
+
"step": 430
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.4219204655674103,
|
| 623 |
+
"grad_norm": 0.13302814960479736,
|
| 624 |
+
"learning_rate": 4.9279278186736674e-05,
|
| 625 |
+
"loss": 0.2242,
|
| 626 |
+
"step": 435
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.4267701260911736,
|
| 630 |
+
"grad_norm": 0.12973704934120178,
|
| 631 |
+
"learning_rate": 4.922518530865693e-05,
|
| 632 |
+
"loss": 0.2329,
|
| 633 |
+
"step": 440
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.43161978661493694,
|
| 637 |
+
"grad_norm": 0.12950606644153595,
|
| 638 |
+
"learning_rate": 4.916916715821053e-05,
|
| 639 |
+
"loss": 0.2132,
|
| 640 |
+
"step": 445
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.4364694471387003,
|
| 644 |
+
"grad_norm": 0.12280883640050888,
|
| 645 |
+
"learning_rate": 4.911122818738394e-05,
|
| 646 |
+
"loss": 0.2255,
|
| 647 |
+
"step": 450
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.44131910766246363,
|
| 651 |
+
"grad_norm": 0.12878793478012085,
|
| 652 |
+
"learning_rate": 4.9051373000818887e-05,
|
| 653 |
+
"loss": 0.2268,
|
| 654 |
+
"step": 455
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.44616876818622697,
|
| 658 |
+
"grad_norm": 0.12943603098392487,
|
| 659 |
+
"learning_rate": 4.898960635544647e-05,
|
| 660 |
+
"loss": 0.2145,
|
| 661 |
+
"step": 460
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.4510184287099903,
|
| 665 |
+
"grad_norm": 0.13643978536128998,
|
| 666 |
+
"learning_rate": 4.8925933160109016e-05,
|
| 667 |
+
"loss": 0.2197,
|
| 668 |
+
"step": 465
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.45586808923375366,
|
| 672 |
+
"grad_norm": 0.1563604176044464,
|
| 673 |
+
"learning_rate": 4.886035847517003e-05,
|
| 674 |
+
"loss": 0.2449,
|
| 675 |
+
"step": 470
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.46071774975751695,
|
| 679 |
+
"grad_norm": 0.13257759809494019,
|
| 680 |
+
"learning_rate": 4.879288751211199e-05,
|
| 681 |
+
"loss": 0.2307,
|
| 682 |
+
"step": 475
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.4655674102812803,
|
| 686 |
+
"grad_norm": 0.13240696489810944,
|
| 687 |
+
"learning_rate": 4.8723525633122193e-05,
|
| 688 |
+
"loss": 0.2136,
|
| 689 |
+
"step": 480
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.47041707080504364,
|
| 693 |
+
"grad_norm": 0.14693856239318848,
|
| 694 |
+
"learning_rate": 4.8652278350666566e-05,
|
| 695 |
+
"loss": 0.2388,
|
| 696 |
+
"step": 485
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.475266731328807,
|
| 700 |
+
"grad_norm": 0.14097219705581665,
|
| 701 |
+
"learning_rate": 4.857915132705162e-05,
|
| 702 |
+
"loss": 0.2178,
|
| 703 |
+
"step": 490
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.4801163918525703,
|
| 707 |
+
"grad_norm": 0.1185833290219307,
|
| 708 |
+
"learning_rate": 4.850415037397439e-05,
|
| 709 |
+
"loss": 0.2283,
|
| 710 |
+
"step": 495
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.48496605237633367,
|
| 714 |
+
"grad_norm": 0.1279655396938324,
|
| 715 |
+
"learning_rate": 4.8427281452060594e-05,
|
| 716 |
+
"loss": 0.2186,
|
| 717 |
+
"step": 500
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.48496605237633367,
|
| 721 |
+
"eval_loss": 0.2229408472776413,
|
| 722 |
+
"eval_runtime": 48.0836,
|
| 723 |
+
"eval_samples_per_second": 14.017,
|
| 724 |
+
"eval_steps_per_second": 0.894,
|
| 725 |
+
"step": 500
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 0.489815712900097,
|
| 729 |
+
"grad_norm": 0.14021804928779602,
|
| 730 |
+
"learning_rate": 4.834855067039088e-05,
|
| 731 |
+
"loss": 0.227,
|
| 732 |
+
"step": 505
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.49466537342386036,
|
| 736 |
+
"grad_norm": 0.13102716207504272,
|
| 737 |
+
"learning_rate": 4.826796428601538e-05,
|
| 738 |
+
"loss": 0.2302,
|
| 739 |
+
"step": 510
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.49951503394762364,
|
| 743 |
+
"grad_norm": 0.11888754367828369,
|
| 744 |
+
"learning_rate": 4.818552870345635e-05,
|
| 745 |
+
"loss": 0.2245,
|
| 746 |
+
"step": 515
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.504364694471387,
|
| 750 |
+
"grad_norm": 0.13540378212928772,
|
| 751 |
+
"learning_rate": 4.810125047419926e-05,
|
| 752 |
+
"loss": 0.233,
|
| 753 |
+
"step": 520
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.5092143549951503,
|
| 757 |
+
"grad_norm": 0.12920723855495453,
|
| 758 |
+
"learning_rate": 4.8015136296172055e-05,
|
| 759 |
+
"loss": 0.2225,
|
| 760 |
+
"step": 525
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 0.5140640155189137,
|
| 764 |
+
"grad_norm": 0.12302406132221222,
|
| 765 |
+
"learning_rate": 4.792719301321289e-05,
|
| 766 |
+
"loss": 0.2155,
|
| 767 |
+
"step": 530
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.518913676042677,
|
| 771 |
+
"grad_norm": 0.12846626341342926,
|
| 772 |
+
"learning_rate": 4.783742761452619e-05,
|
| 773 |
+
"loss": 0.2285,
|
| 774 |
+
"step": 535
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5237633365664404,
|
| 778 |
+
"grad_norm": 0.12521494925022125,
|
| 779 |
+
"learning_rate": 4.7745847234127206e-05,
|
| 780 |
+
"loss": 0.2112,
|
| 781 |
+
"step": 540
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 0.5286129970902037,
|
| 785 |
+
"grad_norm": 0.12830866873264313,
|
| 786 |
+
"learning_rate": 4.7652459150275055e-05,
|
| 787 |
+
"loss": 0.2246,
|
| 788 |
+
"step": 545
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.533462657613967,
|
| 792 |
+
"grad_norm": 0.14525972306728363,
|
| 793 |
+
"learning_rate": 4.7557270784894276e-05,
|
| 794 |
+
"loss": 0.2293,
|
| 795 |
+
"step": 550
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.5383123181377304,
|
| 799 |
+
"grad_norm": 0.14833694696426392,
|
| 800 |
+
"learning_rate": 4.746028970298497e-05,
|
| 801 |
+
"loss": 0.2366,
|
| 802 |
+
"step": 555
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.5431619786614937,
|
| 806 |
+
"grad_norm": 0.13993489742279053,
|
| 807 |
+
"learning_rate": 4.7361523612021616e-05,
|
| 808 |
+
"loss": 0.2456,
|
| 809 |
+
"step": 560
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.5480116391852571,
|
| 813 |
+
"grad_norm": 0.125143900513649,
|
| 814 |
+
"learning_rate": 4.726098036134046e-05,
|
| 815 |
+
"loss": 0.2365,
|
| 816 |
+
"step": 565
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 0.5528612997090203,
|
| 820 |
+
"grad_norm": 0.1390179991722107,
|
| 821 |
+
"learning_rate": 4.715866794151578e-05,
|
| 822 |
+
"loss": 0.233,
|
| 823 |
+
"step": 570
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.5577109602327837,
|
| 827 |
+
"grad_norm": 0.1502588987350464,
|
| 828 |
+
"learning_rate": 4.705459448372478e-05,
|
| 829 |
+
"loss": 0.234,
|
| 830 |
+
"step": 575
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.562560620756547,
|
| 834 |
+
"grad_norm": 0.1441139429807663,
|
| 835 |
+
"learning_rate": 4.6948768259101394e-05,
|
| 836 |
+
"loss": 0.2085,
|
| 837 |
+
"step": 580
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.5674102812803103,
|
| 841 |
+
"grad_norm": 0.136013925075531,
|
| 842 |
+
"learning_rate": 4.6841197678078965e-05,
|
| 843 |
+
"loss": 0.1938,
|
| 844 |
+
"step": 585
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 0.5722599418040737,
|
| 848 |
+
"grad_norm": 0.12127859890460968,
|
| 849 |
+
"learning_rate": 4.6731891289721786e-05,
|
| 850 |
+
"loss": 0.2009,
|
| 851 |
+
"step": 590
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 0.577109602327837,
|
| 855 |
+
"grad_norm": 0.14969481527805328,
|
| 856 |
+
"learning_rate": 4.6620857781045715e-05,
|
| 857 |
+
"loss": 0.2071,
|
| 858 |
+
"step": 595
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.5819592628516004,
|
| 862 |
+
"grad_norm": 0.13548845052719116,
|
| 863 |
+
"learning_rate": 4.65081059763278e-05,
|
| 864 |
+
"loss": 0.2265,
|
| 865 |
+
"step": 600
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.5868089233753637,
|
| 869 |
+
"grad_norm": 0.12766653299331665,
|
| 870 |
+
"learning_rate": 4.639364483640489e-05,
|
| 871 |
+
"loss": 0.2212,
|
| 872 |
+
"step": 605
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.5916585838991271,
|
| 876 |
+
"grad_norm": 0.13482384383678436,
|
| 877 |
+
"learning_rate": 4.627748345796158e-05,
|
| 878 |
+
"loss": 0.2145,
|
| 879 |
+
"step": 610
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.5965082444228904,
|
| 883 |
+
"grad_norm": 0.13684765994548798,
|
| 884 |
+
"learning_rate": 4.615963107280722e-05,
|
| 885 |
+
"loss": 0.2097,
|
| 886 |
+
"step": 615
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 0.6013579049466538,
|
| 890 |
+
"grad_norm": 0.1494571417570114,
|
| 891 |
+
"learning_rate": 4.604009704714219e-05,
|
| 892 |
+
"loss": 0.2282,
|
| 893 |
+
"step": 620
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 0.6062075654704171,
|
| 897 |
+
"grad_norm": 0.14703498780727386,
|
| 898 |
+
"learning_rate": 4.591889088081363e-05,
|
| 899 |
+
"loss": 0.1963,
|
| 900 |
+
"step": 625
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 0.6110572259941804,
|
| 904 |
+
"grad_norm": 0.151000514626503,
|
| 905 |
+
"learning_rate": 4.579602220656032e-05,
|
| 906 |
+
"loss": 0.2267,
|
| 907 |
+
"step": 630
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 0.6159068865179438,
|
| 911 |
+
"grad_norm": 0.1505574882030487,
|
| 912 |
+
"learning_rate": 4.567150078924723e-05,
|
| 913 |
+
"loss": 0.2298,
|
| 914 |
+
"step": 635
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.6207565470417071,
|
| 918 |
+
"grad_norm": 0.1409265249967575,
|
| 919 |
+
"learning_rate": 4.5545336525089444e-05,
|
| 920 |
+
"loss": 0.2051,
|
| 921 |
+
"step": 640
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.6256062075654704,
|
| 925 |
+
"grad_norm": 0.14167818427085876,
|
| 926 |
+
"learning_rate": 4.5417539440865616e-05,
|
| 927 |
+
"loss": 0.2302,
|
| 928 |
+
"step": 645
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.6304558680892337,
|
| 932 |
+
"grad_norm": 0.14084312319755554,
|
| 933 |
+
"learning_rate": 4.528811969312117e-05,
|
| 934 |
+
"loss": 0.2164,
|
| 935 |
+
"step": 650
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 0.635305528612997,
|
| 939 |
+
"grad_norm": 0.14464280009269714,
|
| 940 |
+
"learning_rate": 4.515708756736108e-05,
|
| 941 |
+
"loss": 0.1948,
|
| 942 |
+
"step": 655
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 0.6401551891367604,
|
| 946 |
+
"grad_norm": 0.141952782869339,
|
| 947 |
+
"learning_rate": 4.5024453477232444e-05,
|
| 948 |
+
"loss": 0.2193,
|
| 949 |
+
"step": 660
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.6450048496605237,
|
| 953 |
+
"grad_norm": 0.1446189135313034,
|
| 954 |
+
"learning_rate": 4.4890227963696895e-05,
|
| 955 |
+
"loss": 0.2141,
|
| 956 |
+
"step": 665
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.6498545101842871,
|
| 960 |
+
"grad_norm": 0.15042971074581146,
|
| 961 |
+
"learning_rate": 4.4754421694192835e-05,
|
| 962 |
+
"loss": 0.2125,
|
| 963 |
+
"step": 670
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.6547041707080504,
|
| 967 |
+
"grad_norm": 0.15916042029857635,
|
| 968 |
+
"learning_rate": 4.4617045461787665e-05,
|
| 969 |
+
"loss": 0.2215,
|
| 970 |
+
"step": 675
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 0.6595538312318138,
|
| 974 |
+
"grad_norm": 0.15135568380355835,
|
| 975 |
+
"learning_rate": 4.447811018432002e-05,
|
| 976 |
+
"loss": 0.1974,
|
| 977 |
+
"step": 680
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.6644034917555771,
|
| 981 |
+
"grad_norm": 0.12838397920131683,
|
| 982 |
+
"learning_rate": 4.4337626903532076e-05,
|
| 983 |
+
"loss": 0.2231,
|
| 984 |
+
"step": 685
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.6692531522793405,
|
| 988 |
+
"grad_norm": 0.145597442984581,
|
| 989 |
+
"learning_rate": 4.419560678419203e-05,
|
| 990 |
+
"loss": 0.2193,
|
| 991 |
+
"step": 690
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 0.6741028128031038,
|
| 995 |
+
"grad_norm": 0.14090026915073395,
|
| 996 |
+
"learning_rate": 4.40520611132068e-05,
|
| 997 |
+
"loss": 0.2196,
|
| 998 |
+
"step": 695
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.6789524733268671,
|
| 1002 |
+
"grad_norm": 0.1343528777360916,
|
| 1003 |
+
"learning_rate": 4.390700129872497e-05,
|
| 1004 |
+
"loss": 0.1998,
|
| 1005 |
+
"step": 700
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.6838021338506305,
|
| 1009 |
+
"grad_norm": 0.14349782466888428,
|
| 1010 |
+
"learning_rate": 4.376043886923015e-05,
|
| 1011 |
+
"loss": 0.2073,
|
| 1012 |
+
"step": 705
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 0.6886517943743938,
|
| 1016 |
+
"grad_norm": 0.15432004630565643,
|
| 1017 |
+
"learning_rate": 4.361238547262484e-05,
|
| 1018 |
+
"loss": 0.2184,
|
| 1019 |
+
"step": 710
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 0.6935014548981572,
|
| 1023 |
+
"grad_norm": 0.13481660187244415,
|
| 1024 |
+
"learning_rate": 4.346285287530458e-05,
|
| 1025 |
+
"loss": 0.2248,
|
| 1026 |
+
"step": 715
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 0.6983511154219205,
|
| 1030 |
+
"grad_norm": 0.1398806869983673,
|
| 1031 |
+
"learning_rate": 4.3311852961222966e-05,
|
| 1032 |
+
"loss": 0.2384,
|
| 1033 |
+
"step": 720
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.7032007759456838,
|
| 1037 |
+
"grad_norm": 0.13174192607402802,
|
| 1038 |
+
"learning_rate": 4.315939773094709e-05,
|
| 1039 |
+
"loss": 0.213,
|
| 1040 |
+
"step": 725
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.7080504364694471,
|
| 1044 |
+
"grad_norm": 0.12232716381549835,
|
| 1045 |
+
"learning_rate": 4.300549930070387e-05,
|
| 1046 |
+
"loss": 0.2283,
|
| 1047 |
+
"step": 730
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 0.7129000969932104,
|
| 1051 |
+
"grad_norm": 0.14730586111545563,
|
| 1052 |
+
"learning_rate": 4.2850169901417084e-05,
|
| 1053 |
+
"loss": 0.2098,
|
| 1054 |
+
"step": 735
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.7177497575169738,
|
| 1058 |
+
"grad_norm": 0.13813212513923645,
|
| 1059 |
+
"learning_rate": 4.269342187773532e-05,
|
| 1060 |
+
"loss": 0.2062,
|
| 1061 |
+
"step": 740
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 0.7225994180407371,
|
| 1065 |
+
"grad_norm": 0.13183937966823578,
|
| 1066 |
+
"learning_rate": 4.253526768705097e-05,
|
| 1067 |
+
"loss": 0.2034,
|
| 1068 |
+
"step": 745
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.7274490785645005,
|
| 1072 |
+
"grad_norm": 0.14278775453567505,
|
| 1073 |
+
"learning_rate": 4.237571989851011e-05,
|
| 1074 |
+
"loss": 0.2227,
|
| 1075 |
+
"step": 750
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.7274490785645005,
|
| 1079 |
+
"eval_loss": 0.21041834354400635,
|
| 1080 |
+
"eval_runtime": 48.0966,
|
| 1081 |
+
"eval_samples_per_second": 14.013,
|
| 1082 |
+
"eval_steps_per_second": 0.894,
|
| 1083 |
+
"step": 750
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 0.7322987390882638,
|
| 1087 |
+
"grad_norm": 0.16159804165363312,
|
| 1088 |
+
"learning_rate": 4.221479119201362e-05,
|
| 1089 |
+
"loss": 0.2191,
|
| 1090 |
+
"step": 755
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.7371483996120272,
|
| 1094 |
+
"grad_norm": 0.13423971831798553,
|
| 1095 |
+
"learning_rate": 4.205249435720943e-05,
|
| 1096 |
+
"loss": 0.1993,
|
| 1097 |
+
"step": 760
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 0.7419980601357905,
|
| 1101 |
+
"grad_norm": 0.14568765461444855,
|
| 1102 |
+
"learning_rate": 4.1888842292476126e-05,
|
| 1103 |
+
"loss": 0.2023,
|
| 1104 |
+
"step": 765
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.7468477206595538,
|
| 1108 |
+
"grad_norm": 0.13105235993862152,
|
| 1109 |
+
"learning_rate": 4.172384800389784e-05,
|
| 1110 |
+
"loss": 0.2048,
|
| 1111 |
+
"step": 770
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 0.7516973811833172,
|
| 1115 |
+
"grad_norm": 0.14081120491027832,
|
| 1116 |
+
"learning_rate": 4.155752460423059e-05,
|
| 1117 |
+
"loss": 0.2078,
|
| 1118 |
+
"step": 775
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.7565470417070805,
|
| 1122 |
+
"grad_norm": 0.13213010132312775,
|
| 1123 |
+
"learning_rate": 4.138988531186016e-05,
|
| 1124 |
+
"loss": 0.2183,
|
| 1125 |
+
"step": 780
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.7613967022308439,
|
| 1129 |
+
"grad_norm": 0.1357748806476593,
|
| 1130 |
+
"learning_rate": 4.1220943449751606e-05,
|
| 1131 |
+
"loss": 0.2016,
|
| 1132 |
+
"step": 785
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 0.7662463627546072,
|
| 1136 |
+
"grad_norm": 0.13532240688800812,
|
| 1137 |
+
"learning_rate": 4.105071244439039e-05,
|
| 1138 |
+
"loss": 0.2113,
|
| 1139 |
+
"step": 790
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 0.7710960232783706,
|
| 1143 |
+
"grad_norm": 0.14828477799892426,
|
| 1144 |
+
"learning_rate": 4.0879205824715384e-05,
|
| 1145 |
+
"loss": 0.2159,
|
| 1146 |
+
"step": 795
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.7759456838021338,
|
| 1150 |
+
"grad_norm": 0.1488686501979828,
|
| 1151 |
+
"learning_rate": 4.070643722104358e-05,
|
| 1152 |
+
"loss": 0.2124,
|
| 1153 |
+
"step": 800
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.7807953443258971,
|
| 1157 |
+
"grad_norm": 0.1575794368982315,
|
| 1158 |
+
"learning_rate": 4.053242036398692e-05,
|
| 1159 |
+
"loss": 0.2065,
|
| 1160 |
+
"step": 805
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.7856450048496605,
|
| 1164 |
+
"grad_norm": 0.13456198573112488,
|
| 1165 |
+
"learning_rate": 4.035716908336102e-05,
|
| 1166 |
+
"loss": 0.201,
|
| 1167 |
+
"step": 810
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.7904946653734238,
|
| 1171 |
+
"grad_norm": 0.13958740234375,
|
| 1172 |
+
"learning_rate": 4.01806973070861e-05,
|
| 1173 |
+
"loss": 0.2093,
|
| 1174 |
+
"step": 815
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 0.7953443258971872,
|
| 1178 |
+
"grad_norm": 0.14714553952217102,
|
| 1179 |
+
"learning_rate": 4.000301906008001e-05,
|
| 1180 |
+
"loss": 0.2082,
|
| 1181 |
+
"step": 820
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 0.8001939864209505,
|
| 1185 |
+
"grad_norm": 0.1589927226305008,
|
| 1186 |
+
"learning_rate": 3.9824148463143685e-05,
|
| 1187 |
+
"loss": 0.2279,
|
| 1188 |
+
"step": 825
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 0.8050436469447139,
|
| 1192 |
+
"grad_norm": 0.15132376551628113,
|
| 1193 |
+
"learning_rate": 3.964409973183886e-05,
|
| 1194 |
+
"loss": 0.204,
|
| 1195 |
+
"step": 830
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 0.8098933074684772,
|
| 1199 |
+
"grad_norm": 0.156170055270195,
|
| 1200 |
+
"learning_rate": 3.946288717535833e-05,
|
| 1201 |
+
"loss": 0.2051,
|
| 1202 |
+
"step": 835
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.8147429679922406,
|
| 1206 |
+
"grad_norm": 0.12180455029010773,
|
| 1207 |
+
"learning_rate": 3.928052519538874e-05,
|
| 1208 |
+
"loss": 0.2064,
|
| 1209 |
+
"step": 840
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8195926285160039,
|
| 1213 |
+
"grad_norm": 0.1337326467037201,
|
| 1214 |
+
"learning_rate": 3.9097028284966e-05,
|
| 1215 |
+
"loss": 0.2025,
|
| 1216 |
+
"step": 845
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.8244422890397672,
|
| 1220 |
+
"grad_norm": 0.13768617808818817,
|
| 1221 |
+
"learning_rate": 3.891241102732348e-05,
|
| 1222 |
+
"loss": 0.2037,
|
| 1223 |
+
"step": 850
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 0.8292919495635306,
|
| 1227 |
+
"grad_norm": 0.15049096941947937,
|
| 1228 |
+
"learning_rate": 3.872668809473304e-05,
|
| 1229 |
+
"loss": 0.2028,
|
| 1230 |
+
"step": 855
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 0.8341416100872939,
|
| 1234 |
+
"grad_norm": 0.14802470803260803,
|
| 1235 |
+
"learning_rate": 3.8539874247338956e-05,
|
| 1236 |
+
"loss": 0.2137,
|
| 1237 |
+
"step": 860
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.8389912706110573,
|
| 1241 |
+
"grad_norm": 0.13950683176517487,
|
| 1242 |
+
"learning_rate": 3.835198433198484e-05,
|
| 1243 |
+
"loss": 0.2375,
|
| 1244 |
+
"step": 865
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.8438409311348206,
|
| 1248 |
+
"grad_norm": 0.13893653452396393,
|
| 1249 |
+
"learning_rate": 3.816303328103374e-05,
|
| 1250 |
+
"loss": 0.204,
|
| 1251 |
+
"step": 870
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.8486905916585838,
|
| 1255 |
+
"grad_norm": 0.1808776557445526,
|
| 1256 |
+
"learning_rate": 3.7973036111181405e-05,
|
| 1257 |
+
"loss": 0.2043,
|
| 1258 |
+
"step": 875
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.8535402521823472,
|
| 1262 |
+
"grad_norm": 0.1470966786146164,
|
| 1263 |
+
"learning_rate": 3.7782007922262827e-05,
|
| 1264 |
+
"loss": 0.2117,
|
| 1265 |
+
"step": 880
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 0.8583899127061105,
|
| 1269 |
+
"grad_norm": 0.15209127962589264,
|
| 1270 |
+
"learning_rate": 3.758996389605222e-05,
|
| 1271 |
+
"loss": 0.2187,
|
| 1272 |
+
"step": 885
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 0.8632395732298739,
|
| 1276 |
+
"grad_norm": 0.1539296954870224,
|
| 1277 |
+
"learning_rate": 3.739691929505641e-05,
|
| 1278 |
+
"loss": 0.2062,
|
| 1279 |
+
"step": 890
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 0.8680892337536372,
|
| 1283 |
+
"grad_norm": 0.13484741747379303,
|
| 1284 |
+
"learning_rate": 3.720288946130197e-05,
|
| 1285 |
+
"loss": 0.1985,
|
| 1286 |
+
"step": 895
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.8729388942774006,
|
| 1290 |
+
"grad_norm": 0.15806835889816284,
|
| 1291 |
+
"learning_rate": 3.7007889815115796e-05,
|
| 1292 |
+
"loss": 0.2144,
|
| 1293 |
+
"step": 900
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 0.8777885548011639,
|
| 1297 |
+
"grad_norm": 0.16019117832183838,
|
| 1298 |
+
"learning_rate": 3.681193585389969e-05,
|
| 1299 |
+
"loss": 0.1912,
|
| 1300 |
+
"step": 905
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.8826382153249273,
|
| 1304 |
+
"grad_norm": 0.1545192003250122,
|
| 1305 |
+
"learning_rate": 3.6615043150898674e-05,
|
| 1306 |
+
"loss": 0.2093,
|
| 1307 |
+
"step": 910
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 0.8874878758486906,
|
| 1311 |
+
"grad_norm": 0.1420498490333557,
|
| 1312 |
+
"learning_rate": 3.641722735396336e-05,
|
| 1313 |
+
"loss": 0.205,
|
| 1314 |
+
"step": 915
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 0.8923375363724539,
|
| 1318 |
+
"grad_norm": 0.15403462946414948,
|
| 1319 |
+
"learning_rate": 3.6218504184306295e-05,
|
| 1320 |
+
"loss": 0.1907,
|
| 1321 |
+
"step": 920
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.8971871968962173,
|
| 1325 |
+
"grad_norm": 0.1699294149875641,
|
| 1326 |
+
"learning_rate": 3.6018889435252574e-05,
|
| 1327 |
+
"loss": 0.2399,
|
| 1328 |
+
"step": 925
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.9020368574199806,
|
| 1332 |
+
"grad_norm": 0.1434181183576584,
|
| 1333 |
+
"learning_rate": 3.581839897098468e-05,
|
| 1334 |
+
"loss": 0.2033,
|
| 1335 |
+
"step": 930
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 0.906886517943744,
|
| 1339 |
+
"grad_norm": 0.14518964290618896,
|
| 1340 |
+
"learning_rate": 3.561704872528169e-05,
|
| 1341 |
+
"loss": 0.1908,
|
| 1342 |
+
"step": 935
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.9117361784675073,
|
| 1346 |
+
"grad_norm": 0.1436786949634552,
|
| 1347 |
+
"learning_rate": 3.5414854700252945e-05,
|
| 1348 |
+
"loss": 0.202,
|
| 1349 |
+
"step": 940
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 0.9165858389912707,
|
| 1353 |
+
"grad_norm": 0.1517830491065979,
|
| 1354 |
+
"learning_rate": 3.521183296506628e-05,
|
| 1355 |
+
"loss": 0.2088,
|
| 1356 |
+
"step": 945
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 0.9214354995150339,
|
| 1360 |
+
"grad_norm": 0.1423126757144928,
|
| 1361 |
+
"learning_rate": 3.5007999654671004e-05,
|
| 1362 |
+
"loss": 0.2026,
|
| 1363 |
+
"step": 950
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 0.9262851600387972,
|
| 1367 |
+
"grad_norm": 0.13826879858970642,
|
| 1368 |
+
"learning_rate": 3.4803370968515535e-05,
|
| 1369 |
+
"loss": 0.2048,
|
| 1370 |
+
"step": 955
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.9311348205625606,
|
| 1374 |
+
"grad_norm": 0.15552102029323578,
|
| 1375 |
+
"learning_rate": 3.4597963169259994e-05,
|
| 1376 |
+
"loss": 0.2074,
|
| 1377 |
+
"step": 960
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.9359844810863239,
|
| 1381 |
+
"grad_norm": 0.1478404849767685,
|
| 1382 |
+
"learning_rate": 3.4391792581483724e-05,
|
| 1383 |
+
"loss": 0.2,
|
| 1384 |
+
"step": 965
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 0.9408341416100873,
|
| 1388 |
+
"grad_norm": 0.15055781602859497,
|
| 1389 |
+
"learning_rate": 3.4184875590387924e-05,
|
| 1390 |
+
"loss": 0.2032,
|
| 1391 |
+
"step": 970
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 0.9456838021338506,
|
| 1395 |
+
"grad_norm": 0.14816974103450775,
|
| 1396 |
+
"learning_rate": 3.397722864049347e-05,
|
| 1397 |
+
"loss": 0.2095,
|
| 1398 |
+
"step": 975
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 0.950533462657614,
|
| 1402 |
+
"grad_norm": 0.15103183686733246,
|
| 1403 |
+
"learning_rate": 3.376886823433395e-05,
|
| 1404 |
+
"loss": 0.2133,
|
| 1405 |
+
"step": 980
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 0.9553831231813773,
|
| 1409 |
+
"grad_norm": 0.15531527996063232,
|
| 1410 |
+
"learning_rate": 3.35598109311442e-05,
|
| 1411 |
+
"loss": 0.2125,
|
| 1412 |
+
"step": 985
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.9602327837051406,
|
| 1416 |
+
"grad_norm": 0.13943400979042053,
|
| 1417 |
+
"learning_rate": 3.335007334554423e-05,
|
| 1418 |
+
"loss": 0.1981,
|
| 1419 |
+
"step": 990
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.965082444228904,
|
| 1423 |
+
"grad_norm": 0.14304272830486298,
|
| 1424 |
+
"learning_rate": 3.3139672146218846e-05,
|
| 1425 |
+
"loss": 0.2037,
|
| 1426 |
+
"step": 995
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 0.9699321047526673,
|
| 1430 |
+
"grad_norm": 0.16258351504802704,
|
| 1431 |
+
"learning_rate": 3.2928624054592873e-05,
|
| 1432 |
+
"loss": 0.1965,
|
| 1433 |
+
"step": 1000
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 0.9699321047526673,
|
| 1437 |
+
"eval_loss": 0.20380738377571106,
|
| 1438 |
+
"eval_runtime": 48.08,
|
| 1439 |
+
"eval_samples_per_second": 14.018,
|
| 1440 |
+
"eval_steps_per_second": 0.894,
|
| 1441 |
+
"step": 1000
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.9747817652764307,
|
| 1445 |
+
"grad_norm": 0.14396609365940094,
|
| 1446 |
+
"learning_rate": 3.2716945843502224e-05,
|
| 1447 |
+
"loss": 0.2143,
|
| 1448 |
+
"step": 1005
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.979631425800194,
|
| 1452 |
+
"grad_norm": 0.1358533650636673,
|
| 1453 |
+
"learning_rate": 3.2504654335860986e-05,
|
| 1454 |
+
"loss": 0.2189,
|
| 1455 |
+
"step": 1010
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.9844810863239574,
|
| 1459 |
+
"grad_norm": 0.18899835646152496,
|
| 1460 |
+
"learning_rate": 3.229176640332433e-05,
|
| 1461 |
+
"loss": 0.2089,
|
| 1462 |
+
"step": 1015
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 0.9893307468477207,
|
| 1466 |
+
"grad_norm": 0.14681367576122284,
|
| 1467 |
+
"learning_rate": 3.2078298964947715e-05,
|
| 1468 |
+
"loss": 0.1817,
|
| 1469 |
+
"step": 1020
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 0.9941804073714839,
|
| 1473 |
+
"grad_norm": 0.13952337205410004,
|
| 1474 |
+
"learning_rate": 3.186426898584225e-05,
|
| 1475 |
+
"loss": 0.2128,
|
| 1476 |
+
"step": 1025
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 0.9990300678952473,
|
| 1480 |
+
"grad_norm": 0.14799629151821136,
|
| 1481 |
+
"learning_rate": 3.164969347582639e-05,
|
| 1482 |
+
"loss": 0.2014,
|
| 1483 |
+
"step": 1030
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.0038797284190106,
|
| 1487 |
+
"grad_norm": 0.16220593452453613,
|
| 1488 |
+
"learning_rate": 3.143458948807414e-05,
|
| 1489 |
+
"loss": 0.22,
|
| 1490 |
+
"step": 1035
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 1.008729388942774,
|
| 1494 |
+
"grad_norm": 0.1490844190120697,
|
| 1495 |
+
"learning_rate": 3.1218974117759714e-05,
|
| 1496 |
+
"loss": 0.2157,
|
| 1497 |
+
"step": 1040
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.0135790494665373,
|
| 1501 |
+
"grad_norm": 0.1579083800315857,
|
| 1502 |
+
"learning_rate": 3.100286450069897e-05,
|
| 1503 |
+
"loss": 0.1989,
|
| 1504 |
+
"step": 1045
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 1.0184287099903007,
|
| 1508 |
+
"grad_norm": 0.16225625574588776,
|
| 1509 |
+
"learning_rate": 3.0786277811987505e-05,
|
| 1510 |
+
"loss": 0.1976,
|
| 1511 |
+
"step": 1050
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 1.023278370514064,
|
| 1515 |
+
"grad_norm": 0.14916008710861206,
|
| 1516 |
+
"learning_rate": 3.0569231264635756e-05,
|
| 1517 |
+
"loss": 0.2004,
|
| 1518 |
+
"step": 1055
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 1.0281280310378274,
|
| 1522 |
+
"grad_norm": 0.14502666890621185,
|
| 1523 |
+
"learning_rate": 3.0351742108200888e-05,
|
| 1524 |
+
"loss": 0.2043,
|
| 1525 |
+
"step": 1060
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 1.0329776915615907,
|
| 1529 |
+
"grad_norm": 0.13275770843029022,
|
| 1530 |
+
"learning_rate": 3.0133827627416057e-05,
|
| 1531 |
+
"loss": 0.1969,
|
| 1532 |
+
"step": 1065
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 1.037827352085354,
|
| 1536 |
+
"grad_norm": 0.14670969545841217,
|
| 1537 |
+
"learning_rate": 2.9915505140816597e-05,
|
| 1538 |
+
"loss": 0.2154,
|
| 1539 |
+
"step": 1070
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.0426770126091174,
|
| 1543 |
+
"grad_norm": 0.15070156753063202,
|
| 1544 |
+
"learning_rate": 2.9696791999363727e-05,
|
| 1545 |
+
"loss": 0.2084,
|
| 1546 |
+
"step": 1075
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 1.0475266731328807,
|
| 1550 |
+
"grad_norm": 0.14457589387893677,
|
| 1551 |
+
"learning_rate": 2.9477705585065546e-05,
|
| 1552 |
+
"loss": 0.216,
|
| 1553 |
+
"step": 1080
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 1.052376333656644,
|
| 1557 |
+
"grad_norm": 0.1555938869714737,
|
| 1558 |
+
"learning_rate": 2.925826330959564e-05,
|
| 1559 |
+
"loss": 0.2024,
|
| 1560 |
+
"step": 1085
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 1.0572259941804074,
|
| 1564 |
+
"grad_norm": 0.15301556885242462,
|
| 1565 |
+
"learning_rate": 2.9038482612909335e-05,
|
| 1566 |
+
"loss": 0.2119,
|
| 1567 |
+
"step": 1090
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 1.0620756547041708,
|
| 1571 |
+
"grad_norm": 0.14304795861244202,
|
| 1572 |
+
"learning_rate": 2.8818380961857615e-05,
|
| 1573 |
+
"loss": 0.1878,
|
| 1574 |
+
"step": 1095
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 1.066925315227934,
|
| 1578 |
+
"grad_norm": 0.1455308496952057,
|
| 1579 |
+
"learning_rate": 2.8597975848798992e-05,
|
| 1580 |
+
"loss": 0.2073,
|
| 1581 |
+
"step": 1100
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 1.0717749757516974,
|
| 1585 |
+
"grad_norm": 0.15965287387371063,
|
| 1586 |
+
"learning_rate": 2.837728479020933e-05,
|
| 1587 |
+
"loss": 0.2036,
|
| 1588 |
+
"step": 1105
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 1.0766246362754608,
|
| 1592 |
+
"grad_norm": 0.1591757833957672,
|
| 1593 |
+
"learning_rate": 2.8156325325289717e-05,
|
| 1594 |
+
"loss": 0.1807,
|
| 1595 |
+
"step": 1110
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 1.0814742967992241,
|
| 1599 |
+
"grad_norm": 0.14703893661499023,
|
| 1600 |
+
"learning_rate": 2.7935115014572593e-05,
|
| 1601 |
+
"loss": 0.2166,
|
| 1602 |
+
"step": 1115
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 1.0863239573229875,
|
| 1606 |
+
"grad_norm": 0.14766725897789001,
|
| 1607 |
+
"learning_rate": 2.7713671438526073e-05,
|
| 1608 |
+
"loss": 0.2137,
|
| 1609 |
+
"step": 1120
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 1.0911736178467508,
|
| 1613 |
+
"grad_norm": 0.16928435862064362,
|
| 1614 |
+
"learning_rate": 2.7492012196156835e-05,
|
| 1615 |
+
"loss": 0.2041,
|
| 1616 |
+
"step": 1125
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 1.0960232783705142,
|
| 1620 |
+
"grad_norm": 0.16884630918502808,
|
| 1621 |
+
"learning_rate": 2.7270154903611393e-05,
|
| 1622 |
+
"loss": 0.1964,
|
| 1623 |
+
"step": 1130
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 1.1008729388942773,
|
| 1627 |
+
"grad_norm": 0.15490330755710602,
|
| 1628 |
+
"learning_rate": 2.7048117192776146e-05,
|
| 1629 |
+
"loss": 0.2096,
|
| 1630 |
+
"step": 1135
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 1.1057225994180406,
|
| 1634 |
+
"grad_norm": 0.15356089174747467,
|
| 1635 |
+
"learning_rate": 2.6825916709876013e-05,
|
| 1636 |
+
"loss": 0.2006,
|
| 1637 |
+
"step": 1140
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 1.110572259941804,
|
| 1641 |
+
"grad_norm": 0.14378608763217926,
|
| 1642 |
+
"learning_rate": 2.6603571114072066e-05,
|
| 1643 |
+
"loss": 0.2045,
|
| 1644 |
+
"step": 1145
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 1.1154219204655673,
|
| 1648 |
+
"grad_norm": 0.1541430503129959,
|
| 1649 |
+
"learning_rate": 2.638109807605811e-05,
|
| 1650 |
+
"loss": 0.2145,
|
| 1651 |
+
"step": 1150
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 1.1202715809893307,
|
| 1655 |
+
"grad_norm": 0.14816121757030487,
|
| 1656 |
+
"learning_rate": 2.6158515276656253e-05,
|
| 1657 |
+
"loss": 0.197,
|
| 1658 |
+
"step": 1155
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 1.125121241513094,
|
| 1662 |
+
"grad_norm": 0.15891754627227783,
|
| 1663 |
+
"learning_rate": 2.5935840405411804e-05,
|
| 1664 |
+
"loss": 0.2186,
|
| 1665 |
+
"step": 1160
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.1299709020368573,
|
| 1669 |
+
"grad_norm": 0.16058792173862457,
|
| 1670 |
+
"learning_rate": 2.571309115918738e-05,
|
| 1671 |
+
"loss": 0.2147,
|
| 1672 |
+
"step": 1165
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 1.1348205625606207,
|
| 1676 |
+
"grad_norm": 0.14306621253490448,
|
| 1677 |
+
"learning_rate": 2.549028524075649e-05,
|
| 1678 |
+
"loss": 0.2056,
|
| 1679 |
+
"step": 1170
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 1.139670223084384,
|
| 1683 |
+
"grad_norm": 0.1553443819284439,
|
| 1684 |
+
"learning_rate": 2.5267440357396588e-05,
|
| 1685 |
+
"loss": 0.1925,
|
| 1686 |
+
"step": 1175
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 1.1445198836081474,
|
| 1690 |
+
"grad_norm": 0.15235552191734314,
|
| 1691 |
+
"learning_rate": 2.5044574219481832e-05,
|
| 1692 |
+
"loss": 0.2173,
|
| 1693 |
+
"step": 1180
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 1.1493695441319107,
|
| 1697 |
+
"grad_norm": 0.13989904522895813,
|
| 1698 |
+
"learning_rate": 2.4821704539075577e-05,
|
| 1699 |
+
"loss": 0.1786,
|
| 1700 |
+
"step": 1185
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 1.154219204655674,
|
| 1704 |
+
"grad_norm": 0.1479206681251526,
|
| 1705 |
+
"learning_rate": 2.4598849028522713e-05,
|
| 1706 |
+
"loss": 0.2044,
|
| 1707 |
+
"step": 1190
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.1590688651794374,
|
| 1711 |
+
"grad_norm": 0.17453940212726593,
|
| 1712 |
+
"learning_rate": 2.4376025399041985e-05,
|
| 1713 |
+
"loss": 0.2086,
|
| 1714 |
+
"step": 1195
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 1.1639185257032008,
|
| 1718 |
+
"grad_norm": 0.15538154542446136,
|
| 1719 |
+
"learning_rate": 2.4153251359318404e-05,
|
| 1720 |
+
"loss": 0.1891,
|
| 1721 |
+
"step": 1200
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 1.168768186226964,
|
| 1725 |
+
"grad_norm": 0.14030437171459198,
|
| 1726 |
+
"learning_rate": 2.3930544614095944e-05,
|
| 1727 |
+
"loss": 0.1813,
|
| 1728 |
+
"step": 1205
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 1.1736178467507274,
|
| 1732 |
+
"grad_norm": 0.15048657357692719,
|
| 1733 |
+
"learning_rate": 2.370792286277035e-05,
|
| 1734 |
+
"loss": 0.2068,
|
| 1735 |
+
"step": 1210
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 1.1784675072744908,
|
| 1739 |
+
"grad_norm": 0.14109192788600922,
|
| 1740 |
+
"learning_rate": 2.3485403797982586e-05,
|
| 1741 |
+
"loss": 0.1992,
|
| 1742 |
+
"step": 1215
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 1.1833171677982541,
|
| 1746 |
+
"grad_norm": 0.1549970656633377,
|
| 1747 |
+
"learning_rate": 2.326300510421273e-05,
|
| 1748 |
+
"loss": 0.2067,
|
| 1749 |
+
"step": 1220
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 1.1881668283220175,
|
| 1753 |
+
"grad_norm": 0.15575067698955536,
|
| 1754 |
+
"learning_rate": 2.3040744456374462e-05,
|
| 1755 |
+
"loss": 0.1806,
|
| 1756 |
+
"step": 1225
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 1.1930164888457808,
|
| 1760 |
+
"grad_norm": 0.14550498127937317,
|
| 1761 |
+
"learning_rate": 2.2818639518410414e-05,
|
| 1762 |
+
"loss": 0.2088,
|
| 1763 |
+
"step": 1230
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 1.1978661493695442,
|
| 1767 |
+
"grad_norm": 0.15763843059539795,
|
| 1768 |
+
"learning_rate": 2.259670794188833e-05,
|
| 1769 |
+
"loss": 0.2121,
|
| 1770 |
+
"step": 1235
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 1.2027158098933075,
|
| 1774 |
+
"grad_norm": 0.17174339294433594,
|
| 1775 |
+
"learning_rate": 2.2374967364598222e-05,
|
| 1776 |
+
"loss": 0.1971,
|
| 1777 |
+
"step": 1240
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 1.2075654704170709,
|
| 1781 |
+
"grad_norm": 0.13719911873340607,
|
| 1782 |
+
"learning_rate": 2.2153435409150635e-05,
|
| 1783 |
+
"loss": 0.1851,
|
| 1784 |
+
"step": 1245
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 1.2124151309408342,
|
| 1788 |
+
"grad_norm": 0.1528056263923645,
|
| 1789 |
+
"learning_rate": 2.1932129681576105e-05,
|
| 1790 |
+
"loss": 0.2001,
|
| 1791 |
+
"step": 1250
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 1.2124151309408342,
|
| 1795 |
+
"eval_loss": 0.20025967061519623,
|
| 1796 |
+
"eval_runtime": 48.0897,
|
| 1797 |
+
"eval_samples_per_second": 14.015,
|
| 1798 |
+
"eval_steps_per_second": 0.894,
|
| 1799 |
+
"step": 1250
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 1.2172647914645975,
|
| 1803 |
+
"grad_norm": 0.16801750659942627,
|
| 1804 |
+
"learning_rate": 2.1711067769925914e-05,
|
| 1805 |
+
"loss": 0.2184,
|
| 1806 |
+
"step": 1255
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 1.2221144519883609,
|
| 1810 |
+
"grad_norm": 0.1539103388786316,
|
| 1811 |
+
"learning_rate": 2.149026724287435e-05,
|
| 1812 |
+
"loss": 0.2176,
|
| 1813 |
+
"step": 1260
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 1.2269641125121242,
|
| 1817 |
+
"grad_norm": 0.15657076239585876,
|
| 1818 |
+
"learning_rate": 2.126974564832239e-05,
|
| 1819 |
+
"loss": 0.1989,
|
| 1820 |
+
"step": 1265
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 1.2318137730358876,
|
| 1824 |
+
"grad_norm": 0.15106332302093506,
|
| 1825 |
+
"learning_rate": 2.104952051200315e-05,
|
| 1826 |
+
"loss": 0.2045,
|
| 1827 |
+
"step": 1270
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 1.236663433559651,
|
| 1831 |
+
"grad_norm": 0.14375294744968414,
|
| 1832 |
+
"learning_rate": 2.082960933608903e-05,
|
| 1833 |
+
"loss": 0.1917,
|
| 1834 |
+
"step": 1275
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 1.2415130940834143,
|
| 1838 |
+
"grad_norm": 0.14703327417373657,
|
| 1839 |
+
"learning_rate": 2.061002959780069e-05,
|
| 1840 |
+
"loss": 0.1977,
|
| 1841 |
+
"step": 1280
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 1.2463627546071776,
|
| 1845 |
+
"grad_norm": 0.1680404096841812,
|
| 1846 |
+
"learning_rate": 2.03907987480182e-05,
|
| 1847 |
+
"loss": 0.2127,
|
| 1848 |
+
"step": 1285
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 1.251212415130941,
|
| 1852 |
+
"grad_norm": 0.14668302237987518,
|
| 1853 |
+
"learning_rate": 2.0171934209893994e-05,
|
| 1854 |
+
"loss": 0.2012,
|
| 1855 |
+
"step": 1290
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 1.2560620756547043,
|
| 1859 |
+
"grad_norm": 0.14603851735591888,
|
| 1860 |
+
"learning_rate": 1.9953453377468282e-05,
|
| 1861 |
+
"loss": 0.1996,
|
| 1862 |
+
"step": 1295
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 1.2609117361784676,
|
| 1866 |
+
"grad_norm": 0.1648787260055542,
|
| 1867 |
+
"learning_rate": 1.973537361428665e-05,
|
| 1868 |
+
"loss": 0.2164,
|
| 1869 |
+
"step": 1300
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 1.265761396702231,
|
| 1873 |
+
"grad_norm": 0.16897331178188324,
|
| 1874 |
+
"learning_rate": 1.95177122520201e-05,
|
| 1875 |
+
"loss": 0.1999,
|
| 1876 |
+
"step": 1305
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 1.270611057225994,
|
| 1880 |
+
"grad_norm": 0.16626134514808655,
|
| 1881 |
+
"learning_rate": 1.9300486589087645e-05,
|
| 1882 |
+
"loss": 0.1934,
|
| 1883 |
+
"step": 1310
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 1.2754607177497574,
|
| 1887 |
+
"grad_norm": 0.15668809413909912,
|
| 1888 |
+
"learning_rate": 1.908371388928152e-05,
|
| 1889 |
+
"loss": 0.2078,
|
| 1890 |
+
"step": 1315
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 1.2803103782735208,
|
| 1894 |
+
"grad_norm": 0.14479213953018188,
|
| 1895 |
+
"learning_rate": 1.8867411380395177e-05,
|
| 1896 |
+
"loss": 0.198,
|
| 1897 |
+
"step": 1320
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 1.2851600387972841,
|
| 1901 |
+
"grad_norm": 0.1500495970249176,
|
| 1902 |
+
"learning_rate": 1.865159625285412e-05,
|
| 1903 |
+
"loss": 0.2103,
|
| 1904 |
+
"step": 1325
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 1.2900096993210475,
|
| 1908 |
+
"grad_norm": 0.1683105230331421,
|
| 1909 |
+
"learning_rate": 1.8436285658349694e-05,
|
| 1910 |
+
"loss": 0.2045,
|
| 1911 |
+
"step": 1330
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 1.2948593598448108,
|
| 1915 |
+
"grad_norm": 0.16197173297405243,
|
| 1916 |
+
"learning_rate": 1.822149670847599e-05,
|
| 1917 |
+
"loss": 0.189,
|
| 1918 |
+
"step": 1335
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 1.2997090203685742,
|
| 1922 |
+
"grad_norm": 0.1493465155363083,
|
| 1923 |
+
"learning_rate": 1.800724647336992e-05,
|
| 1924 |
+
"loss": 0.1998,
|
| 1925 |
+
"step": 1340
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 1.3045586808923375,
|
| 1929 |
+
"grad_norm": 0.15609502792358398,
|
| 1930 |
+
"learning_rate": 1.779355198035458e-05,
|
| 1931 |
+
"loss": 0.1944,
|
| 1932 |
+
"step": 1345
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 1.3094083414161009,
|
| 1936 |
+
"grad_norm": 0.16641376912593842,
|
| 1937 |
+
"learning_rate": 1.758043021258601e-05,
|
| 1938 |
+
"loss": 0.2032,
|
| 1939 |
+
"step": 1350
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 1.3142580019398642,
|
| 1943 |
+
"grad_norm": 0.16449637711048126,
|
| 1944 |
+
"learning_rate": 1.7367898107703497e-05,
|
| 1945 |
+
"loss": 0.2099,
|
| 1946 |
+
"step": 1355
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 1.3191076624636275,
|
| 1950 |
+
"grad_norm": 0.14618897438049316,
|
| 1951 |
+
"learning_rate": 1.7155972556483424e-05,
|
| 1952 |
+
"loss": 0.1924,
|
| 1953 |
+
"step": 1360
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 1.3239573229873909,
|
| 1957 |
+
"grad_norm": 0.1407703310251236,
|
| 1958 |
+
"learning_rate": 1.6944670401496997e-05,
|
| 1959 |
+
"loss": 0.1909,
|
| 1960 |
+
"step": 1365
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 1.3288069835111542,
|
| 1964 |
+
"grad_norm": 0.17186439037322998,
|
| 1965 |
+
"learning_rate": 1.673400843577156e-05,
|
| 1966 |
+
"loss": 0.2035,
|
| 1967 |
+
"step": 1370
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 1.3336566440349176,
|
| 1971 |
+
"grad_norm": 0.14018838107585907,
|
| 1972 |
+
"learning_rate": 1.652400340145609e-05,
|
| 1973 |
+
"loss": 0.1949,
|
| 1974 |
+
"step": 1375
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 1.338506304558681,
|
| 1978 |
+
"grad_norm": 0.14712657034397125,
|
| 1979 |
+
"learning_rate": 1.6314671988490603e-05,
|
| 1980 |
+
"loss": 0.1878,
|
| 1981 |
+
"step": 1380
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 1.3433559650824443,
|
| 1985 |
+
"grad_norm": 0.15526027977466583,
|
| 1986 |
+
"learning_rate": 1.610603083327971e-05,
|
| 1987 |
+
"loss": 0.1971,
|
| 1988 |
+
"step": 1385
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 1.3482056256062076,
|
| 1992 |
+
"grad_norm": 0.14195865392684937,
|
| 1993 |
+
"learning_rate": 1.5898096517370498e-05,
|
| 1994 |
+
"loss": 0.196,
|
| 1995 |
+
"step": 1390
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.353055286129971,
|
| 1999 |
+
"grad_norm": 0.16893810033798218,
|
| 2000 |
+
"learning_rate": 1.5690885566134696e-05,
|
| 2001 |
+
"loss": 0.1897,
|
| 2002 |
+
"step": 1395
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 1.3579049466537343,
|
| 2006 |
+
"grad_norm": 0.16904611885547638,
|
| 2007 |
+
"learning_rate": 1.5484414447455348e-05,
|
| 2008 |
+
"loss": 0.1959,
|
| 2009 |
+
"step": 1400
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 1.3627546071774976,
|
| 2013 |
+
"grad_norm": 0.16385570168495178,
|
| 2014 |
+
"learning_rate": 1.5278699570418056e-05,
|
| 2015 |
+
"loss": 0.1971,
|
| 2016 |
+
"step": 1405
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 1.367604267701261,
|
| 2020 |
+
"grad_norm": 0.16677437722682953,
|
| 2021 |
+
"learning_rate": 1.5073757284006868e-05,
|
| 2022 |
+
"loss": 0.1823,
|
| 2023 |
+
"step": 1410
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 1.3724539282250243,
|
| 2027 |
+
"grad_norm": 0.1567675769329071,
|
| 2028 |
+
"learning_rate": 1.4869603875804955e-05,
|
| 2029 |
+
"loss": 0.1941,
|
| 2030 |
+
"step": 1415
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 1.3773035887487877,
|
| 2034 |
+
"grad_norm": 0.1708156019449234,
|
| 2035 |
+
"learning_rate": 1.4666255570700202e-05,
|
| 2036 |
+
"loss": 0.1997,
|
| 2037 |
+
"step": 1420
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 1.3821532492725508,
|
| 2041 |
+
"grad_norm": 0.14904044568538666,
|
| 2042 |
+
"learning_rate": 1.4463728529595719e-05,
|
| 2043 |
+
"loss": 0.1885,
|
| 2044 |
+
"step": 1425
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 1.3870029097963141,
|
| 2048 |
+
"grad_norm": 0.15045498311519623,
|
| 2049 |
+
"learning_rate": 1.4262038848125475e-05,
|
| 2050 |
+
"loss": 0.2168,
|
| 2051 |
+
"step": 1430
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 1.3918525703200775,
|
| 2055 |
+
"grad_norm": 0.1507643461227417,
|
| 2056 |
+
"learning_rate": 1.4061202555375142e-05,
|
| 2057 |
+
"loss": 0.1959,
|
| 2058 |
+
"step": 1435
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 1.3967022308438408,
|
| 2062 |
+
"grad_norm": 0.15534117817878723,
|
| 2063 |
+
"learning_rate": 1.3861235612608142e-05,
|
| 2064 |
+
"loss": 0.1976,
|
| 2065 |
+
"step": 1440
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 1.4015518913676042,
|
| 2069 |
+
"grad_norm": 0.14476318657398224,
|
| 2070 |
+
"learning_rate": 1.366215391199721e-05,
|
| 2071 |
+
"loss": 0.1878,
|
| 2072 |
+
"step": 1445
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 1.4064015518913675,
|
| 2076 |
+
"grad_norm": 0.144943967461586,
|
| 2077 |
+
"learning_rate": 1.346397327536134e-05,
|
| 2078 |
+
"loss": 0.1977,
|
| 2079 |
+
"step": 1450
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 1.4112512124151309,
|
| 2083 |
+
"grad_norm": 0.1705586314201355,
|
| 2084 |
+
"learning_rate": 1.3266709452908348e-05,
|
| 2085 |
+
"loss": 0.2113,
|
| 2086 |
+
"step": 1455
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 1.4161008729388942,
|
| 2090 |
+
"grad_norm": 0.14363010227680206,
|
| 2091 |
+
"learning_rate": 1.3070378121983174e-05,
|
| 2092 |
+
"loss": 0.1902,
|
| 2093 |
+
"step": 1460
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 1.4209505334626575,
|
| 2097 |
+
"grad_norm": 0.17444470524787903,
|
| 2098 |
+
"learning_rate": 1.2874994885821933e-05,
|
| 2099 |
+
"loss": 0.1994,
|
| 2100 |
+
"step": 1465
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 1.4258001939864209,
|
| 2104 |
+
"grad_norm": 0.15788498520851135,
|
| 2105 |
+
"learning_rate": 1.268057527231186e-05,
|
| 2106 |
+
"loss": 0.1999,
|
| 2107 |
+
"step": 1470
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 1.4306498545101842,
|
| 2111 |
+
"grad_norm": 0.1490585207939148,
|
| 2112 |
+
"learning_rate": 1.2487134732757219e-05,
|
| 2113 |
+
"loss": 0.1904,
|
| 2114 |
+
"step": 1475
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 1.4354995150339476,
|
| 2118 |
+
"grad_norm": 0.1542252153158188,
|
| 2119 |
+
"learning_rate": 1.2294688640651355e-05,
|
| 2120 |
+
"loss": 0.2078,
|
| 2121 |
+
"step": 1480
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 1.440349175557711,
|
| 2125 |
+
"grad_norm": 0.1651746928691864,
|
| 2126 |
+
"learning_rate": 1.2103252290454942e-05,
|
| 2127 |
+
"loss": 0.1843,
|
| 2128 |
+
"step": 1485
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 1.4451988360814743,
|
| 2132 |
+
"grad_norm": 0.15188032388687134,
|
| 2133 |
+
"learning_rate": 1.191284089638037e-05,
|
| 2134 |
+
"loss": 0.2107,
|
| 2135 |
+
"step": 1490
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 1.4500484966052376,
|
| 2139 |
+
"grad_norm": 0.15589340031147003,
|
| 2140 |
+
"learning_rate": 1.1723469591182675e-05,
|
| 2141 |
+
"loss": 0.1961,
|
| 2142 |
+
"step": 1495
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 1.454898157129001,
|
| 2146 |
+
"grad_norm": 0.1623036116361618,
|
| 2147 |
+
"learning_rate": 1.1535153424956922e-05,
|
| 2148 |
+
"loss": 0.2087,
|
| 2149 |
+
"step": 1500
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 1.454898157129001,
|
| 2153 |
+
"eval_loss": 0.197998046875,
|
| 2154 |
+
"eval_runtime": 48.0919,
|
| 2155 |
+
"eval_samples_per_second": 14.015,
|
| 2156 |
+
"eval_steps_per_second": 0.894,
|
| 2157 |
+
"step": 1500
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 1.4597478176527643,
|
| 2161 |
+
"grad_norm": 0.1391475796699524,
|
| 2162 |
+
"learning_rate": 1.134790736394199e-05,
|
| 2163 |
+
"loss": 0.1829,
|
| 2164 |
+
"step": 1505
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 1.4645974781765276,
|
| 2168 |
+
"grad_norm": 0.15713264048099518,
|
| 2169 |
+
"learning_rate": 1.116174628933127e-05,
|
| 2170 |
+
"loss": 0.1862,
|
| 2171 |
+
"step": 1510
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.469447138700291,
|
| 2175 |
+
"grad_norm": 0.14922243356704712,
|
| 2176 |
+
"learning_rate": 1.0976684996089911e-05,
|
| 2177 |
+
"loss": 0.2022,
|
| 2178 |
+
"step": 1515
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 1.4742967992240543,
|
| 2182 |
+
"grad_norm": 0.15391802787780762,
|
| 2183 |
+
"learning_rate": 1.0792738191779055e-05,
|
| 2184 |
+
"loss": 0.199,
|
| 2185 |
+
"step": 1520
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.4791464597478177,
|
| 2189 |
+
"grad_norm": 0.16635461151599884,
|
| 2190 |
+
"learning_rate": 1.0609920495386963e-05,
|
| 2191 |
+
"loss": 0.183,
|
| 2192 |
+
"step": 1525
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.483996120271581,
|
| 2196 |
+
"grad_norm": 0.14732369780540466,
|
| 2197 |
+
"learning_rate": 1.0428246436167136e-05,
|
| 2198 |
+
"loss": 0.1946,
|
| 2199 |
+
"step": 1530
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 1.4888457807953444,
|
| 2203 |
+
"grad_norm": 0.15745453536510468,
|
| 2204 |
+
"learning_rate": 1.0247730452483683e-05,
|
| 2205 |
+
"loss": 0.1999,
|
| 2206 |
+
"step": 1535
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.4936954413191077,
|
| 2210 |
+
"grad_norm": 0.1535121500492096,
|
| 2211 |
+
"learning_rate": 1.0068386890663828e-05,
|
| 2212 |
+
"loss": 0.2131,
|
| 2213 |
+
"step": 1540
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 1.498545101842871,
|
| 2217 |
+
"grad_norm": 0.16908058524131775,
|
| 2218 |
+
"learning_rate": 9.890230003857704e-06,
|
| 2219 |
+
"loss": 0.1893,
|
| 2220 |
+
"step": 1545
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 1.5033947623666344,
|
| 2224 |
+
"grad_norm": 0.16990558803081512,
|
| 2225 |
+
"learning_rate": 9.71327395090566e-06,
|
| 2226 |
+
"loss": 0.2033,
|
| 2227 |
+
"step": 1550
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.5082444228903977,
|
| 2231 |
+
"grad_norm": 0.1563047617673874,
|
| 2232 |
+
"learning_rate": 9.537532795212988e-06,
|
| 2233 |
+
"loss": 0.1975,
|
| 2234 |
+
"step": 1555
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 1.513094083414161,
|
| 2238 |
+
"grad_norm": 0.16167347133159637,
|
| 2239 |
+
"learning_rate": 9.363020503632197e-06,
|
| 2240 |
+
"loss": 0.1923,
|
| 2241 |
+
"step": 1560
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.5179437439379244,
|
| 2245 |
+
"grad_norm": 0.1589316427707672,
|
| 2246 |
+
"learning_rate": 9.189750945353082e-06,
|
| 2247 |
+
"loss": 0.2131,
|
| 2248 |
+
"step": 1565
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.5227934044616878,
|
| 2252 |
+
"grad_norm": 0.15486431121826172,
|
| 2253 |
+
"learning_rate": 9.01773789080045e-06,
|
| 2254 |
+
"loss": 0.1961,
|
| 2255 |
+
"step": 1570
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 1.527643064985451,
|
| 2259 |
+
"grad_norm": 0.17133159935474396,
|
| 2260 |
+
"learning_rate": 8.846995010539707e-06,
|
| 2261 |
+
"loss": 0.179,
|
| 2262 |
+
"step": 1575
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 1.5324927255092144,
|
| 2266 |
+
"grad_norm": 0.1626712530851364,
|
| 2267 |
+
"learning_rate": 8.67753587419046e-06,
|
| 2268 |
+
"loss": 0.1977,
|
| 2269 |
+
"step": 1580
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.5373423860329778,
|
| 2273 |
+
"grad_norm": 0.15872539579868317,
|
| 2274 |
+
"learning_rate": 8.509373949348012e-06,
|
| 2275 |
+
"loss": 0.1981,
|
| 2276 |
+
"step": 1585
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.5421920465567411,
|
| 2280 |
+
"grad_norm": 0.15292125940322876,
|
| 2281 |
+
"learning_rate": 8.34252260051314e-06,
|
| 2282 |
+
"loss": 0.1971,
|
| 2283 |
+
"step": 1590
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.5470417070805045,
|
| 2287 |
+
"grad_norm": 0.15812063217163086,
|
| 2288 |
+
"learning_rate": 8.176995088029852e-06,
|
| 2289 |
+
"loss": 0.1964,
|
| 2290 |
+
"step": 1595
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.5518913676042678,
|
| 2294 |
+
"grad_norm": 0.15867389738559723,
|
| 2295 |
+
"learning_rate": 8.01280456703159e-06,
|
| 2296 |
+
"loss": 0.1971,
|
| 2297 |
+
"step": 1600
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.5567410281280312,
|
| 2301 |
+
"grad_norm": 0.1688418984413147,
|
| 2302 |
+
"learning_rate": 7.84996408639576e-06,
|
| 2303 |
+
"loss": 0.2117,
|
| 2304 |
+
"step": 1605
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.5615906886517945,
|
| 2308 |
+
"grad_norm": 0.16065895557403564,
|
| 2309 |
+
"learning_rate": 7.688486587706661e-06,
|
| 2310 |
+
"loss": 0.199,
|
| 2311 |
+
"step": 1610
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.5664403491755579,
|
| 2315 |
+
"grad_norm": 0.15269528329372406,
|
| 2316 |
+
"learning_rate": 7.528384904226932e-06,
|
| 2317 |
+
"loss": 0.1894,
|
| 2318 |
+
"step": 1615
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.5712900096993212,
|
| 2322 |
+
"grad_norm": 0.16433826088905334,
|
| 2323 |
+
"learning_rate": 7.369671759877705e-06,
|
| 2324 |
+
"loss": 0.2061,
|
| 2325 |
+
"step": 1620
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.5761396702230845,
|
| 2329 |
+
"grad_norm": 0.15707933902740479,
|
| 2330 |
+
"learning_rate": 7.212359768227345e-06,
|
| 2331 |
+
"loss": 0.201,
|
| 2332 |
+
"step": 1625
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.5809893307468477,
|
| 2336 |
+
"grad_norm": 0.15439623594284058,
|
| 2337 |
+
"learning_rate": 7.056461431489015e-06,
|
| 2338 |
+
"loss": 0.1866,
|
| 2339 |
+
"step": 1630
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.585838991270611,
|
| 2343 |
+
"grad_norm": 0.1509786695241928,
|
| 2344 |
+
"learning_rate": 6.901989139527048e-06,
|
| 2345 |
+
"loss": 0.1877,
|
| 2346 |
+
"step": 1635
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.5906886517943744,
|
| 2350 |
+
"grad_norm": 0.16154807806015015,
|
| 2351 |
+
"learning_rate": 6.748955168872312e-06,
|
| 2352 |
+
"loss": 0.1974,
|
| 2353 |
+
"step": 1640
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.5955383123181377,
|
| 2357 |
+
"grad_norm": 0.17238792777061462,
|
| 2358 |
+
"learning_rate": 6.5973716817465365e-06,
|
| 2359 |
+
"loss": 0.2212,
|
| 2360 |
+
"step": 1645
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.600387972841901,
|
| 2364 |
+
"grad_norm": 0.1707492172718048,
|
| 2365 |
+
"learning_rate": 6.4472507250956994e-06,
|
| 2366 |
+
"loss": 0.1852,
|
| 2367 |
+
"step": 1650
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.6052376333656644,
|
| 2371 |
+
"grad_norm": 0.16531886160373688,
|
| 2372 |
+
"learning_rate": 6.2986042296326296e-06,
|
| 2373 |
+
"loss": 0.1953,
|
| 2374 |
+
"step": 1655
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.6100872938894277,
|
| 2378 |
+
"grad_norm": 0.154957577586174,
|
| 2379 |
+
"learning_rate": 6.1514440088888735e-06,
|
| 2380 |
+
"loss": 0.2059,
|
| 2381 |
+
"step": 1660
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.614936954413191,
|
| 2385 |
+
"grad_norm": 0.16685739159584045,
|
| 2386 |
+
"learning_rate": 6.005781758275731e-06,
|
| 2387 |
+
"loss": 0.2111,
|
| 2388 |
+
"step": 1665
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.6197866149369544,
|
| 2392 |
+
"grad_norm": 0.16281373798847198,
|
| 2393 |
+
"learning_rate": 5.861629054154852e-06,
|
| 2394 |
+
"loss": 0.2048,
|
| 2395 |
+
"step": 1670
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.6246362754607178,
|
| 2399 |
+
"grad_norm": 0.16481256484985352,
|
| 2400 |
+
"learning_rate": 5.7189973529181705e-06,
|
| 2401 |
+
"loss": 0.2057,
|
| 2402 |
+
"step": 1675
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.629485935984481,
|
| 2406 |
+
"grad_norm": 0.15931367874145508,
|
| 2407 |
+
"learning_rate": 5.577897990077458e-06,
|
| 2408 |
+
"loss": 0.209,
|
| 2409 |
+
"step": 1680
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.6343355965082444,
|
| 2413 |
+
"grad_norm": 0.14951331913471222,
|
| 2414 |
+
"learning_rate": 5.438342179363418e-06,
|
| 2415 |
+
"loss": 0.2012,
|
| 2416 |
+
"step": 1685
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.6391852570320078,
|
| 2420 |
+
"grad_norm": 0.1307000368833542,
|
| 2421 |
+
"learning_rate": 5.300341011834475e-06,
|
| 2422 |
+
"loss": 0.1821,
|
| 2423 |
+
"step": 1690
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.6440349175557711,
|
| 2427 |
+
"grad_norm": 0.16035383939743042,
|
| 2428 |
+
"learning_rate": 5.163905454995372e-06,
|
| 2429 |
+
"loss": 0.1899,
|
| 2430 |
+
"step": 1695
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.6488845780795345,
|
| 2434 |
+
"grad_norm": 0.15784330666065216,
|
| 2435 |
+
"learning_rate": 5.0290463519254925e-06,
|
| 2436 |
+
"loss": 0.1949,
|
| 2437 |
+
"step": 1700
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.6537342386032978,
|
| 2441 |
+
"grad_norm": 0.1625281125307083,
|
| 2442 |
+
"learning_rate": 4.895774420417124e-06,
|
| 2443 |
+
"loss": 0.2106,
|
| 2444 |
+
"step": 1705
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 1.658583899127061,
|
| 2448 |
+
"grad_norm": 0.15070758759975433,
|
| 2449 |
+
"learning_rate": 4.764100252123702e-06,
|
| 2450 |
+
"loss": 0.1786,
|
| 2451 |
+
"step": 1710
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.6634335596508243,
|
| 2455 |
+
"grad_norm": 0.13678708672523499,
|
| 2456 |
+
"learning_rate": 4.634034311718022e-06,
|
| 2457 |
+
"loss": 0.1944,
|
| 2458 |
+
"step": 1715
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.6682832201745876,
|
| 2462 |
+
"grad_norm": 0.16914202272891998,
|
| 2463 |
+
"learning_rate": 4.505586936060566e-06,
|
| 2464 |
+
"loss": 0.2035,
|
| 2465 |
+
"step": 1720
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.673132880698351,
|
| 2469 |
+
"grad_norm": 0.15594607591629028,
|
| 2470 |
+
"learning_rate": 4.378768333378017e-06,
|
| 2471 |
+
"loss": 0.1883,
|
| 2472 |
+
"step": 1725
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 1.6779825412221143,
|
| 2476 |
+
"grad_norm": 0.15200097858905792,
|
| 2477 |
+
"learning_rate": 4.253588582451954e-06,
|
| 2478 |
+
"loss": 0.1786,
|
| 2479 |
+
"step": 1730
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 1.6828322017458777,
|
| 2483 |
+
"grad_norm": 0.1636827439069748,
|
| 2484 |
+
"learning_rate": 4.130057631817838e-06,
|
| 2485 |
+
"loss": 0.191,
|
| 2486 |
+
"step": 1735
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 1.687681862269641,
|
| 2490 |
+
"grad_norm": 0.14671088755130768,
|
| 2491 |
+
"learning_rate": 4.008185298974393e-06,
|
| 2492 |
+
"loss": 0.1992,
|
| 2493 |
+
"step": 1740
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.6925315227934044,
|
| 2497 |
+
"grad_norm": 0.1701505482196808,
|
| 2498 |
+
"learning_rate": 3.887981269603333e-06,
|
| 2499 |
+
"loss": 0.1913,
|
| 2500 |
+
"step": 1745
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.6973811833171677,
|
| 2504 |
+
"grad_norm": 0.15805399417877197,
|
| 2505 |
+
"learning_rate": 3.7694550967996565e-06,
|
| 2506 |
+
"loss": 0.1932,
|
| 2507 |
+
"step": 1750
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.6973811833171677,
|
| 2511 |
+
"eval_loss": 0.19688038527965546,
|
| 2512 |
+
"eval_runtime": 48.0858,
|
| 2513 |
+
"eval_samples_per_second": 14.017,
|
| 2514 |
+
"eval_steps_per_second": 0.894,
|
| 2515 |
+
"step": 1750
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 1.702230843840931,
|
| 2519 |
+
"grad_norm": 0.15937374532222748,
|
| 2520 |
+
"learning_rate": 3.6526162003123615e-06,
|
| 2521 |
+
"loss": 0.1776,
|
| 2522 |
+
"step": 1755
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 1.7070805043646944,
|
| 2526 |
+
"grad_norm": 0.15453214943408966,
|
| 2527 |
+
"learning_rate": 3.5374738657958454e-06,
|
| 2528 |
+
"loss": 0.2025,
|
| 2529 |
+
"step": 1760
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 1.7119301648884577,
|
| 2533 |
+
"grad_norm": 0.15527448058128357,
|
| 2534 |
+
"learning_rate": 3.4240372440719594e-06,
|
| 2535 |
+
"loss": 0.2062,
|
| 2536 |
+
"step": 1765
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 1.716779825412221,
|
| 2540 |
+
"grad_norm": 0.1499599665403366,
|
| 2541 |
+
"learning_rate": 3.3123153504027367e-06,
|
| 2542 |
+
"loss": 0.1999,
|
| 2543 |
+
"step": 1770
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 1.7216294859359844,
|
| 2547 |
+
"grad_norm": 0.1412288248538971,
|
| 2548 |
+
"learning_rate": 3.202317063773891e-06,
|
| 2549 |
+
"loss": 0.1763,
|
| 2550 |
+
"step": 1775
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 1.7264791464597478,
|
| 2554 |
+
"grad_norm": 0.16038767993450165,
|
| 2555 |
+
"learning_rate": 3.0940511261892154e-06,
|
| 2556 |
+
"loss": 0.1941,
|
| 2557 |
+
"step": 1780
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 1.731328806983511,
|
| 2561 |
+
"grad_norm": 0.15138490498065948,
|
| 2562 |
+
"learning_rate": 2.987526141975783e-06,
|
| 2563 |
+
"loss": 0.1969,
|
| 2564 |
+
"step": 1785
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 1.7361784675072744,
|
| 2568 |
+
"grad_norm": 0.15238988399505615,
|
| 2569 |
+
"learning_rate": 2.882750577100149e-06,
|
| 2570 |
+
"loss": 0.1865,
|
| 2571 |
+
"step": 1790
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 1.7410281280310378,
|
| 2575 |
+
"grad_norm": 0.16951079666614532,
|
| 2576 |
+
"learning_rate": 2.7797327584955014e-06,
|
| 2577 |
+
"loss": 0.1931,
|
| 2578 |
+
"step": 1795
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 1.7458777885548011,
|
| 2582 |
+
"grad_norm": 0.1534978598356247,
|
| 2583 |
+
"learning_rate": 2.6784808733999163e-06,
|
| 2584 |
+
"loss": 0.194,
|
| 2585 |
+
"step": 1800
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 1.7507274490785645,
|
| 2589 |
+
"grad_norm": 0.15956641733646393,
|
| 2590 |
+
"learning_rate": 2.579002968705668e-06,
|
| 2591 |
+
"loss": 0.1865,
|
| 2592 |
+
"step": 1805
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 1.7555771096023278,
|
| 2596 |
+
"grad_norm": 0.1396978348493576,
|
| 2597 |
+
"learning_rate": 2.481306950319706e-06,
|
| 2598 |
+
"loss": 0.1871,
|
| 2599 |
+
"step": 1810
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.7604267701260912,
|
| 2603 |
+
"grad_norm": 0.1500706672668457,
|
| 2604 |
+
"learning_rate": 2.3854005825353467e-06,
|
| 2605 |
+
"loss": 0.206,
|
| 2606 |
+
"step": 1815
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 1.7652764306498545,
|
| 2610 |
+
"grad_norm": 0.15206393599510193,
|
| 2611 |
+
"learning_rate": 2.29129148741524e-06,
|
| 2612 |
+
"loss": 0.204,
|
| 2613 |
+
"step": 1820
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 1.7701260911736179,
|
| 2617 |
+
"grad_norm": 0.15631067752838135,
|
| 2618 |
+
"learning_rate": 2.198987144185552e-06,
|
| 2619 |
+
"loss": 0.2022,
|
| 2620 |
+
"step": 1825
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 1.7749757516973812,
|
| 2624 |
+
"grad_norm": 0.16704493761062622,
|
| 2625 |
+
"learning_rate": 2.1084948886416295e-06,
|
| 2626 |
+
"loss": 0.2087,
|
| 2627 |
+
"step": 1830
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 1.7798254122211445,
|
| 2631 |
+
"grad_norm": 0.14626185595989227,
|
| 2632 |
+
"learning_rate": 2.0198219125649396e-06,
|
| 2633 |
+
"loss": 0.2053,
|
| 2634 |
+
"step": 1835
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 1.7846750727449079,
|
| 2638 |
+
"grad_norm": 0.1546330451965332,
|
| 2639 |
+
"learning_rate": 1.9329752631515513e-06,
|
| 2640 |
+
"loss": 0.1893,
|
| 2641 |
+
"step": 1840
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 1.7895247332686712,
|
| 2645 |
+
"grad_norm": 0.1591494381427765,
|
| 2646 |
+
"learning_rate": 1.8479618424520422e-06,
|
| 2647 |
+
"loss": 0.1943,
|
| 2648 |
+
"step": 1845
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 1.7943743937924346,
|
| 2652 |
+
"grad_norm": 0.1502322554588318,
|
| 2653 |
+
"learning_rate": 1.7647884068229676e-06,
|
| 2654 |
+
"loss": 0.1899,
|
| 2655 |
+
"step": 1850
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 1.799224054316198,
|
| 2659 |
+
"grad_norm": 0.1708429455757141,
|
| 2660 |
+
"learning_rate": 1.6834615663899155e-06,
|
| 2661 |
+
"loss": 0.214,
|
| 2662 |
+
"step": 1855
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 1.8040737148399613,
|
| 2666 |
+
"grad_norm": 0.15139105916023254,
|
| 2667 |
+
"learning_rate": 1.6039877845221735e-06,
|
| 2668 |
+
"loss": 0.1921,
|
| 2669 |
+
"step": 1860
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 1.8089233753637246,
|
| 2673 |
+
"grad_norm": 0.15141288936138153,
|
| 2674 |
+
"learning_rate": 1.5263733773190347e-06,
|
| 2675 |
+
"loss": 0.2192,
|
| 2676 |
+
"step": 1865
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 1.813773035887488,
|
| 2680 |
+
"grad_norm": 0.18391959369182587,
|
| 2681 |
+
"learning_rate": 1.4506245131078733e-06,
|
| 2682 |
+
"loss": 0.2072,
|
| 2683 |
+
"step": 1870
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.8186226964112513,
|
| 2687 |
+
"grad_norm": 0.15688742697238922,
|
| 2688 |
+
"learning_rate": 1.376747211953891e-06,
|
| 2689 |
+
"loss": 0.1824,
|
| 2690 |
+
"step": 1875
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 1.8234723569350146,
|
| 2694 |
+
"grad_norm": 0.15558670461177826,
|
| 2695 |
+
"learning_rate": 1.304747345181681e-06,
|
| 2696 |
+
"loss": 0.1975,
|
| 2697 |
+
"step": 1880
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.828322017458778,
|
| 2701 |
+
"grad_norm": 0.18685834109783173,
|
| 2702 |
+
"learning_rate": 1.2346306349086262e-06,
|
| 2703 |
+
"loss": 0.1957,
|
| 2704 |
+
"step": 1885
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 1.8331716779825413,
|
| 2708 |
+
"grad_norm": 0.16459019482135773,
|
| 2709 |
+
"learning_rate": 1.1664026535901318e-06,
|
| 2710 |
+
"loss": 0.2068,
|
| 2711 |
+
"step": 1890
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 1.8380213385063047,
|
| 2715 |
+
"grad_norm": 0.156924769282341,
|
| 2716 |
+
"learning_rate": 1.1000688235767525e-06,
|
| 2717 |
+
"loss": 0.2086,
|
| 2718 |
+
"step": 1895
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 1.842870999030068,
|
| 2722 |
+
"grad_norm": 0.16359376907348633,
|
| 2723 |
+
"learning_rate": 1.0356344166832648e-06,
|
| 2724 |
+
"loss": 0.2163,
|
| 2725 |
+
"step": 1900
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 1.8477206595538314,
|
| 2729 |
+
"grad_norm": 0.16292302310466766,
|
| 2730 |
+
"learning_rate": 9.73104553769691e-07,
|
| 2731 |
+
"loss": 0.2034,
|
| 2732 |
+
"step": 1905
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 1.8525703200775947,
|
| 2736 |
+
"grad_norm": 0.17051053047180176,
|
| 2737 |
+
"learning_rate": 9.124842043343407e-07,
|
| 2738 |
+
"loss": 0.2168,
|
| 2739 |
+
"step": 1910
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.857419980601358,
|
| 2743 |
+
"grad_norm": 0.17267897725105286,
|
| 2744 |
+
"learning_rate": 8.537781861188282e-07,
|
| 2745 |
+
"loss": 0.2234,
|
| 2746 |
+
"step": 1915
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 1.8622696411251214,
|
| 2750 |
+
"grad_norm": 0.16032524406909943,
|
| 2751 |
+
"learning_rate": 7.969911647252193e-07,
|
| 2752 |
+
"loss": 0.1986,
|
| 2753 |
+
"step": 1920
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 1.8671193016488847,
|
| 2757 |
+
"grad_norm": 0.1543329358100891,
|
| 2758 |
+
"learning_rate": 7.421276532452337e-07,
|
| 2759 |
+
"loss": 0.173,
|
| 2760 |
+
"step": 1925
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 1.871968962172648,
|
| 2764 |
+
"grad_norm": 0.15669037401676178,
|
| 2765 |
+
"learning_rate": 6.891920119015571e-07,
|
| 2766 |
+
"loss": 0.1987,
|
| 2767 |
+
"step": 1930
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 1.8768186226964112,
|
| 2771 |
+
"grad_norm": 0.1615551859140396,
|
| 2772 |
+
"learning_rate": 6.381884477013239e-07,
|
| 2773 |
+
"loss": 0.1831,
|
| 2774 |
+
"step": 1935
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 1.8816682832201745,
|
| 2778 |
+
"grad_norm": 0.1686321347951889,
|
| 2779 |
+
"learning_rate": 5.891210141017733e-07,
|
| 2780 |
+
"loss": 0.2029,
|
| 2781 |
+
"step": 1940
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.8865179437439379,
|
| 2785 |
+
"grad_norm": 0.15177270770072937,
|
| 2786 |
+
"learning_rate": 5.419936106881018e-07,
|
| 2787 |
+
"loss": 0.2034,
|
| 2788 |
+
"step": 1945
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 1.8913676042677012,
|
| 2792 |
+
"grad_norm": 0.155603289604187,
|
| 2793 |
+
"learning_rate": 4.968099828635525e-07,
|
| 2794 |
+
"loss": 0.1962,
|
| 2795 |
+
"step": 1950
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 1.8962172647914646,
|
| 2799 |
+
"grad_norm": 0.14843294024467468,
|
| 2800 |
+
"learning_rate": 4.535737215517366e-07,
|
| 2801 |
+
"loss": 0.1949,
|
| 2802 |
+
"step": 1955
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 1.901066925315228,
|
| 2806 |
+
"grad_norm": 0.15314973890781403,
|
| 2807 |
+
"learning_rate": 4.122882629112701e-07,
|
| 2808 |
+
"loss": 0.2104,
|
| 2809 |
+
"step": 1960
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 1.9059165858389913,
|
| 2813 |
+
"grad_norm": 0.1589522510766983,
|
| 2814 |
+
"learning_rate": 3.729568880626755e-07,
|
| 2815 |
+
"loss": 0.193,
|
| 2816 |
+
"step": 1965
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 1.9107662463627546,
|
| 2820 |
+
"grad_norm": 0.14796093106269836,
|
| 2821 |
+
"learning_rate": 3.355827228276182e-07,
|
| 2822 |
+
"loss": 0.1712,
|
| 2823 |
+
"step": 1970
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 1.915615906886518,
|
| 2827 |
+
"grad_norm": 0.1589411497116089,
|
| 2828 |
+
"learning_rate": 3.00168737480494e-07,
|
| 2829 |
+
"loss": 0.1878,
|
| 2830 |
+
"step": 1975
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 1.9204655674102813,
|
| 2834 |
+
"grad_norm": 0.14060987532138824,
|
| 2835 |
+
"learning_rate": 2.6671774651235437e-07,
|
| 2836 |
+
"loss": 0.1762,
|
| 2837 |
+
"step": 1980
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.9253152279340446,
|
| 2841 |
+
"grad_norm": 0.15994738042354584,
|
| 2842 |
+
"learning_rate": 2.3523240840724325e-07,
|
| 2843 |
+
"loss": 0.1924,
|
| 2844 |
+
"step": 1985
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 1.930164888457808,
|
| 2848 |
+
"grad_norm": 0.15940771996974945,
|
| 2849 |
+
"learning_rate": 2.0571522543090526e-07,
|
| 2850 |
+
"loss": 0.2053,
|
| 2851 |
+
"step": 1990
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 1.9350145489815713,
|
| 2855 |
+
"grad_norm": 0.14268319308757782,
|
| 2856 |
+
"learning_rate": 1.7816854343193357e-07,
|
| 2857 |
+
"loss": 0.1847,
|
| 2858 |
+
"step": 1995
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 1.9398642095053347,
|
| 2862 |
+
"grad_norm": 0.15468433499336243,
|
| 2863 |
+
"learning_rate": 1.5259455165531068e-07,
|
| 2864 |
+
"loss": 0.2044,
|
| 2865 |
+
"step": 2000
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 1.9398642095053347,
|
| 2869 |
+
"eval_loss": 0.19655267894268036,
|
| 2870 |
+
"eval_runtime": 48.0808,
|
| 2871 |
+
"eval_samples_per_second": 14.018,
|
| 2872 |
+
"eval_steps_per_second": 0.894,
|
| 2873 |
+
"step": 2000
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.944713870029098,
|
| 2877 |
+
"grad_norm": 0.1507444828748703,
|
| 2878 |
+
"learning_rate": 1.2899528256845873e-07,
|
| 2879 |
+
"loss": 0.1824,
|
| 2880 |
+
"step": 2005
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 1.9495635305528611,
|
| 2884 |
+
"grad_norm": 0.14235784113407135,
|
| 2885 |
+
"learning_rate": 1.073726116996715e-07,
|
| 2886 |
+
"loss": 0.1954,
|
| 2887 |
+
"step": 2010
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 1.9544131910766245,
|
| 2891 |
+
"grad_norm": 0.14553847908973694,
|
| 2892 |
+
"learning_rate": 8.77282574890892e-08,
|
| 2893 |
+
"loss": 0.2163,
|
| 2894 |
+
"step": 2015
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.9592628516003878,
|
| 2898 |
+
"grad_norm": 0.16772744059562683,
|
| 2899 |
+
"learning_rate": 7.006378115210766e-08,
|
| 2900 |
+
"loss": 0.1822,
|
| 2901 |
+
"step": 2020
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.9641125121241512,
|
| 2905 |
+
"grad_norm": 0.15165811777114868,
|
| 2906 |
+
"learning_rate": 5.43805865553082e-08,
|
| 2907 |
+
"loss": 0.1862,
|
| 2908 |
+
"step": 2025
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.9689621726479145,
|
| 2912 |
+
"grad_norm": 0.1540442407131195,
|
| 2913 |
+
"learning_rate": 4.067992010489685e-08,
|
| 2914 |
+
"loss": 0.2106,
|
| 2915 |
+
"step": 2030
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.9738118331716779,
|
| 2919 |
+
"grad_norm": 0.14102330803871155,
|
| 2920 |
+
"learning_rate": 2.8962870647630814e-08,
|
| 2921 |
+
"loss": 0.1879,
|
| 2922 |
+
"step": 2035
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 1.9786614936954412,
|
| 2926 |
+
"grad_norm": 0.14310438930988312,
|
| 2927 |
+
"learning_rate": 1.9230369384293233e-08,
|
| 2928 |
+
"loss": 0.1789,
|
| 2929 |
+
"step": 2040
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 1.9835111542192045,
|
| 2933 |
+
"grad_norm": 0.1545654982328415,
|
| 2934 |
+
"learning_rate": 1.148318979568297e-08,
|
| 2935 |
+
"loss": 0.193,
|
| 2936 |
+
"step": 2045
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.9883608147429679,
|
| 2940 |
+
"grad_norm": 0.15825237333774567,
|
| 2941 |
+
"learning_rate": 5.7219475811526316e-09,
|
| 2942 |
+
"loss": 0.1901,
|
| 2943 |
+
"step": 2050
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.9932104752667312,
|
| 2947 |
+
"grad_norm": 0.1504404991865158,
|
| 2948 |
+
"learning_rate": 1.947100609661634e-09,
|
| 2949 |
+
"loss": 0.2034,
|
| 2950 |
+
"step": 2055
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.9980601357904946,
|
| 2954 |
+
"grad_norm": 0.14141124486923218,
|
| 2955 |
+
"learning_rate": 1.5894888339418324e-10,
|
| 2956 |
+
"loss": 0.1773,
|
| 2957 |
+
"step": 2060
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 2.0,
|
| 2961 |
+
"step": 2062,
|
| 2962 |
+
"total_flos": 2.6820605084455404e+18,
|
| 2963 |
+
"train_loss": 0.2813772309560665,
|
| 2964 |
+
"train_runtime": 15083.292,
|
| 2965 |
+
"train_samples_per_second": 4.374,
|
| 2966 |
+
"train_steps_per_second": 0.137
|
| 2967 |
+
}
|
| 2968 |
+
],
|
| 2969 |
+
"logging_steps": 5,
|
| 2970 |
+
"max_steps": 2062,
|
| 2971 |
+
"num_input_tokens_seen": 0,
|
| 2972 |
+
"num_train_epochs": 2,
|
| 2973 |
+
"save_steps": 250,
|
| 2974 |
+
"total_flos": 2.6820605084455404e+18,
|
| 2975 |
+
"train_batch_size": 4,
|
| 2976 |
+
"trial_name": null,
|
| 2977 |
+
"trial_params": null
|
| 2978 |
+
}
|