Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +1883 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1883 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2630,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03802281368821293,
|
| 13 |
+
"grad_norm": 10.531242370605469,
|
| 14 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 15 |
+
"loss": 1.5247,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.07604562737642585,
|
| 20 |
+
"grad_norm": 5.058220863342285,
|
| 21 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 22 |
+
"loss": 1.1238,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.11406844106463879,
|
| 27 |
+
"grad_norm": 2.8926784992218018,
|
| 28 |
+
"learning_rate": 2.272727272727273e-05,
|
| 29 |
+
"loss": 0.4906,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.1520912547528517,
|
| 34 |
+
"grad_norm": 2.6709837913513184,
|
| 35 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 36 |
+
"loss": 0.3857,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.19011406844106463,
|
| 41 |
+
"grad_norm": 0.935093879699707,
|
| 42 |
+
"learning_rate": 3.787878787878788e-05,
|
| 43 |
+
"loss": 0.2874,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.22813688212927757,
|
| 48 |
+
"grad_norm": 1.3582843542099,
|
| 49 |
+
"learning_rate": 4.545454545454546e-05,
|
| 50 |
+
"loss": 0.2348,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.2661596958174905,
|
| 55 |
+
"grad_norm": 0.8589798212051392,
|
| 56 |
+
"learning_rate": 5.303030303030303e-05,
|
| 57 |
+
"loss": 0.1967,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.3041825095057034,
|
| 62 |
+
"grad_norm": 1.0151523351669312,
|
| 63 |
+
"learning_rate": 6.060606060606061e-05,
|
| 64 |
+
"loss": 0.1742,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.34220532319391633,
|
| 69 |
+
"grad_norm": 1.0646296739578247,
|
| 70 |
+
"learning_rate": 6.818181818181818e-05,
|
| 71 |
+
"loss": 0.1598,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.38022813688212925,
|
| 76 |
+
"grad_norm": 0.8589377403259277,
|
| 77 |
+
"learning_rate": 7.575757575757576e-05,
|
| 78 |
+
"loss": 0.1532,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.41825095057034223,
|
| 83 |
+
"grad_norm": 0.825715184211731,
|
| 84 |
+
"learning_rate": 8.333333333333334e-05,
|
| 85 |
+
"loss": 0.1468,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.45627376425855515,
|
| 90 |
+
"grad_norm": 0.9612506031990051,
|
| 91 |
+
"learning_rate": 9.090909090909092e-05,
|
| 92 |
+
"loss": 0.1401,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.49429657794676807,
|
| 97 |
+
"grad_norm": 0.8467527627944946,
|
| 98 |
+
"learning_rate": 9.848484848484849e-05,
|
| 99 |
+
"loss": 0.1281,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.532319391634981,
|
| 104 |
+
"grad_norm": 1.0839829444885254,
|
| 105 |
+
"learning_rate": 9.999746935517458e-05,
|
| 106 |
+
"loss": 0.1331,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.5703422053231939,
|
| 111 |
+
"grad_norm": 1.5719529390335083,
|
| 112 |
+
"learning_rate": 9.99871890496051e-05,
|
| 113 |
+
"loss": 0.1237,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.6083650190114068,
|
| 118 |
+
"grad_norm": 0.6265313625335693,
|
| 119 |
+
"learning_rate": 9.996900254271908e-05,
|
| 120 |
+
"loss": 0.1178,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.6463878326996197,
|
| 125 |
+
"grad_norm": 0.9848031997680664,
|
| 126 |
+
"learning_rate": 9.99429127109772e-05,
|
| 127 |
+
"loss": 0.103,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.6844106463878327,
|
| 132 |
+
"grad_norm": 1.1421473026275635,
|
| 133 |
+
"learning_rate": 9.9908923680866e-05,
|
| 134 |
+
"loss": 0.1079,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.7224334600760456,
|
| 139 |
+
"grad_norm": 0.986842691898346,
|
| 140 |
+
"learning_rate": 9.986704082824516e-05,
|
| 141 |
+
"loss": 0.1017,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.7604562737642585,
|
| 146 |
+
"grad_norm": 0.8210304975509644,
|
| 147 |
+
"learning_rate": 9.981727077749728e-05,
|
| 148 |
+
"loss": 0.0995,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.7984790874524715,
|
| 153 |
+
"grad_norm": 1.1234544515609741,
|
| 154 |
+
"learning_rate": 9.975962140048007e-05,
|
| 155 |
+
"loss": 0.0937,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.8365019011406845,
|
| 160 |
+
"grad_norm": 0.9322770833969116,
|
| 161 |
+
"learning_rate": 9.969410181528138e-05,
|
| 162 |
+
"loss": 0.0908,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.8745247148288974,
|
| 167 |
+
"grad_norm": 0.960360050201416,
|
| 168 |
+
"learning_rate": 9.962072238477698e-05,
|
| 169 |
+
"loss": 0.0973,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.9125475285171103,
|
| 174 |
+
"grad_norm": 0.7615262269973755,
|
| 175 |
+
"learning_rate": 9.953949471499157e-05,
|
| 176 |
+
"loss": 0.0934,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.9505703422053232,
|
| 181 |
+
"grad_norm": 0.8841362595558167,
|
| 182 |
+
"learning_rate": 9.945043165326309e-05,
|
| 183 |
+
"loss": 0.088,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.9885931558935361,
|
| 188 |
+
"grad_norm": 0.574943482875824,
|
| 189 |
+
"learning_rate": 9.935354728621069e-05,
|
| 190 |
+
"loss": 0.0858,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.026615969581749,
|
| 195 |
+
"grad_norm": 0.584782063961029,
|
| 196 |
+
"learning_rate": 9.92488569375068e-05,
|
| 197 |
+
"loss": 0.0841,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.064638783269962,
|
| 202 |
+
"grad_norm": 0.7055894732475281,
|
| 203 |
+
"learning_rate": 9.913637716545343e-05,
|
| 204 |
+
"loss": 0.0827,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.102661596958175,
|
| 209 |
+
"grad_norm": 0.46008625626564026,
|
| 210 |
+
"learning_rate": 9.901612576036325e-05,
|
| 211 |
+
"loss": 0.0813,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.1406844106463878,
|
| 216 |
+
"grad_norm": 0.592440664768219,
|
| 217 |
+
"learning_rate": 9.888812174174573e-05,
|
| 218 |
+
"loss": 0.0877,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.1787072243346008,
|
| 223 |
+
"grad_norm": 0.5659314393997192,
|
| 224 |
+
"learning_rate": 9.875238535529905e-05,
|
| 225 |
+
"loss": 0.0855,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.2167300380228137,
|
| 230 |
+
"grad_norm": 0.6649744510650635,
|
| 231 |
+
"learning_rate": 9.860893806970781e-05,
|
| 232 |
+
"loss": 0.0857,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.2547528517110267,
|
| 237 |
+
"grad_norm": 0.4853374660015106,
|
| 238 |
+
"learning_rate": 9.845780257324755e-05,
|
| 239 |
+
"loss": 0.0796,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.2927756653992395,
|
| 244 |
+
"grad_norm": 0.6634009480476379,
|
| 245 |
+
"learning_rate": 9.829900277019624e-05,
|
| 246 |
+
"loss": 0.0715,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.3307984790874525,
|
| 251 |
+
"grad_norm": 0.4123174250125885,
|
| 252 |
+
"learning_rate": 9.81325637770534e-05,
|
| 253 |
+
"loss": 0.0719,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.3688212927756653,
|
| 258 |
+
"grad_norm": 0.7861672043800354,
|
| 259 |
+
"learning_rate": 9.795851191856773e-05,
|
| 260 |
+
"loss": 0.0739,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 1.4068441064638784,
|
| 265 |
+
"grad_norm": 0.39750707149505615,
|
| 266 |
+
"learning_rate": 9.777687472357324e-05,
|
| 267 |
+
"loss": 0.0698,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 1.4448669201520912,
|
| 272 |
+
"grad_norm": 0.722964882850647,
|
| 273 |
+
"learning_rate": 9.758768092063536e-05,
|
| 274 |
+
"loss": 0.0699,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.4828897338403042,
|
| 279 |
+
"grad_norm": 0.6673786044120789,
|
| 280 |
+
"learning_rate": 9.739096043350699e-05,
|
| 281 |
+
"loss": 0.075,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.5209125475285172,
|
| 286 |
+
"grad_norm": 0.4799659252166748,
|
| 287 |
+
"learning_rate": 9.71867443763957e-05,
|
| 288 |
+
"loss": 0.0757,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 1.55893536121673,
|
| 293 |
+
"grad_norm": 0.485879510641098,
|
| 294 |
+
"learning_rate": 9.697506504904246e-05,
|
| 295 |
+
"loss": 0.0631,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.5969581749049429,
|
| 300 |
+
"grad_norm": 0.5157936811447144,
|
| 301 |
+
"learning_rate": 9.675595593161305e-05,
|
| 302 |
+
"loss": 0.0663,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.6349809885931559,
|
| 307 |
+
"grad_norm": 0.3810887634754181,
|
| 308 |
+
"learning_rate": 9.65294516794027e-05,
|
| 309 |
+
"loss": 0.0704,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.673003802281369,
|
| 314 |
+
"grad_norm": 0.5194770097732544,
|
| 315 |
+
"learning_rate": 9.629558811735476e-05,
|
| 316 |
+
"loss": 0.0771,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.7110266159695817,
|
| 321 |
+
"grad_norm": 0.7283788323402405,
|
| 322 |
+
"learning_rate": 9.605440223439453e-05,
|
| 323 |
+
"loss": 0.0695,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.7490494296577945,
|
| 328 |
+
"grad_norm": 0.6051397919654846,
|
| 329 |
+
"learning_rate": 9.580593217757894e-05,
|
| 330 |
+
"loss": 0.0733,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 1.7870722433460076,
|
| 335 |
+
"grad_norm": 0.48710307478904724,
|
| 336 |
+
"learning_rate": 9.555021724606298e-05,
|
| 337 |
+
"loss": 0.0688,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.8250950570342206,
|
| 342 |
+
"grad_norm": 0.44367125630378723,
|
| 343 |
+
"learning_rate": 9.5287297884884e-05,
|
| 344 |
+
"loss": 0.0632,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.8631178707224336,
|
| 349 |
+
"grad_norm": 0.6892393231391907,
|
| 350 |
+
"learning_rate": 9.501721567856474e-05,
|
| 351 |
+
"loss": 0.0686,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.9011406844106464,
|
| 356 |
+
"grad_norm": 0.7118890881538391,
|
| 357 |
+
"learning_rate": 9.474001334453612e-05,
|
| 358 |
+
"loss": 0.0648,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.9391634980988592,
|
| 363 |
+
"grad_norm": 0.4286198318004608,
|
| 364 |
+
"learning_rate": 9.445573472638085e-05,
|
| 365 |
+
"loss": 0.059,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.9771863117870723,
|
| 370 |
+
"grad_norm": 0.5651097297668457,
|
| 371 |
+
"learning_rate": 9.416442478689898e-05,
|
| 372 |
+
"loss": 0.0657,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 2.0152091254752853,
|
| 377 |
+
"grad_norm": 0.41622453927993774,
|
| 378 |
+
"learning_rate": 9.38661296009963e-05,
|
| 379 |
+
"loss": 0.0619,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.053231939163498,
|
| 384 |
+
"grad_norm": 0.40556129813194275,
|
| 385 |
+
"learning_rate": 9.3560896348397e-05,
|
| 386 |
+
"loss": 0.0598,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 2.091254752851711,
|
| 391 |
+
"grad_norm": 0.3720885217189789,
|
| 392 |
+
"learning_rate": 9.324877330618143e-05,
|
| 393 |
+
"loss": 0.0594,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.129277566539924,
|
| 398 |
+
"grad_norm": 0.3962866961956024,
|
| 399 |
+
"learning_rate": 9.292980984115047e-05,
|
| 400 |
+
"loss": 0.0638,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 2.167300380228137,
|
| 405 |
+
"grad_norm": 0.5914015173912048,
|
| 406 |
+
"learning_rate": 9.260405640201737e-05,
|
| 407 |
+
"loss": 0.0618,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 2.20532319391635,
|
| 412 |
+
"grad_norm": 0.3687211275100708,
|
| 413 |
+
"learning_rate": 9.227156451142862e-05,
|
| 414 |
+
"loss": 0.0565,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.2433460076045626,
|
| 419 |
+
"grad_norm": 0.4098389148712158,
|
| 420 |
+
"learning_rate": 9.193238675781482e-05,
|
| 421 |
+
"loss": 0.0596,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.2813688212927756,
|
| 426 |
+
"grad_norm": 0.43146541714668274,
|
| 427 |
+
"learning_rate": 9.15865767870731e-05,
|
| 428 |
+
"loss": 0.0595,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 2.3193916349809887,
|
| 433 |
+
"grad_norm": 0.4519984722137451,
|
| 434 |
+
"learning_rate": 9.123418929408225e-05,
|
| 435 |
+
"loss": 0.0535,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 2.3574144486692017,
|
| 440 |
+
"grad_norm": 0.41412219405174255,
|
| 441 |
+
"learning_rate": 9.087528001405194e-05,
|
| 442 |
+
"loss": 0.0569,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 2.3954372623574143,
|
| 447 |
+
"grad_norm": 0.3652435541152954,
|
| 448 |
+
"learning_rate": 9.05099057137073e-05,
|
| 449 |
+
"loss": 0.0551,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.4334600760456273,
|
| 454 |
+
"grad_norm": 0.49258697032928467,
|
| 455 |
+
"learning_rate": 9.013812418231056e-05,
|
| 456 |
+
"loss": 0.0553,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 2.4714828897338403,
|
| 461 |
+
"grad_norm": 0.4152522683143616,
|
| 462 |
+
"learning_rate": 8.975999422252071e-05,
|
| 463 |
+
"loss": 0.0542,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.5095057034220534,
|
| 468 |
+
"grad_norm": 0.6340664625167847,
|
| 469 |
+
"learning_rate": 8.937557564109308e-05,
|
| 470 |
+
"loss": 0.0555,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.5475285171102664,
|
| 475 |
+
"grad_norm": 0.2747473120689392,
|
| 476 |
+
"learning_rate": 8.898492923942007e-05,
|
| 477 |
+
"loss": 0.0499,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.585551330798479,
|
| 482 |
+
"grad_norm": 0.42373764514923096,
|
| 483 |
+
"learning_rate": 8.858811680391441e-05,
|
| 484 |
+
"loss": 0.0613,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 2.623574144486692,
|
| 489 |
+
"grad_norm": 0.46734005212783813,
|
| 490 |
+
"learning_rate": 8.818520109623687e-05,
|
| 491 |
+
"loss": 0.0572,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 2.661596958174905,
|
| 496 |
+
"grad_norm": 0.551827073097229,
|
| 497 |
+
"learning_rate": 8.777624584336943e-05,
|
| 498 |
+
"loss": 0.0663,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 2.6996197718631176,
|
| 503 |
+
"grad_norm": 0.3376104235649109,
|
| 504 |
+
"learning_rate": 8.736131572753614e-05,
|
| 505 |
+
"loss": 0.0576,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.7376425855513307,
|
| 510 |
+
"grad_norm": 0.4250059723854065,
|
| 511 |
+
"learning_rate": 8.694047637597245e-05,
|
| 512 |
+
"loss": 0.055,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.7756653992395437,
|
| 517 |
+
"grad_norm": 0.4221857190132141,
|
| 518 |
+
"learning_rate": 8.65137943505455e-05,
|
| 519 |
+
"loss": 0.0556,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.8136882129277567,
|
| 524 |
+
"grad_norm": 0.29908323287963867,
|
| 525 |
+
"learning_rate": 8.608133713722632e-05,
|
| 526 |
+
"loss": 0.0528,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 2.8517110266159698,
|
| 531 |
+
"grad_norm": 0.5638275742530823,
|
| 532 |
+
"learning_rate": 8.56431731354159e-05,
|
| 533 |
+
"loss": 0.0581,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.8897338403041823,
|
| 538 |
+
"grad_norm": 0.4976101815700531,
|
| 539 |
+
"learning_rate": 8.519937164712691e-05,
|
| 540 |
+
"loss": 0.0581,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 2.9277566539923954,
|
| 545 |
+
"grad_norm": 0.5850965976715088,
|
| 546 |
+
"learning_rate": 8.475000286602252e-05,
|
| 547 |
+
"loss": 0.0555,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.9657794676806084,
|
| 552 |
+
"grad_norm": 0.5351647734642029,
|
| 553 |
+
"learning_rate": 8.429513786631426e-05,
|
| 554 |
+
"loss": 0.0598,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.0038022813688214,
|
| 559 |
+
"grad_norm": 0.3065849840641022,
|
| 560 |
+
"learning_rate": 8.383484859152062e-05,
|
| 561 |
+
"loss": 0.0549,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 3.041825095057034,
|
| 566 |
+
"grad_norm": 0.47382888197898865,
|
| 567 |
+
"learning_rate": 8.336920784308813e-05,
|
| 568 |
+
"loss": 0.0528,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.079847908745247,
|
| 573 |
+
"grad_norm": 0.41480347514152527,
|
| 574 |
+
"learning_rate": 8.289828926887672e-05,
|
| 575 |
+
"loss": 0.0528,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 3.11787072243346,
|
| 580 |
+
"grad_norm": 0.4022481441497803,
|
| 581 |
+
"learning_rate": 8.24221673515113e-05,
|
| 582 |
+
"loss": 0.0523,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 3.155893536121673,
|
| 587 |
+
"grad_norm": 0.3361530900001526,
|
| 588 |
+
"learning_rate": 8.19409173966012e-05,
|
| 589 |
+
"loss": 0.056,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.1939163498098857,
|
| 594 |
+
"grad_norm": 0.6117600798606873,
|
| 595 |
+
"learning_rate": 8.145461552082947e-05,
|
| 596 |
+
"loss": 0.0601,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.2319391634980987,
|
| 601 |
+
"grad_norm": 0.5725025534629822,
|
| 602 |
+
"learning_rate": 8.096333863991402e-05,
|
| 603 |
+
"loss": 0.0547,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 3.2699619771863118,
|
| 608 |
+
"grad_norm": 0.41603338718414307,
|
| 609 |
+
"learning_rate": 8.046716445644219e-05,
|
| 610 |
+
"loss": 0.0487,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 3.307984790874525,
|
| 615 |
+
"grad_norm": 0.358055979013443,
|
| 616 |
+
"learning_rate": 7.996617144758094e-05,
|
| 617 |
+
"loss": 0.0512,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 3.346007604562738,
|
| 622 |
+
"grad_norm": 0.25234535336494446,
|
| 623 |
+
"learning_rate": 7.946043885266465e-05,
|
| 624 |
+
"loss": 0.0463,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 3.3840304182509504,
|
| 629 |
+
"grad_norm": 0.28671276569366455,
|
| 630 |
+
"learning_rate": 7.895004666066213e-05,
|
| 631 |
+
"loss": 0.0445,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.4220532319391634,
|
| 636 |
+
"grad_norm": 0.4477708339691162,
|
| 637 |
+
"learning_rate": 7.843507559752537e-05,
|
| 638 |
+
"loss": 0.048,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.4600760456273765,
|
| 643 |
+
"grad_norm": 0.4737355411052704,
|
| 644 |
+
"learning_rate": 7.791560711342135e-05,
|
| 645 |
+
"loss": 0.0475,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 3.4980988593155895,
|
| 650 |
+
"grad_norm": 0.4270668625831604,
|
| 651 |
+
"learning_rate": 7.739172336984967e-05,
|
| 652 |
+
"loss": 0.0493,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 3.5361216730038025,
|
| 657 |
+
"grad_norm": 0.3703446388244629,
|
| 658 |
+
"learning_rate": 7.68635072266475e-05,
|
| 659 |
+
"loss": 0.0496,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 3.574144486692015,
|
| 664 |
+
"grad_norm": 0.41201433539390564,
|
| 665 |
+
"learning_rate": 7.633104222888396e-05,
|
| 666 |
+
"loss": 0.0514,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 3.612167300380228,
|
| 671 |
+
"grad_norm": 0.8504674434661865,
|
| 672 |
+
"learning_rate": 7.579441259364642e-05,
|
| 673 |
+
"loss": 0.0546,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.650190114068441,
|
| 678 |
+
"grad_norm": 0.46347907185554504,
|
| 679 |
+
"learning_rate": 7.525370319672025e-05,
|
| 680 |
+
"loss": 0.0545,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 3.6882129277566538,
|
| 685 |
+
"grad_norm": 0.42974913120269775,
|
| 686 |
+
"learning_rate": 7.470899955916445e-05,
|
| 687 |
+
"loss": 0.0497,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 3.726235741444867,
|
| 692 |
+
"grad_norm": 0.3402682840824127,
|
| 693 |
+
"learning_rate": 7.416038783378539e-05,
|
| 694 |
+
"loss": 0.0529,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 3.76425855513308,
|
| 699 |
+
"grad_norm": 0.4092146158218384,
|
| 700 |
+
"learning_rate": 7.360795479151027e-05,
|
| 701 |
+
"loss": 0.0455,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.802281368821293,
|
| 706 |
+
"grad_norm": 0.35827386379241943,
|
| 707 |
+
"learning_rate": 7.305178780766332e-05,
|
| 708 |
+
"loss": 0.0488,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 3.840304182509506,
|
| 713 |
+
"grad_norm": 0.3754422664642334,
|
| 714 |
+
"learning_rate": 7.249197484814578e-05,
|
| 715 |
+
"loss": 0.055,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 3.8783269961977185,
|
| 720 |
+
"grad_norm": 0.2807624638080597,
|
| 721 |
+
"learning_rate": 7.192860445552316e-05,
|
| 722 |
+
"loss": 0.046,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 3.9163498098859315,
|
| 727 |
+
"grad_norm": 0.2535790503025055,
|
| 728 |
+
"learning_rate": 7.136176573502069e-05,
|
| 729 |
+
"loss": 0.042,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 3.9543726235741445,
|
| 734 |
+
"grad_norm": 0.3779744803905487,
|
| 735 |
+
"learning_rate": 7.079154834043013e-05,
|
| 736 |
+
"loss": 0.0465,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 3.9923954372623576,
|
| 741 |
+
"grad_norm": 0.4343276917934418,
|
| 742 |
+
"learning_rate": 7.021804245992972e-05,
|
| 743 |
+
"loss": 0.0416,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 4.030418250950571,
|
| 748 |
+
"grad_norm": 0.47514665126800537,
|
| 749 |
+
"learning_rate": 6.964133880181963e-05,
|
| 750 |
+
"loss": 0.0429,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 4.068441064638783,
|
| 755 |
+
"grad_norm": 0.36156734824180603,
|
| 756 |
+
"learning_rate": 6.906152858017501e-05,
|
| 757 |
+
"loss": 0.0431,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.106463878326996,
|
| 762 |
+
"grad_norm": 0.4116564393043518,
|
| 763 |
+
"learning_rate": 6.847870350041934e-05,
|
| 764 |
+
"loss": 0.0503,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 4.144486692015209,
|
| 769 |
+
"grad_norm": 0.3581191301345825,
|
| 770 |
+
"learning_rate": 6.789295574481969e-05,
|
| 771 |
+
"loss": 0.0424,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 4.182509505703422,
|
| 776 |
+
"grad_norm": 0.3734349012374878,
|
| 777 |
+
"learning_rate": 6.730437795790689e-05,
|
| 778 |
+
"loss": 0.0481,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.220532319391635,
|
| 783 |
+
"grad_norm": 0.30373069643974304,
|
| 784 |
+
"learning_rate": 6.671306323182239e-05,
|
| 785 |
+
"loss": 0.0403,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.258555133079848,
|
| 790 |
+
"grad_norm": 0.33292144536972046,
|
| 791 |
+
"learning_rate": 6.611910509159443e-05,
|
| 792 |
+
"loss": 0.0419,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 4.2965779467680605,
|
| 797 |
+
"grad_norm": 0.4385669231414795,
|
| 798 |
+
"learning_rate": 6.552259748034555e-05,
|
| 799 |
+
"loss": 0.049,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.334600760456274,
|
| 804 |
+
"grad_norm": 0.3568478524684906,
|
| 805 |
+
"learning_rate": 6.49236347444344e-05,
|
| 806 |
+
"loss": 0.0434,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 4.3726235741444865,
|
| 811 |
+
"grad_norm": 0.47680339217185974,
|
| 812 |
+
"learning_rate": 6.432231161853317e-05,
|
| 813 |
+
"loss": 0.0453,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 4.4106463878327,
|
| 818 |
+
"grad_norm": 0.4149623513221741,
|
| 819 |
+
"learning_rate": 6.371872321064414e-05,
|
| 820 |
+
"loss": 0.0494,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.448669201520913,
|
| 825 |
+
"grad_norm": 0.32099780440330505,
|
| 826 |
+
"learning_rate": 6.311296498705691e-05,
|
| 827 |
+
"loss": 0.0449,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 4.486692015209125,
|
| 832 |
+
"grad_norm": 0.2793867588043213,
|
| 833 |
+
"learning_rate": 6.250513275724896e-05,
|
| 834 |
+
"loss": 0.0444,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 4.524714828897339,
|
| 839 |
+
"grad_norm": 0.40786534547805786,
|
| 840 |
+
"learning_rate": 6.189532265873209e-05,
|
| 841 |
+
"loss": 0.0391,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.562737642585551,
|
| 846 |
+
"grad_norm": 0.3745206892490387,
|
| 847 |
+
"learning_rate": 6.128363114184676e-05,
|
| 848 |
+
"loss": 0.0384,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 4.600760456273765,
|
| 853 |
+
"grad_norm": 0.43438687920570374,
|
| 854 |
+
"learning_rate": 6.067015495450715e-05,
|
| 855 |
+
"loss": 0.0404,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 4.638783269961977,
|
| 860 |
+
"grad_norm": 0.4575098156929016,
|
| 861 |
+
"learning_rate": 6.005499112689905e-05,
|
| 862 |
+
"loss": 0.0475,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 4.67680608365019,
|
| 867 |
+
"grad_norm": 0.39668184518814087,
|
| 868 |
+
"learning_rate": 5.9438236956133076e-05,
|
| 869 |
+
"loss": 0.0428,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 4.714828897338403,
|
| 874 |
+
"grad_norm": 0.3513396382331848,
|
| 875 |
+
"learning_rate": 5.881998999085583e-05,
|
| 876 |
+
"loss": 0.0461,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 4.752851711026616,
|
| 881 |
+
"grad_norm": 0.5033044219017029,
|
| 882 |
+
"learning_rate": 5.820034801582101e-05,
|
| 883 |
+
"loss": 0.0447,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 4.7908745247148286,
|
| 888 |
+
"grad_norm": 0.43220236897468567,
|
| 889 |
+
"learning_rate": 5.7579409036423424e-05,
|
| 890 |
+
"loss": 0.0418,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 4.828897338403042,
|
| 895 |
+
"grad_norm": 0.3379021883010864,
|
| 896 |
+
"learning_rate": 5.6957271263198045e-05,
|
| 897 |
+
"loss": 0.0404,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 4.866920152091255,
|
| 902 |
+
"grad_norm": 0.3024837374687195,
|
| 903 |
+
"learning_rate": 5.633403309628645e-05,
|
| 904 |
+
"loss": 0.0396,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 4.904942965779467,
|
| 909 |
+
"grad_norm": 0.5330707430839539,
|
| 910 |
+
"learning_rate": 5.5709793109873584e-05,
|
| 911 |
+
"loss": 0.0443,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 4.942965779467681,
|
| 916 |
+
"grad_norm": 0.38940224051475525,
|
| 917 |
+
"learning_rate": 5.508465003659663e-05,
|
| 918 |
+
"loss": 0.0436,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 4.980988593155893,
|
| 923 |
+
"grad_norm": 0.4433404803276062,
|
| 924 |
+
"learning_rate": 5.445870275192918e-05,
|
| 925 |
+
"loss": 0.0345,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.019011406844107,
|
| 930 |
+
"grad_norm": 0.4395654797554016,
|
| 931 |
+
"learning_rate": 5.3832050258542474e-05,
|
| 932 |
+
"loss": 0.044,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 5.057034220532319,
|
| 937 |
+
"grad_norm": 0.39197906851768494,
|
| 938 |
+
"learning_rate": 5.320479167064687e-05,
|
| 939 |
+
"loss": 0.0431,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 5.095057034220532,
|
| 944 |
+
"grad_norm": 0.2870258390903473,
|
| 945 |
+
"learning_rate": 5.2577026198315315e-05,
|
| 946 |
+
"loss": 0.04,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 5.133079847908745,
|
| 951 |
+
"grad_norm": 0.28162604570388794,
|
| 952 |
+
"learning_rate": 5.1948853131791955e-05,
|
| 953 |
+
"loss": 0.0342,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 5.171102661596958,
|
| 958 |
+
"grad_norm": 0.4061802327632904,
|
| 959 |
+
"learning_rate": 5.132037182578785e-05,
|
| 960 |
+
"loss": 0.0381,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.2091254752851714,
|
| 965 |
+
"grad_norm": 0.41974517703056335,
|
| 966 |
+
"learning_rate": 5.069168168376663e-05,
|
| 967 |
+
"loss": 0.0388,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 5.247148288973384,
|
| 972 |
+
"grad_norm": 0.2790164053440094,
|
| 973 |
+
"learning_rate": 5.006288214222241e-05,
|
| 974 |
+
"loss": 0.0364,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 5.285171102661597,
|
| 979 |
+
"grad_norm": 0.23911532759666443,
|
| 980 |
+
"learning_rate": 4.943407265495238e-05,
|
| 981 |
+
"loss": 0.0383,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 5.32319391634981,
|
| 986 |
+
"grad_norm": 0.47133246064186096,
|
| 987 |
+
"learning_rate": 4.8805352677326854e-05,
|
| 988 |
+
"loss": 0.0424,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 5.361216730038023,
|
| 993 |
+
"grad_norm": 0.3483254909515381,
|
| 994 |
+
"learning_rate": 4.817682165055882e-05,
|
| 995 |
+
"loss": 0.033,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 5.399239543726236,
|
| 1000 |
+
"grad_norm": 0.6127725839614868,
|
| 1001 |
+
"learning_rate": 4.7548578985975984e-05,
|
| 1002 |
+
"loss": 0.0358,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.437262357414449,
|
| 1007 |
+
"grad_norm": 0.4287540912628174,
|
| 1008 |
+
"learning_rate": 4.692072404929739e-05,
|
| 1009 |
+
"loss": 0.0365,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.475285171102661,
|
| 1014 |
+
"grad_norm": 0.41925495862960815,
|
| 1015 |
+
"learning_rate": 4.629335614491725e-05,
|
| 1016 |
+
"loss": 0.033,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 5.513307984790875,
|
| 1021 |
+
"grad_norm": 0.4788515865802765,
|
| 1022 |
+
"learning_rate": 4.566657450019849e-05,
|
| 1023 |
+
"loss": 0.0342,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 5.551330798479087,
|
| 1028 |
+
"grad_norm": 0.42283207178115845,
|
| 1029 |
+
"learning_rate": 4.504047824977862e-05,
|
| 1030 |
+
"loss": 0.0368,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 5.589353612167301,
|
| 1035 |
+
"grad_norm": 0.4737829267978668,
|
| 1036 |
+
"learning_rate": 4.441516641989002e-05,
|
| 1037 |
+
"loss": 0.0358,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 5.6273764258555135,
|
| 1042 |
+
"grad_norm": 0.24999397993087769,
|
| 1043 |
+
"learning_rate": 4.3790737912697586e-05,
|
| 1044 |
+
"loss": 0.0388,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 5.665399239543726,
|
| 1049 |
+
"grad_norm": 0.3402468264102936,
|
| 1050 |
+
"learning_rate": 4.316729149065579e-05,
|
| 1051 |
+
"loss": 0.0388,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.7034220532319395,
|
| 1056 |
+
"grad_norm": 0.32290977239608765,
|
| 1057 |
+
"learning_rate": 4.254492576088816e-05,
|
| 1058 |
+
"loss": 0.0376,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 5.741444866920152,
|
| 1063 |
+
"grad_norm": 0.2910301685333252,
|
| 1064 |
+
"learning_rate": 4.192373915959093e-05,
|
| 1065 |
+
"loss": 0.0396,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 5.779467680608365,
|
| 1070 |
+
"grad_norm": 0.2554992437362671,
|
| 1071 |
+
"learning_rate": 4.130382993646407e-05,
|
| 1072 |
+
"loss": 0.0374,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 5.817490494296578,
|
| 1077 |
+
"grad_norm": 0.23737633228302002,
|
| 1078 |
+
"learning_rate": 4.068529613917154e-05,
|
| 1079 |
+
"loss": 0.0374,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 5.855513307984791,
|
| 1084 |
+
"grad_norm": 0.36381277441978455,
|
| 1085 |
+
"learning_rate": 4.006823559783374e-05,
|
| 1086 |
+
"loss": 0.0345,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 5.893536121673003,
|
| 1091 |
+
"grad_norm": 0.5745224952697754,
|
| 1092 |
+
"learning_rate": 3.94527459095542e-05,
|
| 1093 |
+
"loss": 0.0327,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.931558935361217,
|
| 1098 |
+
"grad_norm": 0.2845105528831482,
|
| 1099 |
+
"learning_rate": 3.883892442298318e-05,
|
| 1100 |
+
"loss": 0.0348,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 5.969581749049429,
|
| 1105 |
+
"grad_norm": 0.21019914746284485,
|
| 1106 |
+
"learning_rate": 3.822686822292052e-05,
|
| 1107 |
+
"loss": 0.0301,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 6.007604562737643,
|
| 1112 |
+
"grad_norm": 0.46327492594718933,
|
| 1113 |
+
"learning_rate": 3.761667411496037e-05,
|
| 1114 |
+
"loss": 0.0356,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 6.0456273764258555,
|
| 1119 |
+
"grad_norm": 0.4197891652584076,
|
| 1120 |
+
"learning_rate": 3.70084386101799e-05,
|
| 1121 |
+
"loss": 0.0349,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 6.083650190114068,
|
| 1126 |
+
"grad_norm": 0.7197537422180176,
|
| 1127 |
+
"learning_rate": 3.640225790987469e-05,
|
| 1128 |
+
"loss": 0.0346,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 6.1216730038022815,
|
| 1133 |
+
"grad_norm": 0.5813512802124023,
|
| 1134 |
+
"learning_rate": 3.5798227890343056e-05,
|
| 1135 |
+
"loss": 0.0357,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.159695817490494,
|
| 1140 |
+
"grad_norm": 0.29964444041252136,
|
| 1141 |
+
"learning_rate": 3.519644408772191e-05,
|
| 1142 |
+
"loss": 0.0336,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.197718631178708,
|
| 1147 |
+
"grad_norm": 0.3644579350948334,
|
| 1148 |
+
"learning_rate": 3.459700168287624e-05,
|
| 1149 |
+
"loss": 0.03,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 6.23574144486692,
|
| 1154 |
+
"grad_norm": 0.31609275937080383,
|
| 1155 |
+
"learning_rate": 3.399999548634494e-05,
|
| 1156 |
+
"loss": 0.0357,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 6.273764258555133,
|
| 1161 |
+
"grad_norm": 0.4375005066394806,
|
| 1162 |
+
"learning_rate": 3.340551992334505e-05,
|
| 1163 |
+
"loss": 0.0299,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 6.311787072243346,
|
| 1168 |
+
"grad_norm": 0.23590423166751862,
|
| 1169 |
+
"learning_rate": 3.2813669018837134e-05,
|
| 1170 |
+
"loss": 0.0298,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 6.349809885931559,
|
| 1175 |
+
"grad_norm": 0.21977555751800537,
|
| 1176 |
+
"learning_rate": 3.2224536382653844e-05,
|
| 1177 |
+
"loss": 0.0384,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.387832699619771,
|
| 1182 |
+
"grad_norm": 0.506341278553009,
|
| 1183 |
+
"learning_rate": 3.163821519469411e-05,
|
| 1184 |
+
"loss": 0.0319,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.425855513307985,
|
| 1189 |
+
"grad_norm": 0.42202290892601013,
|
| 1190 |
+
"learning_rate": 3.1054798190185406e-05,
|
| 1191 |
+
"loss": 0.0328,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 6.4638783269961975,
|
| 1196 |
+
"grad_norm": 0.2345847338438034,
|
| 1197 |
+
"learning_rate": 3.0474377645016356e-05,
|
| 1198 |
+
"loss": 0.0303,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 6.501901140684411,
|
| 1203 |
+
"grad_norm": 0.2911858558654785,
|
| 1204 |
+
"learning_rate": 2.9897045361141955e-05,
|
| 1205 |
+
"loss": 0.0316,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 6.5399239543726235,
|
| 1210 |
+
"grad_norm": 0.3863067626953125,
|
| 1211 |
+
"learning_rate": 2.9322892652063694e-05,
|
| 1212 |
+
"loss": 0.0294,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 6.577946768060836,
|
| 1217 |
+
"grad_norm": 0.32410648465156555,
|
| 1218 |
+
"learning_rate": 2.8752010328387056e-05,
|
| 1219 |
+
"loss": 0.028,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.61596958174905,
|
| 1224 |
+
"grad_norm": 0.3410639762878418,
|
| 1225 |
+
"learning_rate": 2.818448868345851e-05,
|
| 1226 |
+
"loss": 0.0416,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 6.653992395437262,
|
| 1231 |
+
"grad_norm": 0.27167966961860657,
|
| 1232 |
+
"learning_rate": 2.7620417479084275e-05,
|
| 1233 |
+
"loss": 0.0316,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 6.692015209125476,
|
| 1238 |
+
"grad_norm": 0.31542375683784485,
|
| 1239 |
+
"learning_rate": 2.7059885931333163e-05,
|
| 1240 |
+
"loss": 0.033,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 6.730038022813688,
|
| 1245 |
+
"grad_norm": 0.32455578446388245,
|
| 1246 |
+
"learning_rate": 2.650298269642578e-05,
|
| 1247 |
+
"loss": 0.0327,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 6.768060836501901,
|
| 1252 |
+
"grad_norm": 0.26266616582870483,
|
| 1253 |
+
"learning_rate": 2.59497958567123e-05,
|
| 1254 |
+
"loss": 0.0281,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 6.806083650190114,
|
| 1259 |
+
"grad_norm": 0.3133273720741272,
|
| 1260 |
+
"learning_rate": 2.540041290674091e-05,
|
| 1261 |
+
"loss": 0.0299,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 6.844106463878327,
|
| 1266 |
+
"grad_norm": 0.23578499257564545,
|
| 1267 |
+
"learning_rate": 2.4854920739419257e-05,
|
| 1268 |
+
"loss": 0.0298,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 6.8821292775665395,
|
| 1273 |
+
"grad_norm": 0.2830640375614166,
|
| 1274 |
+
"learning_rate": 2.4313405632271085e-05,
|
| 1275 |
+
"loss": 0.0308,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 6.920152091254753,
|
| 1280 |
+
"grad_norm": 0.20085899531841278,
|
| 1281 |
+
"learning_rate": 2.377595323379026e-05,
|
| 1282 |
+
"loss": 0.0317,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 6.9581749049429655,
|
| 1287 |
+
"grad_norm": 0.18935348093509674,
|
| 1288 |
+
"learning_rate": 2.3242648549894136e-05,
|
| 1289 |
+
"loss": 0.0274,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 6.996197718631179,
|
| 1294 |
+
"grad_norm": 0.33719274401664734,
|
| 1295 |
+
"learning_rate": 2.27135759304786e-05,
|
| 1296 |
+
"loss": 0.0322,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 7.034220532319392,
|
| 1301 |
+
"grad_norm": 0.4374171197414398,
|
| 1302 |
+
"learning_rate": 2.2188819056077032e-05,
|
| 1303 |
+
"loss": 0.0318,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 7.072243346007604,
|
| 1308 |
+
"grad_norm": 0.3346361815929413,
|
| 1309 |
+
"learning_rate": 2.1668460924624873e-05,
|
| 1310 |
+
"loss": 0.0314,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 7.110266159695818,
|
| 1315 |
+
"grad_norm": 0.328633189201355,
|
| 1316 |
+
"learning_rate": 2.1152583838332423e-05,
|
| 1317 |
+
"loss": 0.0335,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 7.14828897338403,
|
| 1322 |
+
"grad_norm": 0.3334602415561676,
|
| 1323 |
+
"learning_rate": 2.0641269390667435e-05,
|
| 1324 |
+
"loss": 0.029,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.186311787072244,
|
| 1329 |
+
"grad_norm": 0.2650391161441803,
|
| 1330 |
+
"learning_rate": 2.0134598453449998e-05,
|
| 1331 |
+
"loss": 0.0272,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 7.224334600760456,
|
| 1336 |
+
"grad_norm": 0.24867604672908783,
|
| 1337 |
+
"learning_rate": 1.963265116406147e-05,
|
| 1338 |
+
"loss": 0.0365,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 7.262357414448669,
|
| 1343 |
+
"grad_norm": 0.3868961036205292,
|
| 1344 |
+
"learning_rate": 1.9135506912769534e-05,
|
| 1345 |
+
"loss": 0.0271,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 7.300380228136882,
|
| 1350 |
+
"grad_norm": 0.20158067345619202,
|
| 1351 |
+
"learning_rate": 1.864324433017151e-05,
|
| 1352 |
+
"loss": 0.0277,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 7.338403041825095,
|
| 1357 |
+
"grad_norm": 0.32405081391334534,
|
| 1358 |
+
"learning_rate": 1.815594127475781e-05,
|
| 1359 |
+
"loss": 0.0297,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 7.3764258555133075,
|
| 1364 |
+
"grad_norm": 0.25651466846466064,
|
| 1365 |
+
"learning_rate": 1.7673674820597458e-05,
|
| 1366 |
+
"loss": 0.0327,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 7.414448669201521,
|
| 1371 |
+
"grad_norm": 0.44135525822639465,
|
| 1372 |
+
"learning_rate": 1.7196521245147667e-05,
|
| 1373 |
+
"loss": 0.0283,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 7.452471482889734,
|
| 1378 |
+
"grad_norm": 0.22677309811115265,
|
| 1379 |
+
"learning_rate": 1.6724556017189492e-05,
|
| 1380 |
+
"loss": 0.0303,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 7.490494296577947,
|
| 1385 |
+
"grad_norm": 0.27086296677589417,
|
| 1386 |
+
"learning_rate": 1.6257853784891348e-05,
|
| 1387 |
+
"loss": 0.03,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 7.52851711026616,
|
| 1392 |
+
"grad_norm": 0.38143807649612427,
|
| 1393 |
+
"learning_rate": 1.5796488364002317e-05,
|
| 1394 |
+
"loss": 0.0287,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 7.566539923954372,
|
| 1399 |
+
"grad_norm": 0.28342220187187195,
|
| 1400 |
+
"learning_rate": 1.5340532726177108e-05,
|
| 1401 |
+
"loss": 0.0273,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 7.604562737642586,
|
| 1406 |
+
"grad_norm": 0.3746427297592163,
|
| 1407 |
+
"learning_rate": 1.4890058987434435e-05,
|
| 1408 |
+
"loss": 0.0318,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 7.642585551330798,
|
| 1413 |
+
"grad_norm": 0.6539424061775208,
|
| 1414 |
+
"learning_rate": 1.4445138396750995e-05,
|
| 1415 |
+
"loss": 0.0265,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 7.680608365019012,
|
| 1420 |
+
"grad_norm": 0.2562049925327301,
|
| 1421 |
+
"learning_rate": 1.4005841324792296e-05,
|
| 1422 |
+
"loss": 0.026,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 7.718631178707224,
|
| 1427 |
+
"grad_norm": 0.2805837094783783,
|
| 1428 |
+
"learning_rate": 1.3572237252782527e-05,
|
| 1429 |
+
"loss": 0.0307,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 7.756653992395437,
|
| 1434 |
+
"grad_norm": 0.42774826288223267,
|
| 1435 |
+
"learning_rate": 1.3144394761515077e-05,
|
| 1436 |
+
"loss": 0.0277,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 7.79467680608365,
|
| 1441 |
+
"grad_norm": 0.26309627294540405,
|
| 1442 |
+
"learning_rate": 1.272238152050551e-05,
|
| 1443 |
+
"loss": 0.0352,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 7.832699619771863,
|
| 1448 |
+
"grad_norm": 0.32600271701812744,
|
| 1449 |
+
"learning_rate": 1.2306264277288654e-05,
|
| 1450 |
+
"loss": 0.0291,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 7.870722433460076,
|
| 1455 |
+
"grad_norm": 0.4313303828239441,
|
| 1456 |
+
"learning_rate": 1.1896108846861497e-05,
|
| 1457 |
+
"loss": 0.0301,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 7.908745247148289,
|
| 1462 |
+
"grad_norm": 0.30629509687423706,
|
| 1463 |
+
"learning_rate": 1.1491980101273553e-05,
|
| 1464 |
+
"loss": 0.0276,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 7.946768060836502,
|
| 1469 |
+
"grad_norm": 0.24339847266674042,
|
| 1470 |
+
"learning_rate": 1.1093941959366476e-05,
|
| 1471 |
+
"loss": 0.0306,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 7.984790874524715,
|
| 1476 |
+
"grad_norm": 0.307231068611145,
|
| 1477 |
+
"learning_rate": 1.070205737666427e-05,
|
| 1478 |
+
"loss": 0.0292,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 8.022813688212928,
|
| 1483 |
+
"grad_norm": 0.30481287837028503,
|
| 1484 |
+
"learning_rate": 1.0316388335416066e-05,
|
| 1485 |
+
"loss": 0.0268,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 8.060836501901141,
|
| 1490 |
+
"grad_norm": 0.2899107336997986,
|
| 1491 |
+
"learning_rate": 9.936995834792623e-06,
|
| 1492 |
+
"loss": 0.0223,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 8.098859315589353,
|
| 1497 |
+
"grad_norm": 0.2853327989578247,
|
| 1498 |
+
"learning_rate": 9.56393988123851e-06,
|
| 1499 |
+
"loss": 0.0288,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 8.136882129277566,
|
| 1504 |
+
"grad_norm": 0.28443998098373413,
|
| 1505 |
+
"learning_rate": 9.197279478981136e-06,
|
| 1506 |
+
"loss": 0.0274,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 8.17490494296578,
|
| 1511 |
+
"grad_norm": 0.24939852952957153,
|
| 1512 |
+
"learning_rate": 8.83707262069845e-06,
|
| 1513 |
+
"loss": 0.0255,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 8.212927756653992,
|
| 1518 |
+
"grad_norm": 0.28874605894088745,
|
| 1519 |
+
"learning_rate": 8.483376278346467e-06,
|
| 1520 |
+
"loss": 0.0238,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 8.250950570342205,
|
| 1525 |
+
"grad_norm": 0.3146332800388336,
|
| 1526 |
+
"learning_rate": 8.13624639414841e-06,
|
| 1527 |
+
"loss": 0.0255,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 8.288973384030419,
|
| 1532 |
+
"grad_norm": 0.33142825961112976,
|
| 1533 |
+
"learning_rate": 7.795737871746611e-06,
|
| 1534 |
+
"loss": 0.0259,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 8.326996197718632,
|
| 1539 |
+
"grad_norm": 0.22175247967243195,
|
| 1540 |
+
"learning_rate": 7.46190456751863e-06,
|
| 1541 |
+
"loss": 0.0274,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 8.365019011406844,
|
| 1546 |
+
"grad_norm": 0.3944449722766876,
|
| 1547 |
+
"learning_rate": 7.134799282059168e-06,
|
| 1548 |
+
"loss": 0.0295,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 8.403041825095057,
|
| 1553 |
+
"grad_norm": 0.25309473276138306,
|
| 1554 |
+
"learning_rate": 6.814473751828871e-06,
|
| 1555 |
+
"loss": 0.0278,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 8.44106463878327,
|
| 1560 |
+
"grad_norm": 0.27433377504348755,
|
| 1561 |
+
"learning_rate": 6.500978640971445e-06,
|
| 1562 |
+
"loss": 0.0305,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 8.479087452471482,
|
| 1567 |
+
"grad_norm": 0.2707982063293457,
|
| 1568 |
+
"learning_rate": 6.194363533300351e-06,
|
| 1569 |
+
"loss": 0.0302,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 8.517110266159696,
|
| 1574 |
+
"grad_norm": 0.4330613315105438,
|
| 1575 |
+
"learning_rate": 5.894676924456454e-06,
|
| 1576 |
+
"loss": 0.0251,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 8.55513307984791,
|
| 1581 |
+
"grad_norm": 0.32701870799064636,
|
| 1582 |
+
"learning_rate": 5.601966214237714e-06,
|
| 1583 |
+
"loss": 0.0278,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 8.593155893536121,
|
| 1588 |
+
"grad_norm": 0.181648850440979,
|
| 1589 |
+
"learning_rate": 5.316277699102223e-06,
|
| 1590 |
+
"loss": 0.0278,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 8.631178707224334,
|
| 1595 |
+
"grad_norm": 0.24093398451805115,
|
| 1596 |
+
"learning_rate": 5.037656564845711e-06,
|
| 1597 |
+
"loss": 0.0244,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 8.669201520912548,
|
| 1602 |
+
"grad_norm": 0.2580304741859436,
|
| 1603 |
+
"learning_rate": 4.76614687945483e-06,
|
| 1604 |
+
"loss": 0.0272,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 8.70722433460076,
|
| 1609 |
+
"grad_norm": 0.41510969400405884,
|
| 1610 |
+
"learning_rate": 4.5017915861370605e-06,
|
| 1611 |
+
"loss": 0.0294,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 8.745247148288973,
|
| 1616 |
+
"grad_norm": 0.4753967225551605,
|
| 1617 |
+
"learning_rate": 4.244632496528716e-06,
|
| 1618 |
+
"loss": 0.0259,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 8.783269961977187,
|
| 1623 |
+
"grad_norm": 0.23562125861644745,
|
| 1624 |
+
"learning_rate": 3.9947102840817595e-06,
|
| 1625 |
+
"loss": 0.0256,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 8.8212927756654,
|
| 1630 |
+
"grad_norm": 0.3789341449737549,
|
| 1631 |
+
"learning_rate": 3.75206447763074e-06,
|
| 1632 |
+
"loss": 0.032,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 8.859315589353612,
|
| 1637 |
+
"grad_norm": 0.36058199405670166,
|
| 1638 |
+
"learning_rate": 3.516733455140692e-06,
|
| 1639 |
+
"loss": 0.0284,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 8.897338403041825,
|
| 1644 |
+
"grad_norm": 0.42400893568992615,
|
| 1645 |
+
"learning_rate": 3.288754437637165e-06,
|
| 1646 |
+
"loss": 0.0295,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 8.935361216730039,
|
| 1651 |
+
"grad_norm": 0.33166080713272095,
|
| 1652 |
+
"learning_rate": 3.068163483319103e-06,
|
| 1653 |
+
"loss": 0.0291,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 8.97338403041825,
|
| 1658 |
+
"grad_norm": 0.5982539653778076,
|
| 1659 |
+
"learning_rate": 2.8549954818557877e-06,
|
| 1660 |
+
"loss": 0.0266,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 9.011406844106464,
|
| 1665 |
+
"grad_norm": 0.16449838876724243,
|
| 1666 |
+
"learning_rate": 2.649284148868475e-06,
|
| 1667 |
+
"loss": 0.026,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 9.049429657794677,
|
| 1672 |
+
"grad_norm": 0.5010789036750793,
|
| 1673 |
+
"learning_rate": 2.4510620205978e-06,
|
| 1674 |
+
"loss": 0.0277,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 9.087452471482889,
|
| 1679 |
+
"grad_norm": 0.24876657128334045,
|
| 1680 |
+
"learning_rate": 2.2603604487576922e-06,
|
| 1681 |
+
"loss": 0.0233,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 9.125475285171103,
|
| 1686 |
+
"grad_norm": 0.2562822997570038,
|
| 1687 |
+
"learning_rate": 2.0772095955766625e-06,
|
| 1688 |
+
"loss": 0.0265,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 9.163498098859316,
|
| 1693 |
+
"grad_norm": 0.29126426577568054,
|
| 1694 |
+
"learning_rate": 1.9016384290271805e-06,
|
| 1695 |
+
"loss": 0.0302,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 9.201520912547528,
|
| 1700 |
+
"grad_norm": 0.36330851912498474,
|
| 1701 |
+
"learning_rate": 1.7336747182440105e-06,
|
| 1702 |
+
"loss": 0.0244,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 9.239543726235741,
|
| 1707 |
+
"grad_norm": 0.3469178378582001,
|
| 1708 |
+
"learning_rate": 1.573345029132034e-06,
|
| 1709 |
+
"loss": 0.0308,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 9.277566539923955,
|
| 1714 |
+
"grad_norm": 0.3596096634864807,
|
| 1715 |
+
"learning_rate": 1.4206747201645777e-06,
|
| 1716 |
+
"loss": 0.0245,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 9.315589353612168,
|
| 1721 |
+
"grad_norm": 0.33200764656066895,
|
| 1722 |
+
"learning_rate": 1.2756879383725162e-06,
|
| 1723 |
+
"loss": 0.0276,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 9.35361216730038,
|
| 1728 |
+
"grad_norm": 0.26584237813949585,
|
| 1729 |
+
"learning_rate": 1.1384076155251223e-06,
|
| 1730 |
+
"loss": 0.0304,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 9.391634980988593,
|
| 1735 |
+
"grad_norm": 0.39666983485221863,
|
| 1736 |
+
"learning_rate": 1.0088554645030169e-06,
|
| 1737 |
+
"loss": 0.0265,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 9.429657794676807,
|
| 1742 |
+
"grad_norm": 0.3055102825164795,
|
| 1743 |
+
"learning_rate": 8.870519758640172e-07,
|
| 1744 |
+
"loss": 0.0228,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 9.467680608365018,
|
| 1749 |
+
"grad_norm": 0.2553930878639221,
|
| 1750 |
+
"learning_rate": 7.730164146022179e-07,
|
| 1751 |
+
"loss": 0.026,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 9.505703422053232,
|
| 1756 |
+
"grad_norm": 0.29695436358451843,
|
| 1757 |
+
"learning_rate": 6.667668171009889e-07,
|
| 1758 |
+
"loss": 0.0277,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 9.543726235741445,
|
| 1763 |
+
"grad_norm": 0.28649795055389404,
|
| 1764 |
+
"learning_rate": 5.683199882802081e-07,
|
| 1765 |
+
"loss": 0.0246,
|
| 1766 |
+
"step": 2510
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 9.581749049429657,
|
| 1770 |
+
"grad_norm": 0.26851099729537964,
|
| 1771 |
+
"learning_rate": 4.776914989383763e-07,
|
| 1772 |
+
"loss": 0.0264,
|
| 1773 |
+
"step": 2520
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 9.61977186311787,
|
| 1777 |
+
"grad_norm": 0.5427762269973755,
|
| 1778 |
+
"learning_rate": 3.948956832898143e-07,
|
| 1779 |
+
"loss": 0.0283,
|
| 1780 |
+
"step": 2530
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 9.657794676806084,
|
| 1784 |
+
"grad_norm": 0.30882003903388977,
|
| 1785 |
+
"learning_rate": 3.1994563669754953e-07,
|
| 1786 |
+
"loss": 0.024,
|
| 1787 |
+
"step": 2540
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 9.695817490494296,
|
| 1791 |
+
"grad_norm": 0.22906525433063507,
|
| 1792 |
+
"learning_rate": 2.528532136020445e-07,
|
| 1793 |
+
"loss": 0.0308,
|
| 1794 |
+
"step": 2550
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 9.73384030418251,
|
| 1798 |
+
"grad_norm": 0.2328147441148758,
|
| 1799 |
+
"learning_rate": 1.9362902564628583e-07,
|
| 1800 |
+
"loss": 0.0239,
|
| 1801 |
+
"step": 2560
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 9.771863117870723,
|
| 1805 |
+
"grad_norm": 0.3251258134841919,
|
| 1806 |
+
"learning_rate": 1.42282439997371e-07,
|
| 1807 |
+
"loss": 0.0296,
|
| 1808 |
+
"step": 2570
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 9.809885931558936,
|
| 1812 |
+
"grad_norm": 0.39408111572265625,
|
| 1813 |
+
"learning_rate": 9.882157786497149e-08,
|
| 1814 |
+
"loss": 0.0289,
|
| 1815 |
+
"step": 2580
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 9.847908745247148,
|
| 1819 |
+
"grad_norm": 0.17637042701244354,
|
| 1820 |
+
"learning_rate": 6.32533132168378e-08,
|
| 1821 |
+
"loss": 0.0217,
|
| 1822 |
+
"step": 2590
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 9.885931558935361,
|
| 1826 |
+
"grad_norm": 0.42764726281166077,
|
| 1827 |
+
"learning_rate": 3.558327169158604e-08,
|
| 1828 |
+
"loss": 0.0272,
|
| 1829 |
+
"step": 2600
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 9.923954372623575,
|
| 1833 |
+
"grad_norm": 0.27274248003959656,
|
| 1834 |
+
"learning_rate": 1.581582970890394e-08,
|
| 1835 |
+
"loss": 0.0248,
|
| 1836 |
+
"step": 2610
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 9.961977186311787,
|
| 1840 |
+
"grad_norm": 0.22425174713134766,
|
| 1841 |
+
"learning_rate": 3.954113777382418e-09,
|
| 1842 |
+
"loss": 0.0274,
|
| 1843 |
+
"step": 2620
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 10.0,
|
| 1847 |
+
"grad_norm": 0.19350482523441315,
|
| 1848 |
+
"learning_rate": 0.0,
|
| 1849 |
+
"loss": 0.0251,
|
| 1850 |
+
"step": 2630
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 10.0,
|
| 1854 |
+
"step": 2630,
|
| 1855 |
+
"total_flos": 2.7759792547300608e+17,
|
| 1856 |
+
"train_loss": 0.062387093805314923,
|
| 1857 |
+
"train_runtime": 2695.4145,
|
| 1858 |
+
"train_samples_per_second": 47.807,
|
| 1859 |
+
"train_steps_per_second": 0.976
|
| 1860 |
+
}
|
| 1861 |
+
],
|
| 1862 |
+
"logging_steps": 10,
|
| 1863 |
+
"max_steps": 2630,
|
| 1864 |
+
"num_input_tokens_seen": 0,
|
| 1865 |
+
"num_train_epochs": 10,
|
| 1866 |
+
"save_steps": 10000,
|
| 1867 |
+
"stateful_callbacks": {
|
| 1868 |
+
"TrainerControl": {
|
| 1869 |
+
"args": {
|
| 1870 |
+
"should_epoch_stop": false,
|
| 1871 |
+
"should_evaluate": false,
|
| 1872 |
+
"should_log": false,
|
| 1873 |
+
"should_save": true,
|
| 1874 |
+
"should_training_stop": true
|
| 1875 |
+
},
|
| 1876 |
+
"attributes": {}
|
| 1877 |
+
}
|
| 1878 |
+
},
|
| 1879 |
+
"total_flos": 2.7759792547300608e+17,
|
| 1880 |
+
"train_batch_size": 49,
|
| 1881 |
+
"trial_name": null,
|
| 1882 |
+
"trial_params": null
|
| 1883 |
+
}
|