Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +2345 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2345 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.2782892974889872,
|
| 3 |
+
"best_model_checkpoint": "./whisper-medium_new_data/checkpoint-8000",
|
| 4 |
+
"epoch": 3.261312678353037,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 8000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010191602119853241,
|
| 13 |
+
"grad_norm": 29.000947952270508,
|
| 14 |
+
"learning_rate": 2.1000000000000003e-07,
|
| 15 |
+
"loss": 3.777,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.020383204239706482,
|
| 20 |
+
"grad_norm": 16.35529136657715,
|
| 21 |
+
"learning_rate": 4.6000000000000004e-07,
|
| 22 |
+
"loss": 3.379,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.030574806359559722,
|
| 27 |
+
"grad_norm": 11.864029884338379,
|
| 28 |
+
"learning_rate": 7.1e-07,
|
| 29 |
+
"loss": 2.7049,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.040766408479412965,
|
| 34 |
+
"grad_norm": 9.735739707946777,
|
| 35 |
+
"learning_rate": 9.600000000000001e-07,
|
| 36 |
+
"loss": 2.1205,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.050958010599266204,
|
| 41 |
+
"grad_norm": 7.762232780456543,
|
| 42 |
+
"learning_rate": 1.21e-06,
|
| 43 |
+
"loss": 1.731,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.061149612719119444,
|
| 48 |
+
"grad_norm": 8.16604995727539,
|
| 49 |
+
"learning_rate": 1.46e-06,
|
| 50 |
+
"loss": 1.4442,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07134121483897268,
|
| 55 |
+
"grad_norm": 7.53511905670166,
|
| 56 |
+
"learning_rate": 1.7100000000000004e-06,
|
| 57 |
+
"loss": 1.244,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.08153281695882593,
|
| 62 |
+
"grad_norm": 7.6938557624816895,
|
| 63 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 64 |
+
"loss": 1.0535,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.09172441907867916,
|
| 69 |
+
"grad_norm": 7.2217912673950195,
|
| 70 |
+
"learning_rate": 2.21e-06,
|
| 71 |
+
"loss": 0.9467,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.10191602119853241,
|
| 76 |
+
"grad_norm": 6.709696292877197,
|
| 77 |
+
"learning_rate": 2.46e-06,
|
| 78 |
+
"loss": 0.8746,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.11210762331838565,
|
| 83 |
+
"grad_norm": 6.30186653137207,
|
| 84 |
+
"learning_rate": 2.7100000000000003e-06,
|
| 85 |
+
"loss": 0.7813,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.12229922543823889,
|
| 90 |
+
"grad_norm": 7.001236915588379,
|
| 91 |
+
"learning_rate": 2.96e-06,
|
| 92 |
+
"loss": 0.7312,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.13249082755809213,
|
| 97 |
+
"grad_norm": 6.0768232345581055,
|
| 98 |
+
"learning_rate": 3.21e-06,
|
| 99 |
+
"loss": 0.6602,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.14268242967794537,
|
| 104 |
+
"grad_norm": 6.0217790603637695,
|
| 105 |
+
"learning_rate": 3.46e-06,
|
| 106 |
+
"loss": 0.6294,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15287403179779863,
|
| 111 |
+
"grad_norm": 6.146008491516113,
|
| 112 |
+
"learning_rate": 3.7100000000000005e-06,
|
| 113 |
+
"loss": 0.6184,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.16306563391765186,
|
| 118 |
+
"grad_norm": 5.988298416137695,
|
| 119 |
+
"learning_rate": 3.96e-06,
|
| 120 |
+
"loss": 0.5581,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.1732572360375051,
|
| 125 |
+
"grad_norm": 6.24209451675415,
|
| 126 |
+
"learning_rate": 4.21e-06,
|
| 127 |
+
"loss": 0.526,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.18344883815735832,
|
| 132 |
+
"grad_norm": 5.728849411010742,
|
| 133 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 134 |
+
"loss": 0.4973,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.19364044027721158,
|
| 139 |
+
"grad_norm": 5.146097660064697,
|
| 140 |
+
"learning_rate": 4.71e-06,
|
| 141 |
+
"loss": 0.4877,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.20383204239706482,
|
| 146 |
+
"grad_norm": 5.125807285308838,
|
| 147 |
+
"learning_rate": 4.960000000000001e-06,
|
| 148 |
+
"loss": 0.4718,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.21402364451691805,
|
| 153 |
+
"grad_norm": 5.3624653816223145,
|
| 154 |
+
"learning_rate": 5.210000000000001e-06,
|
| 155 |
+
"loss": 0.455,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2242152466367713,
|
| 160 |
+
"grad_norm": 5.1377854347229,
|
| 161 |
+
"learning_rate": 5.460000000000001e-06,
|
| 162 |
+
"loss": 0.4426,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.23440684875662454,
|
| 167 |
+
"grad_norm": 5.11069393157959,
|
| 168 |
+
"learning_rate": 5.71e-06,
|
| 169 |
+
"loss": 0.4176,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.24459845087647777,
|
| 174 |
+
"grad_norm": 5.100331783294678,
|
| 175 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 176 |
+
"loss": 0.4044,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.254790052996331,
|
| 181 |
+
"grad_norm": 5.771068572998047,
|
| 182 |
+
"learning_rate": 6.210000000000001e-06,
|
| 183 |
+
"loss": 0.3994,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.26498165511618427,
|
| 188 |
+
"grad_norm": 5.4431986808776855,
|
| 189 |
+
"learning_rate": 6.460000000000001e-06,
|
| 190 |
+
"loss": 0.366,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.2751732572360375,
|
| 195 |
+
"grad_norm": 5.490501880645752,
|
| 196 |
+
"learning_rate": 6.710000000000001e-06,
|
| 197 |
+
"loss": 0.3742,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.28536485935589073,
|
| 202 |
+
"grad_norm": 5.428119659423828,
|
| 203 |
+
"learning_rate": 6.96e-06,
|
| 204 |
+
"loss": 0.3577,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.295556461475744,
|
| 209 |
+
"grad_norm": 4.746588230133057,
|
| 210 |
+
"learning_rate": 7.2100000000000004e-06,
|
| 211 |
+
"loss": 0.3465,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.30574806359559725,
|
| 216 |
+
"grad_norm": 4.743890762329102,
|
| 217 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 218 |
+
"loss": 0.3226,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.31593966571545046,
|
| 223 |
+
"grad_norm": 5.242619037628174,
|
| 224 |
+
"learning_rate": 7.71e-06,
|
| 225 |
+
"loss": 0.3185,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.3261312678353037,
|
| 230 |
+
"grad_norm": 4.178338050842285,
|
| 231 |
+
"learning_rate": 7.960000000000002e-06,
|
| 232 |
+
"loss": 0.32,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.336322869955157,
|
| 237 |
+
"grad_norm": 4.850783824920654,
|
| 238 |
+
"learning_rate": 8.210000000000001e-06,
|
| 239 |
+
"loss": 0.327,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.3465144720750102,
|
| 244 |
+
"grad_norm": 5.147852897644043,
|
| 245 |
+
"learning_rate": 8.46e-06,
|
| 246 |
+
"loss": 0.3334,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.35670607419486344,
|
| 251 |
+
"grad_norm": 4.720470905303955,
|
| 252 |
+
"learning_rate": 8.710000000000001e-06,
|
| 253 |
+
"loss": 0.2821,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.36689767631471665,
|
| 258 |
+
"grad_norm": 4.410367965698242,
|
| 259 |
+
"learning_rate": 8.96e-06,
|
| 260 |
+
"loss": 0.2894,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3770892784345699,
|
| 265 |
+
"grad_norm": 4.329741477966309,
|
| 266 |
+
"learning_rate": 9.210000000000002e-06,
|
| 267 |
+
"loss": 0.2792,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.38728088055442317,
|
| 272 |
+
"grad_norm": 4.4180803298950195,
|
| 273 |
+
"learning_rate": 9.460000000000001e-06,
|
| 274 |
+
"loss": 0.2842,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3974724826742764,
|
| 279 |
+
"grad_norm": 4.213972568511963,
|
| 280 |
+
"learning_rate": 9.71e-06,
|
| 281 |
+
"loss": 0.2854,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.40766408479412963,
|
| 286 |
+
"grad_norm": 3.454677104949951,
|
| 287 |
+
"learning_rate": 9.960000000000001e-06,
|
| 288 |
+
"loss": 0.2733,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.40766408479412963,
|
| 293 |
+
"eval_loss": 0.2585422396659851,
|
| 294 |
+
"eval_runtime": 2430.2946,
|
| 295 |
+
"eval_samples_per_second": 4.195,
|
| 296 |
+
"eval_steps_per_second": 0.131,
|
| 297 |
+
"eval_wer": 32.5923736874671,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.4178556869139829,
|
| 302 |
+
"grad_norm": 3.999333620071411,
|
| 303 |
+
"learning_rate": 9.970000000000001e-06,
|
| 304 |
+
"loss": 0.2649,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.4280472890338361,
|
| 309 |
+
"grad_norm": 3.658658504486084,
|
| 310 |
+
"learning_rate": 9.934285714285715e-06,
|
| 311 |
+
"loss": 0.2494,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.43823889115368936,
|
| 316 |
+
"grad_norm": 4.042171955108643,
|
| 317 |
+
"learning_rate": 9.89857142857143e-06,
|
| 318 |
+
"loss": 0.2626,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.4484304932735426,
|
| 323 |
+
"grad_norm": 4.251903057098389,
|
| 324 |
+
"learning_rate": 9.862857142857144e-06,
|
| 325 |
+
"loss": 0.2457,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.4586220953933958,
|
| 330 |
+
"grad_norm": 4.49993896484375,
|
| 331 |
+
"learning_rate": 9.827142857142859e-06,
|
| 332 |
+
"loss": 0.2598,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.4688136975132491,
|
| 337 |
+
"grad_norm": 4.043424129486084,
|
| 338 |
+
"learning_rate": 9.791428571428571e-06,
|
| 339 |
+
"loss": 0.2447,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.47900529963310234,
|
| 344 |
+
"grad_norm": 3.787559747695923,
|
| 345 |
+
"learning_rate": 9.755714285714286e-06,
|
| 346 |
+
"loss": 0.2328,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.48919690175295555,
|
| 351 |
+
"grad_norm": 3.875075340270996,
|
| 352 |
+
"learning_rate": 9.72e-06,
|
| 353 |
+
"loss": 0.2359,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.4993885038728088,
|
| 358 |
+
"grad_norm": 4.148637294769287,
|
| 359 |
+
"learning_rate": 9.684285714285715e-06,
|
| 360 |
+
"loss": 0.2234,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.509580105992662,
|
| 365 |
+
"grad_norm": 3.9952499866485596,
|
| 366 |
+
"learning_rate": 9.648571428571429e-06,
|
| 367 |
+
"loss": 0.225,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.5197717081125153,
|
| 372 |
+
"grad_norm": 3.0705387592315674,
|
| 373 |
+
"learning_rate": 9.612857142857144e-06,
|
| 374 |
+
"loss": 0.2183,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.5299633102323685,
|
| 379 |
+
"grad_norm": 3.5826079845428467,
|
| 380 |
+
"learning_rate": 9.577142857142858e-06,
|
| 381 |
+
"loss": 0.2125,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.5401549123522218,
|
| 386 |
+
"grad_norm": 3.478602170944214,
|
| 387 |
+
"learning_rate": 9.541428571428572e-06,
|
| 388 |
+
"loss": 0.2067,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.550346514472075,
|
| 393 |
+
"grad_norm": 3.646801233291626,
|
| 394 |
+
"learning_rate": 9.505714285714287e-06,
|
| 395 |
+
"loss": 0.2132,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.5605381165919282,
|
| 400 |
+
"grad_norm": 4.240411758422852,
|
| 401 |
+
"learning_rate": 9.47e-06,
|
| 402 |
+
"loss": 0.2077,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.5707297187117815,
|
| 407 |
+
"grad_norm": 3.531076669692993,
|
| 408 |
+
"learning_rate": 9.434285714285714e-06,
|
| 409 |
+
"loss": 0.1978,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.5809213208316347,
|
| 414 |
+
"grad_norm": 4.578576564788818,
|
| 415 |
+
"learning_rate": 9.39857142857143e-06,
|
| 416 |
+
"loss": 0.1976,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.591112922951488,
|
| 421 |
+
"grad_norm": 3.237501382827759,
|
| 422 |
+
"learning_rate": 9.362857142857143e-06,
|
| 423 |
+
"loss": 0.1829,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.6013045250713412,
|
| 428 |
+
"grad_norm": 3.9498815536499023,
|
| 429 |
+
"learning_rate": 9.327142857142857e-06,
|
| 430 |
+
"loss": 0.1926,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.6114961271911945,
|
| 435 |
+
"grad_norm": 3.386265277862549,
|
| 436 |
+
"learning_rate": 9.291428571428572e-06,
|
| 437 |
+
"loss": 0.1829,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.6216877293110477,
|
| 442 |
+
"grad_norm": 3.7739858627319336,
|
| 443 |
+
"learning_rate": 9.255714285714286e-06,
|
| 444 |
+
"loss": 0.1807,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.6318793314309009,
|
| 449 |
+
"grad_norm": 2.9203953742980957,
|
| 450 |
+
"learning_rate": 9.220000000000002e-06,
|
| 451 |
+
"loss": 0.1816,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.6420709335507542,
|
| 456 |
+
"grad_norm": 3.465407371520996,
|
| 457 |
+
"learning_rate": 9.184285714285715e-06,
|
| 458 |
+
"loss": 0.1739,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6522625356706074,
|
| 463 |
+
"grad_norm": 3.1187219619750977,
|
| 464 |
+
"learning_rate": 9.148571428571429e-06,
|
| 465 |
+
"loss": 0.177,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.6624541377904607,
|
| 470 |
+
"grad_norm": 3.4814960956573486,
|
| 471 |
+
"learning_rate": 9.112857142857142e-06,
|
| 472 |
+
"loss": 0.1718,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.672645739910314,
|
| 477 |
+
"grad_norm": 3.3516340255737305,
|
| 478 |
+
"learning_rate": 9.077142857142858e-06,
|
| 479 |
+
"loss": 0.1682,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.6828373420301671,
|
| 484 |
+
"grad_norm": 3.460319757461548,
|
| 485 |
+
"learning_rate": 9.041428571428572e-06,
|
| 486 |
+
"loss": 0.1645,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.6930289441500204,
|
| 491 |
+
"grad_norm": 2.6608901023864746,
|
| 492 |
+
"learning_rate": 9.005714285714287e-06,
|
| 493 |
+
"loss": 0.1721,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.7032205462698736,
|
| 498 |
+
"grad_norm": 3.2124173641204834,
|
| 499 |
+
"learning_rate": 8.97e-06,
|
| 500 |
+
"loss": 0.1647,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.7134121483897269,
|
| 505 |
+
"grad_norm": 3.007030725479126,
|
| 506 |
+
"learning_rate": 8.934285714285716e-06,
|
| 507 |
+
"loss": 0.163,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.7236037505095801,
|
| 512 |
+
"grad_norm": 2.7804579734802246,
|
| 513 |
+
"learning_rate": 8.89857142857143e-06,
|
| 514 |
+
"loss": 0.1732,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.7337953526294333,
|
| 519 |
+
"grad_norm": 3.615187644958496,
|
| 520 |
+
"learning_rate": 8.862857142857143e-06,
|
| 521 |
+
"loss": 0.1578,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.7439869547492866,
|
| 526 |
+
"grad_norm": 2.8256614208221436,
|
| 527 |
+
"learning_rate": 8.827142857142857e-06,
|
| 528 |
+
"loss": 0.1553,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.7541785568691398,
|
| 533 |
+
"grad_norm": 2.9422616958618164,
|
| 534 |
+
"learning_rate": 8.791428571428572e-06,
|
| 535 |
+
"loss": 0.1483,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.7643701589889931,
|
| 540 |
+
"grad_norm": 2.9863955974578857,
|
| 541 |
+
"learning_rate": 8.755714285714286e-06,
|
| 542 |
+
"loss": 0.1446,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.7745617611088463,
|
| 547 |
+
"grad_norm": 3.2235851287841797,
|
| 548 |
+
"learning_rate": 8.720000000000001e-06,
|
| 549 |
+
"loss": 0.1488,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.7847533632286996,
|
| 554 |
+
"grad_norm": 3.379194974899292,
|
| 555 |
+
"learning_rate": 8.684285714285715e-06,
|
| 556 |
+
"loss": 0.1433,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.7949449653485527,
|
| 561 |
+
"grad_norm": 3.080751895904541,
|
| 562 |
+
"learning_rate": 8.64857142857143e-06,
|
| 563 |
+
"loss": 0.1428,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.805136567468406,
|
| 568 |
+
"grad_norm": 2.6511690616607666,
|
| 569 |
+
"learning_rate": 8.612857142857144e-06,
|
| 570 |
+
"loss": 0.1351,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.8153281695882593,
|
| 575 |
+
"grad_norm": 3.0043816566467285,
|
| 576 |
+
"learning_rate": 8.577142857142858e-06,
|
| 577 |
+
"loss": 0.1527,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.8153281695882593,
|
| 582 |
+
"eval_loss": 0.12457678467035294,
|
| 583 |
+
"eval_runtime": 2416.9884,
|
| 584 |
+
"eval_samples_per_second": 4.218,
|
| 585 |
+
"eval_steps_per_second": 0.132,
|
| 586 |
+
"eval_wer": 16.723770120886936,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.8255197717081125,
|
| 591 |
+
"grad_norm": 2.428893804550171,
|
| 592 |
+
"learning_rate": 8.541428571428571e-06,
|
| 593 |
+
"loss": 0.1329,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.8357113738279658,
|
| 598 |
+
"grad_norm": 2.5849223136901855,
|
| 599 |
+
"learning_rate": 8.505714285714287e-06,
|
| 600 |
+
"loss": 0.1355,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.845902975947819,
|
| 605 |
+
"grad_norm": 2.7230465412139893,
|
| 606 |
+
"learning_rate": 8.47e-06,
|
| 607 |
+
"loss": 0.1409,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.8560945780676722,
|
| 612 |
+
"grad_norm": 2.889012336730957,
|
| 613 |
+
"learning_rate": 8.434285714285716e-06,
|
| 614 |
+
"loss": 0.1339,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.8662861801875255,
|
| 619 |
+
"grad_norm": 2.4034271240234375,
|
| 620 |
+
"learning_rate": 8.39857142857143e-06,
|
| 621 |
+
"loss": 0.1397,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.8764777823073787,
|
| 626 |
+
"grad_norm": 2.2975339889526367,
|
| 627 |
+
"learning_rate": 8.362857142857143e-06,
|
| 628 |
+
"loss": 0.1254,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.886669384427232,
|
| 633 |
+
"grad_norm": 2.7864370346069336,
|
| 634 |
+
"learning_rate": 8.327142857142858e-06,
|
| 635 |
+
"loss": 0.1298,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.8968609865470852,
|
| 640 |
+
"grad_norm": 3.3211889266967773,
|
| 641 |
+
"learning_rate": 8.291428571428572e-06,
|
| 642 |
+
"loss": 0.1195,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.9070525886669385,
|
| 647 |
+
"grad_norm": 2.9740707874298096,
|
| 648 |
+
"learning_rate": 8.255714285714287e-06,
|
| 649 |
+
"loss": 0.1236,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.9172441907867916,
|
| 654 |
+
"grad_norm": 2.8324942588806152,
|
| 655 |
+
"learning_rate": 8.220000000000001e-06,
|
| 656 |
+
"loss": 0.1256,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.9274357929066449,
|
| 661 |
+
"grad_norm": 2.7244651317596436,
|
| 662 |
+
"learning_rate": 8.184285714285715e-06,
|
| 663 |
+
"loss": 0.1232,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.9376273950264982,
|
| 668 |
+
"grad_norm": 2.6477315425872803,
|
| 669 |
+
"learning_rate": 8.148571428571428e-06,
|
| 670 |
+
"loss": 0.1194,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.9478189971463514,
|
| 675 |
+
"grad_norm": 2.737881660461426,
|
| 676 |
+
"learning_rate": 8.112857142857144e-06,
|
| 677 |
+
"loss": 0.1184,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.9580105992662047,
|
| 682 |
+
"grad_norm": 3.096505641937256,
|
| 683 |
+
"learning_rate": 8.077142857142857e-06,
|
| 684 |
+
"loss": 0.1316,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.9682022013860578,
|
| 689 |
+
"grad_norm": 2.7507877349853516,
|
| 690 |
+
"learning_rate": 8.041428571428573e-06,
|
| 691 |
+
"loss": 0.109,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.9783938035059111,
|
| 696 |
+
"grad_norm": 2.7306649684906006,
|
| 697 |
+
"learning_rate": 8.005714285714286e-06,
|
| 698 |
+
"loss": 0.1263,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.9885854056257644,
|
| 703 |
+
"grad_norm": 3.1028060913085938,
|
| 704 |
+
"learning_rate": 7.970000000000002e-06,
|
| 705 |
+
"loss": 0.1135,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.9987770077456176,
|
| 710 |
+
"grad_norm": 2.456820011138916,
|
| 711 |
+
"learning_rate": 7.934285714285715e-06,
|
| 712 |
+
"loss": 0.1192,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.0089686098654709,
|
| 717 |
+
"grad_norm": 2.0729124546051025,
|
| 718 |
+
"learning_rate": 7.898571428571429e-06,
|
| 719 |
+
"loss": 0.0789,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.019160211985324,
|
| 724 |
+
"grad_norm": 2.0284624099731445,
|
| 725 |
+
"learning_rate": 7.862857142857143e-06,
|
| 726 |
+
"loss": 0.0741,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.0293518141051774,
|
| 731 |
+
"grad_norm": 1.9724788665771484,
|
| 732 |
+
"learning_rate": 7.827142857142858e-06,
|
| 733 |
+
"loss": 0.068,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.0395434162250305,
|
| 738 |
+
"grad_norm": 2.1046948432922363,
|
| 739 |
+
"learning_rate": 7.791428571428572e-06,
|
| 740 |
+
"loss": 0.0785,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.049735018344884,
|
| 745 |
+
"grad_norm": 2.5051779747009277,
|
| 746 |
+
"learning_rate": 7.755714285714287e-06,
|
| 747 |
+
"loss": 0.0788,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.059926620464737,
|
| 752 |
+
"grad_norm": 2.253950357437134,
|
| 753 |
+
"learning_rate": 7.72e-06,
|
| 754 |
+
"loss": 0.0748,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.0701182225845902,
|
| 759 |
+
"grad_norm": 2.03273868560791,
|
| 760 |
+
"learning_rate": 7.684285714285716e-06,
|
| 761 |
+
"loss": 0.0795,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.0803098247044436,
|
| 766 |
+
"grad_norm": 1.9357993602752686,
|
| 767 |
+
"learning_rate": 7.64857142857143e-06,
|
| 768 |
+
"loss": 0.0688,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.0905014268242967,
|
| 773 |
+
"grad_norm": 2.5450971126556396,
|
| 774 |
+
"learning_rate": 7.612857142857143e-06,
|
| 775 |
+
"loss": 0.0807,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.10069302894415,
|
| 780 |
+
"grad_norm": 2.2401187419891357,
|
| 781 |
+
"learning_rate": 7.577142857142857e-06,
|
| 782 |
+
"loss": 0.0741,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.1108846310640033,
|
| 787 |
+
"grad_norm": 2.1422953605651855,
|
| 788 |
+
"learning_rate": 7.5414285714285715e-06,
|
| 789 |
+
"loss": 0.0761,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.1210762331838564,
|
| 794 |
+
"grad_norm": 2.204946994781494,
|
| 795 |
+
"learning_rate": 7.505714285714286e-06,
|
| 796 |
+
"loss": 0.0726,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.1312678353037098,
|
| 801 |
+
"grad_norm": 2.2459845542907715,
|
| 802 |
+
"learning_rate": 7.4700000000000005e-06,
|
| 803 |
+
"loss": 0.0751,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.141459437423563,
|
| 808 |
+
"grad_norm": 2.2025928497314453,
|
| 809 |
+
"learning_rate": 7.434285714285715e-06,
|
| 810 |
+
"loss": 0.074,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.1516510395434163,
|
| 815 |
+
"grad_norm": 1.94180166721344,
|
| 816 |
+
"learning_rate": 7.3985714285714295e-06,
|
| 817 |
+
"loss": 0.0768,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.1618426416632694,
|
| 822 |
+
"grad_norm": 2.4948198795318604,
|
| 823 |
+
"learning_rate": 7.362857142857144e-06,
|
| 824 |
+
"loss": 0.0737,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.1720342437831226,
|
| 829 |
+
"grad_norm": 2.136427640914917,
|
| 830 |
+
"learning_rate": 7.3271428571428585e-06,
|
| 831 |
+
"loss": 0.0711,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.182225845902976,
|
| 836 |
+
"grad_norm": 1.7289948463439941,
|
| 837 |
+
"learning_rate": 7.291428571428571e-06,
|
| 838 |
+
"loss": 0.0713,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.1924174480228291,
|
| 843 |
+
"grad_norm": 2.254673957824707,
|
| 844 |
+
"learning_rate": 7.255714285714286e-06,
|
| 845 |
+
"loss": 0.0696,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.2026090501426825,
|
| 850 |
+
"grad_norm": 2.0539391040802,
|
| 851 |
+
"learning_rate": 7.22e-06,
|
| 852 |
+
"loss": 0.0669,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.2128006522625356,
|
| 857 |
+
"grad_norm": 2.1277451515197754,
|
| 858 |
+
"learning_rate": 7.184285714285715e-06,
|
| 859 |
+
"loss": 0.0682,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.222992254382389,
|
| 864 |
+
"grad_norm": 1.7704741954803467,
|
| 865 |
+
"learning_rate": 7.148571428571429e-06,
|
| 866 |
+
"loss": 0.0655,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.222992254382389,
|
| 871 |
+
"eval_loss": 0.07764188200235367,
|
| 872 |
+
"eval_runtime": 2406.4856,
|
| 873 |
+
"eval_samples_per_second": 4.237,
|
| 874 |
+
"eval_steps_per_second": 0.133,
|
| 875 |
+
"eval_wer": 10.566755631077823,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.2331838565022422,
|
| 880 |
+
"grad_norm": 2.0602128505706787,
|
| 881 |
+
"learning_rate": 7.112857142857144e-06,
|
| 882 |
+
"loss": 0.0667,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.2433754586220953,
|
| 887 |
+
"grad_norm": 2.0332977771759033,
|
| 888 |
+
"learning_rate": 7.077142857142858e-06,
|
| 889 |
+
"loss": 0.0656,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.2535670607419487,
|
| 894 |
+
"grad_norm": 2.4050822257995605,
|
| 895 |
+
"learning_rate": 7.041428571428572e-06,
|
| 896 |
+
"loss": 0.0624,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.2637586628618018,
|
| 901 |
+
"grad_norm": 2.1430561542510986,
|
| 902 |
+
"learning_rate": 7.0057142857142865e-06,
|
| 903 |
+
"loss": 0.0805,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.273950264981655,
|
| 908 |
+
"grad_norm": 2.0134341716766357,
|
| 909 |
+
"learning_rate": 6.97e-06,
|
| 910 |
+
"loss": 0.0664,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.2841418671015083,
|
| 915 |
+
"grad_norm": 2.217721462249756,
|
| 916 |
+
"learning_rate": 6.934285714285715e-06,
|
| 917 |
+
"loss": 0.0745,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.2943334692213617,
|
| 922 |
+
"grad_norm": 1.5617090463638306,
|
| 923 |
+
"learning_rate": 6.898571428571429e-06,
|
| 924 |
+
"loss": 0.0648,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.3045250713412149,
|
| 929 |
+
"grad_norm": 1.9269059896469116,
|
| 930 |
+
"learning_rate": 6.862857142857144e-06,
|
| 931 |
+
"loss": 0.0618,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.314716673461068,
|
| 936 |
+
"grad_norm": 2.0952699184417725,
|
| 937 |
+
"learning_rate": 6.827142857142857e-06,
|
| 938 |
+
"loss": 0.0706,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.3249082755809214,
|
| 943 |
+
"grad_norm": 1.764196753501892,
|
| 944 |
+
"learning_rate": 6.791428571428572e-06,
|
| 945 |
+
"loss": 0.0658,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.3350998777007745,
|
| 950 |
+
"grad_norm": 2.4948887825012207,
|
| 951 |
+
"learning_rate": 6.755714285714286e-06,
|
| 952 |
+
"loss": 0.0573,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.3452914798206277,
|
| 957 |
+
"grad_norm": 2.2995638847351074,
|
| 958 |
+
"learning_rate": 6.720000000000001e-06,
|
| 959 |
+
"loss": 0.0584,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.355483081940481,
|
| 964 |
+
"grad_norm": 1.6779873371124268,
|
| 965 |
+
"learning_rate": 6.684285714285715e-06,
|
| 966 |
+
"loss": 0.0618,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.3656746840603342,
|
| 971 |
+
"grad_norm": 2.435180425643921,
|
| 972 |
+
"learning_rate": 6.648571428571429e-06,
|
| 973 |
+
"loss": 0.0611,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.3758662861801876,
|
| 978 |
+
"grad_norm": 1.7600586414337158,
|
| 979 |
+
"learning_rate": 6.612857142857143e-06,
|
| 980 |
+
"loss": 0.0638,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.3860578883000407,
|
| 985 |
+
"grad_norm": 2.179009199142456,
|
| 986 |
+
"learning_rate": 6.577142857142857e-06,
|
| 987 |
+
"loss": 0.0577,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.396249490419894,
|
| 992 |
+
"grad_norm": 1.8130829334259033,
|
| 993 |
+
"learning_rate": 6.541428571428572e-06,
|
| 994 |
+
"loss": 0.0572,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.4064410925397473,
|
| 999 |
+
"grad_norm": 2.2357475757598877,
|
| 1000 |
+
"learning_rate": 6.505714285714286e-06,
|
| 1001 |
+
"loss": 0.0617,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.4166326946596004,
|
| 1006 |
+
"grad_norm": 2.523681163787842,
|
| 1007 |
+
"learning_rate": 6.470000000000001e-06,
|
| 1008 |
+
"loss": 0.0592,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.4268242967794538,
|
| 1013 |
+
"grad_norm": 1.9920995235443115,
|
| 1014 |
+
"learning_rate": 6.434285714285715e-06,
|
| 1015 |
+
"loss": 0.0594,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.437015898899307,
|
| 1020 |
+
"grad_norm": 1.876678705215454,
|
| 1021 |
+
"learning_rate": 6.39857142857143e-06,
|
| 1022 |
+
"loss": 0.0588,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.4472075010191603,
|
| 1027 |
+
"grad_norm": 2.1072115898132324,
|
| 1028 |
+
"learning_rate": 6.3628571428571426e-06,
|
| 1029 |
+
"loss": 0.0584,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.4573991031390134,
|
| 1034 |
+
"grad_norm": 1.9492688179016113,
|
| 1035 |
+
"learning_rate": 6.327142857142857e-06,
|
| 1036 |
+
"loss": 0.0619,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.4675907052588668,
|
| 1041 |
+
"grad_norm": 1.9974353313446045,
|
| 1042 |
+
"learning_rate": 6.2914285714285716e-06,
|
| 1043 |
+
"loss": 0.0526,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.47778230737872,
|
| 1048 |
+
"grad_norm": 2.0630576610565186,
|
| 1049 |
+
"learning_rate": 6.255714285714286e-06,
|
| 1050 |
+
"loss": 0.0587,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.487973909498573,
|
| 1055 |
+
"grad_norm": 2.312988042831421,
|
| 1056 |
+
"learning_rate": 6.220000000000001e-06,
|
| 1057 |
+
"loss": 0.0539,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.4981655116184265,
|
| 1062 |
+
"grad_norm": 2.3259403705596924,
|
| 1063 |
+
"learning_rate": 6.184285714285715e-06,
|
| 1064 |
+
"loss": 0.0547,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.5083571137382796,
|
| 1069 |
+
"grad_norm": 1.7665891647338867,
|
| 1070 |
+
"learning_rate": 6.14857142857143e-06,
|
| 1071 |
+
"loss": 0.0546,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.5185487158581328,
|
| 1076 |
+
"grad_norm": 1.8702497482299805,
|
| 1077 |
+
"learning_rate": 6.112857142857144e-06,
|
| 1078 |
+
"loss": 0.059,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.5287403179779862,
|
| 1083 |
+
"grad_norm": 1.725127935409546,
|
| 1084 |
+
"learning_rate": 6.077142857142858e-06,
|
| 1085 |
+
"loss": 0.0593,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.5389319200978395,
|
| 1090 |
+
"grad_norm": 1.6888962984085083,
|
| 1091 |
+
"learning_rate": 6.0414285714285714e-06,
|
| 1092 |
+
"loss": 0.0492,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.5491235222176927,
|
| 1097 |
+
"grad_norm": 1.8041000366210938,
|
| 1098 |
+
"learning_rate": 6.005714285714286e-06,
|
| 1099 |
+
"loss": 0.0524,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.5593151243375458,
|
| 1104 |
+
"grad_norm": 2.0858354568481445,
|
| 1105 |
+
"learning_rate": 5.9700000000000004e-06,
|
| 1106 |
+
"loss": 0.0532,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.5695067264573992,
|
| 1111 |
+
"grad_norm": 2.3499724864959717,
|
| 1112 |
+
"learning_rate": 5.934285714285715e-06,
|
| 1113 |
+
"loss": 0.0509,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.5796983285772523,
|
| 1118 |
+
"grad_norm": 1.6974126100540161,
|
| 1119 |
+
"learning_rate": 5.8985714285714295e-06,
|
| 1120 |
+
"loss": 0.0525,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.5898899306971055,
|
| 1125 |
+
"grad_norm": 2.386164426803589,
|
| 1126 |
+
"learning_rate": 5.862857142857143e-06,
|
| 1127 |
+
"loss": 0.056,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.6000815328169589,
|
| 1132 |
+
"grad_norm": 2.2181949615478516,
|
| 1133 |
+
"learning_rate": 5.827142857142858e-06,
|
| 1134 |
+
"loss": 0.0532,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 1.610273134936812,
|
| 1139 |
+
"grad_norm": 2.883762836456299,
|
| 1140 |
+
"learning_rate": 5.791428571428572e-06,
|
| 1141 |
+
"loss": 0.0554,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.6204647370566652,
|
| 1146 |
+
"grad_norm": 2.2483813762664795,
|
| 1147 |
+
"learning_rate": 5.755714285714287e-06,
|
| 1148 |
+
"loss": 0.053,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.6306563391765185,
|
| 1153 |
+
"grad_norm": 1.992173433303833,
|
| 1154 |
+
"learning_rate": 5.72e-06,
|
| 1155 |
+
"loss": 0.0455,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.6306563391765185,
|
| 1160 |
+
"eval_loss": 0.05135625973343849,
|
| 1161 |
+
"eval_runtime": 2404.8469,
|
| 1162 |
+
"eval_samples_per_second": 4.24,
|
| 1163 |
+
"eval_steps_per_second": 0.133,
|
| 1164 |
+
"eval_wer": 6.767451954600445,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.640847941296372,
|
| 1169 |
+
"grad_norm": 1.7579491138458252,
|
| 1170 |
+
"learning_rate": 5.684285714285715e-06,
|
| 1171 |
+
"loss": 0.0521,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.651039543416225,
|
| 1176 |
+
"grad_norm": 1.9420897960662842,
|
| 1177 |
+
"learning_rate": 5.6485714285714285e-06,
|
| 1178 |
+
"loss": 0.0542,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.6612311455360782,
|
| 1183 |
+
"grad_norm": 2.2500171661376953,
|
| 1184 |
+
"learning_rate": 5.612857142857143e-06,
|
| 1185 |
+
"loss": 0.0516,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.6714227476559316,
|
| 1190 |
+
"grad_norm": 1.7925150394439697,
|
| 1191 |
+
"learning_rate": 5.5771428571428575e-06,
|
| 1192 |
+
"loss": 0.0493,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.6816143497757847,
|
| 1197 |
+
"grad_norm": 2.5120983123779297,
|
| 1198 |
+
"learning_rate": 5.541428571428572e-06,
|
| 1199 |
+
"loss": 0.0496,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.6918059518956379,
|
| 1204 |
+
"grad_norm": 1.6827411651611328,
|
| 1205 |
+
"learning_rate": 5.5057142857142865e-06,
|
| 1206 |
+
"loss": 0.048,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 1.7019975540154912,
|
| 1211 |
+
"grad_norm": 1.951802134513855,
|
| 1212 |
+
"learning_rate": 5.470000000000001e-06,
|
| 1213 |
+
"loss": 0.0491,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.7121891561353446,
|
| 1218 |
+
"grad_norm": 1.688637375831604,
|
| 1219 |
+
"learning_rate": 5.4342857142857155e-06,
|
| 1220 |
+
"loss": 0.0515,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.7223807582551978,
|
| 1225 |
+
"grad_norm": 2.034071207046509,
|
| 1226 |
+
"learning_rate": 5.398571428571428e-06,
|
| 1227 |
+
"loss": 0.0516,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 1.732572360375051,
|
| 1232 |
+
"grad_norm": 1.9581401348114014,
|
| 1233 |
+
"learning_rate": 5.362857142857143e-06,
|
| 1234 |
+
"loss": 0.0536,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.7427639624949043,
|
| 1239 |
+
"grad_norm": 1.8898799419403076,
|
| 1240 |
+
"learning_rate": 5.327142857142857e-06,
|
| 1241 |
+
"loss": 0.0491,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.7529555646147574,
|
| 1246 |
+
"grad_norm": 2.1761457920074463,
|
| 1247 |
+
"learning_rate": 5.291428571428572e-06,
|
| 1248 |
+
"loss": 0.0496,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 1.7631471667346106,
|
| 1253 |
+
"grad_norm": 1.9841150045394897,
|
| 1254 |
+
"learning_rate": 5.255714285714286e-06,
|
| 1255 |
+
"loss": 0.0469,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.773338768854464,
|
| 1260 |
+
"grad_norm": 1.3026888370513916,
|
| 1261 |
+
"learning_rate": 5.220000000000001e-06,
|
| 1262 |
+
"loss": 0.0482,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.7835303709743173,
|
| 1267 |
+
"grad_norm": 1.3047564029693604,
|
| 1268 |
+
"learning_rate": 5.184285714285715e-06,
|
| 1269 |
+
"loss": 0.0493,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.7937219730941703,
|
| 1274 |
+
"grad_norm": 1.6902995109558105,
|
| 1275 |
+
"learning_rate": 5.14857142857143e-06,
|
| 1276 |
+
"loss": 0.0494,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.8039135752140236,
|
| 1281 |
+
"grad_norm": 2.262953758239746,
|
| 1282 |
+
"learning_rate": 5.112857142857143e-06,
|
| 1283 |
+
"loss": 0.0476,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.814105177333877,
|
| 1288 |
+
"grad_norm": 1.777170181274414,
|
| 1289 |
+
"learning_rate": 5.077142857142857e-06,
|
| 1290 |
+
"loss": 0.0471,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 1.8242967794537301,
|
| 1295 |
+
"grad_norm": 1.9193710088729858,
|
| 1296 |
+
"learning_rate": 5.041428571428572e-06,
|
| 1297 |
+
"loss": 0.0467,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.8344883815735833,
|
| 1302 |
+
"grad_norm": 1.8790152072906494,
|
| 1303 |
+
"learning_rate": 5.005714285714286e-06,
|
| 1304 |
+
"loss": 0.0487,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.8446799836934367,
|
| 1309 |
+
"grad_norm": 2.3190362453460693,
|
| 1310 |
+
"learning_rate": 4.970000000000001e-06,
|
| 1311 |
+
"loss": 0.0466,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 1.8548715858132898,
|
| 1316 |
+
"grad_norm": 1.821876049041748,
|
| 1317 |
+
"learning_rate": 4.934285714285715e-06,
|
| 1318 |
+
"loss": 0.0456,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.865063187933143,
|
| 1323 |
+
"grad_norm": 1.3371105194091797,
|
| 1324 |
+
"learning_rate": 4.898571428571429e-06,
|
| 1325 |
+
"loss": 0.0414,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.8752547900529963,
|
| 1330 |
+
"grad_norm": 1.5593416690826416,
|
| 1331 |
+
"learning_rate": 4.862857142857143e-06,
|
| 1332 |
+
"loss": 0.0435,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.8854463921728497,
|
| 1337 |
+
"grad_norm": 1.9036990404129028,
|
| 1338 |
+
"learning_rate": 4.827142857142858e-06,
|
| 1339 |
+
"loss": 0.0448,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.8956379942927029,
|
| 1344 |
+
"grad_norm": 1.8190467357635498,
|
| 1345 |
+
"learning_rate": 4.7914285714285715e-06,
|
| 1346 |
+
"loss": 0.0388,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.905829596412556,
|
| 1351 |
+
"grad_norm": 1.7529155015945435,
|
| 1352 |
+
"learning_rate": 4.755714285714286e-06,
|
| 1353 |
+
"loss": 0.0421,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.9160211985324094,
|
| 1358 |
+
"grad_norm": 2.084097385406494,
|
| 1359 |
+
"learning_rate": 4.7200000000000005e-06,
|
| 1360 |
+
"loss": 0.0464,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.9262128006522625,
|
| 1365 |
+
"grad_norm": 1.569948673248291,
|
| 1366 |
+
"learning_rate": 4.684285714285714e-06,
|
| 1367 |
+
"loss": 0.0438,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.9364044027721157,
|
| 1372 |
+
"grad_norm": 2.0869407653808594,
|
| 1373 |
+
"learning_rate": 4.648571428571429e-06,
|
| 1374 |
+
"loss": 0.0421,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.946596004891969,
|
| 1379 |
+
"grad_norm": 1.8481653928756714,
|
| 1380 |
+
"learning_rate": 4.612857142857143e-06,
|
| 1381 |
+
"loss": 0.0429,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.9567876070118224,
|
| 1386 |
+
"grad_norm": 2.0351133346557617,
|
| 1387 |
+
"learning_rate": 4.577142857142858e-06,
|
| 1388 |
+
"loss": 0.0431,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.9669792091316753,
|
| 1393 |
+
"grad_norm": 1.88765549659729,
|
| 1394 |
+
"learning_rate": 4.541428571428571e-06,
|
| 1395 |
+
"loss": 0.0448,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.9771708112515287,
|
| 1400 |
+
"grad_norm": 1.8413023948669434,
|
| 1401 |
+
"learning_rate": 4.505714285714286e-06,
|
| 1402 |
+
"loss": 0.0391,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.987362413371382,
|
| 1407 |
+
"grad_norm": 2.0287082195281982,
|
| 1408 |
+
"learning_rate": 4.47e-06,
|
| 1409 |
+
"loss": 0.0417,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.9975540154912352,
|
| 1414 |
+
"grad_norm": 2.1331636905670166,
|
| 1415 |
+
"learning_rate": 4.434285714285715e-06,
|
| 1416 |
+
"loss": 0.0434,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.0077456176110884,
|
| 1421 |
+
"grad_norm": 1.133825421333313,
|
| 1422 |
+
"learning_rate": 4.3985714285714286e-06,
|
| 1423 |
+
"loss": 0.0246,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.0179372197309418,
|
| 1428 |
+
"grad_norm": 1.3025981187820435,
|
| 1429 |
+
"learning_rate": 4.362857142857143e-06,
|
| 1430 |
+
"loss": 0.017,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.028128821850795,
|
| 1435 |
+
"grad_norm": 1.008670687675476,
|
| 1436 |
+
"learning_rate": 4.327142857142858e-06,
|
| 1437 |
+
"loss": 0.0174,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.038320423970648,
|
| 1442 |
+
"grad_norm": 0.9607629179954529,
|
| 1443 |
+
"learning_rate": 4.291428571428572e-06,
|
| 1444 |
+
"loss": 0.0162,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.038320423970648,
|
| 1449 |
+
"eval_loss": 0.0353175513446331,
|
| 1450 |
+
"eval_runtime": 2408.5196,
|
| 1451 |
+
"eval_samples_per_second": 4.233,
|
| 1452 |
+
"eval_steps_per_second": 0.132,
|
| 1453 |
+
"eval_wer": 4.477157079135229,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 2.0485120260905014,
|
| 1458 |
+
"grad_norm": 1.6613744497299194,
|
| 1459 |
+
"learning_rate": 4.255714285714286e-06,
|
| 1460 |
+
"loss": 0.0213,
|
| 1461 |
+
"step": 5025
|
| 1462 |
+
},
|
| 1463 |
+
{
|
| 1464 |
+
"epoch": 2.058703628210355,
|
| 1465 |
+
"grad_norm": 0.6007469296455383,
|
| 1466 |
+
"learning_rate": 4.22e-06,
|
| 1467 |
+
"loss": 0.015,
|
| 1468 |
+
"step": 5050
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 2.0688952303302077,
|
| 1472 |
+
"grad_norm": 1.5992757081985474,
|
| 1473 |
+
"learning_rate": 4.184285714285715e-06,
|
| 1474 |
+
"loss": 0.0191,
|
| 1475 |
+
"step": 5075
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 2.079086832450061,
|
| 1479 |
+
"grad_norm": 0.913711667060852,
|
| 1480 |
+
"learning_rate": 4.148571428571429e-06,
|
| 1481 |
+
"loss": 0.0155,
|
| 1482 |
+
"step": 5100
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.0892784345699145,
|
| 1486 |
+
"grad_norm": 1.1569514274597168,
|
| 1487 |
+
"learning_rate": 4.112857142857144e-06,
|
| 1488 |
+
"loss": 0.0164,
|
| 1489 |
+
"step": 5125
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 2.099470036689768,
|
| 1493 |
+
"grad_norm": 1.0935266017913818,
|
| 1494 |
+
"learning_rate": 4.0771428571428574e-06,
|
| 1495 |
+
"loss": 0.0146,
|
| 1496 |
+
"step": 5150
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 2.1096616388096208,
|
| 1500 |
+
"grad_norm": 0.8869621157646179,
|
| 1501 |
+
"learning_rate": 4.041428571428572e-06,
|
| 1502 |
+
"loss": 0.0168,
|
| 1503 |
+
"step": 5175
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 2.119853240929474,
|
| 1507 |
+
"grad_norm": 1.3779696226119995,
|
| 1508 |
+
"learning_rate": 4.0057142857142864e-06,
|
| 1509 |
+
"loss": 0.0164,
|
| 1510 |
+
"step": 5200
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 2.1300448430493275,
|
| 1514 |
+
"grad_norm": 1.2132755517959595,
|
| 1515 |
+
"learning_rate": 3.97e-06,
|
| 1516 |
+
"loss": 0.0178,
|
| 1517 |
+
"step": 5225
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.1402364451691804,
|
| 1521 |
+
"grad_norm": 1.8169926404953003,
|
| 1522 |
+
"learning_rate": 3.934285714285715e-06,
|
| 1523 |
+
"loss": 0.0181,
|
| 1524 |
+
"step": 5250
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 2.150428047289034,
|
| 1528 |
+
"grad_norm": 0.7188290357589722,
|
| 1529 |
+
"learning_rate": 3.898571428571429e-06,
|
| 1530 |
+
"loss": 0.0172,
|
| 1531 |
+
"step": 5275
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 2.160619649408887,
|
| 1535 |
+
"grad_norm": 1.0363893508911133,
|
| 1536 |
+
"learning_rate": 3.862857142857143e-06,
|
| 1537 |
+
"loss": 0.0151,
|
| 1538 |
+
"step": 5300
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 2.1708112515287405,
|
| 1542 |
+
"grad_norm": 0.7467776536941528,
|
| 1543 |
+
"learning_rate": 3.827142857142857e-06,
|
| 1544 |
+
"loss": 0.0159,
|
| 1545 |
+
"step": 5325
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 2.1810028536485935,
|
| 1549 |
+
"grad_norm": 1.4140534400939941,
|
| 1550 |
+
"learning_rate": 3.7914285714285722e-06,
|
| 1551 |
+
"loss": 0.0179,
|
| 1552 |
+
"step": 5350
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 2.191194455768447,
|
| 1556 |
+
"grad_norm": 0.831071138381958,
|
| 1557 |
+
"learning_rate": 3.755714285714286e-06,
|
| 1558 |
+
"loss": 0.018,
|
| 1559 |
+
"step": 5375
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 2.2013860578883,
|
| 1563 |
+
"grad_norm": 0.978151261806488,
|
| 1564 |
+
"learning_rate": 3.7200000000000004e-06,
|
| 1565 |
+
"loss": 0.0164,
|
| 1566 |
+
"step": 5400
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 2.211577660008153,
|
| 1570 |
+
"grad_norm": 1.2061834335327148,
|
| 1571 |
+
"learning_rate": 3.684285714285715e-06,
|
| 1572 |
+
"loss": 0.0155,
|
| 1573 |
+
"step": 5425
|
| 1574 |
+
},
|
| 1575 |
+
{
|
| 1576 |
+
"epoch": 2.2217692621280065,
|
| 1577 |
+
"grad_norm": 1.0183775424957275,
|
| 1578 |
+
"learning_rate": 3.648571428571429e-06,
|
| 1579 |
+
"loss": 0.0166,
|
| 1580 |
+
"step": 5450
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 2.23196086424786,
|
| 1584 |
+
"grad_norm": 0.7747207283973694,
|
| 1585 |
+
"learning_rate": 3.612857142857143e-06,
|
| 1586 |
+
"loss": 0.017,
|
| 1587 |
+
"step": 5475
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 2.242152466367713,
|
| 1591 |
+
"grad_norm": 1.3631898164749146,
|
| 1592 |
+
"learning_rate": 3.5771428571428576e-06,
|
| 1593 |
+
"loss": 0.0147,
|
| 1594 |
+
"step": 5500
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 2.252344068487566,
|
| 1598 |
+
"grad_norm": 2.3850202560424805,
|
| 1599 |
+
"learning_rate": 3.5414285714285716e-06,
|
| 1600 |
+
"loss": 0.0145,
|
| 1601 |
+
"step": 5525
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 2.2625356706074196,
|
| 1605 |
+
"grad_norm": 1.2588169574737549,
|
| 1606 |
+
"learning_rate": 3.505714285714286e-06,
|
| 1607 |
+
"loss": 0.0191,
|
| 1608 |
+
"step": 5550
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 2.2727272727272725,
|
| 1612 |
+
"grad_norm": 1.3466730117797852,
|
| 1613 |
+
"learning_rate": 3.4700000000000002e-06,
|
| 1614 |
+
"loss": 0.0177,
|
| 1615 |
+
"step": 5575
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 2.282918874847126,
|
| 1619 |
+
"grad_norm": 1.3090860843658447,
|
| 1620 |
+
"learning_rate": 3.4342857142857143e-06,
|
| 1621 |
+
"loss": 0.014,
|
| 1622 |
+
"step": 5600
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.2931104769669792,
|
| 1626 |
+
"grad_norm": 1.4974130392074585,
|
| 1627 |
+
"learning_rate": 3.398571428571429e-06,
|
| 1628 |
+
"loss": 0.0152,
|
| 1629 |
+
"step": 5625
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.3033020790868326,
|
| 1633 |
+
"grad_norm": 1.342842936515808,
|
| 1634 |
+
"learning_rate": 3.3628571428571433e-06,
|
| 1635 |
+
"loss": 0.0165,
|
| 1636 |
+
"step": 5650
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"epoch": 2.3134936812066855,
|
| 1640 |
+
"grad_norm": 0.9690020084381104,
|
| 1641 |
+
"learning_rate": 3.327142857142858e-06,
|
| 1642 |
+
"loss": 0.0146,
|
| 1643 |
+
"step": 5675
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 2.323685283326539,
|
| 1647 |
+
"grad_norm": 0.4699741303920746,
|
| 1648 |
+
"learning_rate": 3.2914285714285715e-06,
|
| 1649 |
+
"loss": 0.0178,
|
| 1650 |
+
"step": 5700
|
| 1651 |
+
},
|
| 1652 |
+
{
|
| 1653 |
+
"epoch": 2.3338768854463923,
|
| 1654 |
+
"grad_norm": 1.166608214378357,
|
| 1655 |
+
"learning_rate": 3.255714285714286e-06,
|
| 1656 |
+
"loss": 0.0152,
|
| 1657 |
+
"step": 5725
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 2.344068487566245,
|
| 1661 |
+
"grad_norm": 0.7220166921615601,
|
| 1662 |
+
"learning_rate": 3.2200000000000005e-06,
|
| 1663 |
+
"loss": 0.0142,
|
| 1664 |
+
"step": 5750
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 2.3542600896860986,
|
| 1668 |
+
"grad_norm": 0.8474377989768982,
|
| 1669 |
+
"learning_rate": 3.1842857142857146e-06,
|
| 1670 |
+
"loss": 0.0186,
|
| 1671 |
+
"step": 5775
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 2.364451691805952,
|
| 1675 |
+
"grad_norm": 1.246468186378479,
|
| 1676 |
+
"learning_rate": 3.1485714285714287e-06,
|
| 1677 |
+
"loss": 0.0143,
|
| 1678 |
+
"step": 5800
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 2.3746432939258053,
|
| 1682 |
+
"grad_norm": 0.9267088770866394,
|
| 1683 |
+
"learning_rate": 3.112857142857143e-06,
|
| 1684 |
+
"loss": 0.0157,
|
| 1685 |
+
"step": 5825
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 2.3848348960456582,
|
| 1689 |
+
"grad_norm": 1.0727224349975586,
|
| 1690 |
+
"learning_rate": 3.0771428571428573e-06,
|
| 1691 |
+
"loss": 0.0145,
|
| 1692 |
+
"step": 5850
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 2.3950264981655116,
|
| 1696 |
+
"grad_norm": 1.3142344951629639,
|
| 1697 |
+
"learning_rate": 3.0414285714285718e-06,
|
| 1698 |
+
"loss": 0.0133,
|
| 1699 |
+
"step": 5875
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 2.405218100285365,
|
| 1703 |
+
"grad_norm": 1.3349863290786743,
|
| 1704 |
+
"learning_rate": 3.005714285714286e-06,
|
| 1705 |
+
"loss": 0.0137,
|
| 1706 |
+
"step": 5900
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.415409702405218,
|
| 1710 |
+
"grad_norm": 0.7742011547088623,
|
| 1711 |
+
"learning_rate": 2.97e-06,
|
| 1712 |
+
"loss": 0.0149,
|
| 1713 |
+
"step": 5925
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 2.4256013045250713,
|
| 1717 |
+
"grad_norm": 1.1534104347229004,
|
| 1718 |
+
"learning_rate": 2.9342857142857144e-06,
|
| 1719 |
+
"loss": 0.0156,
|
| 1720 |
+
"step": 5950
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 2.4357929066449246,
|
| 1724 |
+
"grad_norm": 1.1034826040267944,
|
| 1725 |
+
"learning_rate": 2.898571428571429e-06,
|
| 1726 |
+
"loss": 0.0136,
|
| 1727 |
+
"step": 5975
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 2.445984508764778,
|
| 1731 |
+
"grad_norm": 1.3665175437927246,
|
| 1732 |
+
"learning_rate": 2.8628571428571435e-06,
|
| 1733 |
+
"loss": 0.0129,
|
| 1734 |
+
"step": 6000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.445984508764778,
|
| 1738 |
+
"eval_loss": 0.027366982772946358,
|
| 1739 |
+
"eval_runtime": 2410.6864,
|
| 1740 |
+
"eval_samples_per_second": 4.23,
|
| 1741 |
+
"eval_steps_per_second": 0.132,
|
| 1742 |
+
"eval_wer": 3.436365819195996,
|
| 1743 |
+
"step": 6000
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.456176110884631,
|
| 1747 |
+
"grad_norm": 0.9552819728851318,
|
| 1748 |
+
"learning_rate": 2.827142857142857e-06,
|
| 1749 |
+
"loss": 0.015,
|
| 1750 |
+
"step": 6025
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.4663677130044843,
|
| 1754 |
+
"grad_norm": 0.8750997185707092,
|
| 1755 |
+
"learning_rate": 2.7914285714285716e-06,
|
| 1756 |
+
"loss": 0.014,
|
| 1757 |
+
"step": 6050
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 2.4765593151243377,
|
| 1761 |
+
"grad_norm": 1.3504140377044678,
|
| 1762 |
+
"learning_rate": 2.755714285714286e-06,
|
| 1763 |
+
"loss": 0.0172,
|
| 1764 |
+
"step": 6075
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.4867509172441906,
|
| 1768 |
+
"grad_norm": 1.59817373752594,
|
| 1769 |
+
"learning_rate": 2.7200000000000002e-06,
|
| 1770 |
+
"loss": 0.0153,
|
| 1771 |
+
"step": 6100
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 2.496942519364044,
|
| 1775 |
+
"grad_norm": 1.6574933528900146,
|
| 1776 |
+
"learning_rate": 2.6842857142857143e-06,
|
| 1777 |
+
"loss": 0.0167,
|
| 1778 |
+
"step": 6125
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.5071341214838974,
|
| 1782 |
+
"grad_norm": 0.7508680820465088,
|
| 1783 |
+
"learning_rate": 2.648571428571429e-06,
|
| 1784 |
+
"loss": 0.013,
|
| 1785 |
+
"step": 6150
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 2.5173257236037507,
|
| 1789 |
+
"grad_norm": 1.0604500770568848,
|
| 1790 |
+
"learning_rate": 2.612857142857143e-06,
|
| 1791 |
+
"loss": 0.0162,
|
| 1792 |
+
"step": 6175
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.5275173257236037,
|
| 1796 |
+
"grad_norm": 1.3438018560409546,
|
| 1797 |
+
"learning_rate": 2.5771428571428574e-06,
|
| 1798 |
+
"loss": 0.0141,
|
| 1799 |
+
"step": 6200
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.537708927843457,
|
| 1803 |
+
"grad_norm": 1.3562772274017334,
|
| 1804 |
+
"learning_rate": 2.541428571428572e-06,
|
| 1805 |
+
"loss": 0.0134,
|
| 1806 |
+
"step": 6225
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.54790052996331,
|
| 1810 |
+
"grad_norm": 0.681122362613678,
|
| 1811 |
+
"learning_rate": 2.5057142857142856e-06,
|
| 1812 |
+
"loss": 0.0123,
|
| 1813 |
+
"step": 6250
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 2.5580921320831633,
|
| 1817 |
+
"grad_norm": 0.8060537576675415,
|
| 1818 |
+
"learning_rate": 2.47e-06,
|
| 1819 |
+
"loss": 0.0143,
|
| 1820 |
+
"step": 6275
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 2.5682837342030167,
|
| 1824 |
+
"grad_norm": 0.9794915318489075,
|
| 1825 |
+
"learning_rate": 2.4342857142857146e-06,
|
| 1826 |
+
"loss": 0.0128,
|
| 1827 |
+
"step": 6300
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.57847533632287,
|
| 1831 |
+
"grad_norm": 0.7749600410461426,
|
| 1832 |
+
"learning_rate": 2.3985714285714287e-06,
|
| 1833 |
+
"loss": 0.0132,
|
| 1834 |
+
"step": 6325
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.5886669384427234,
|
| 1838 |
+
"grad_norm": 0.808312177658081,
|
| 1839 |
+
"learning_rate": 2.362857142857143e-06,
|
| 1840 |
+
"loss": 0.0166,
|
| 1841 |
+
"step": 6350
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 2.5988585405625764,
|
| 1845 |
+
"grad_norm": 1.0200546979904175,
|
| 1846 |
+
"learning_rate": 2.3271428571428572e-06,
|
| 1847 |
+
"loss": 0.0138,
|
| 1848 |
+
"step": 6375
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 2.6090501426824297,
|
| 1852 |
+
"grad_norm": 1.4276586771011353,
|
| 1853 |
+
"learning_rate": 2.2914285714285718e-06,
|
| 1854 |
+
"loss": 0.0119,
|
| 1855 |
+
"step": 6400
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 2.6192417448022827,
|
| 1859 |
+
"grad_norm": 1.1479567289352417,
|
| 1860 |
+
"learning_rate": 2.255714285714286e-06,
|
| 1861 |
+
"loss": 0.0127,
|
| 1862 |
+
"step": 6425
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 2.629433346922136,
|
| 1866 |
+
"grad_norm": 0.9421886205673218,
|
| 1867 |
+
"learning_rate": 2.2200000000000003e-06,
|
| 1868 |
+
"loss": 0.0132,
|
| 1869 |
+
"step": 6450
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.6396249490419894,
|
| 1873 |
+
"grad_norm": 0.573397696018219,
|
| 1874 |
+
"learning_rate": 2.1842857142857144e-06,
|
| 1875 |
+
"loss": 0.0131,
|
| 1876 |
+
"step": 6475
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 2.649816551161843,
|
| 1880 |
+
"grad_norm": 1.2187764644622803,
|
| 1881 |
+
"learning_rate": 2.148571428571429e-06,
|
| 1882 |
+
"loss": 0.0125,
|
| 1883 |
+
"step": 6500
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 2.660008153281696,
|
| 1887 |
+
"grad_norm": 0.7359505891799927,
|
| 1888 |
+
"learning_rate": 2.112857142857143e-06,
|
| 1889 |
+
"loss": 0.0141,
|
| 1890 |
+
"step": 6525
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.670199755401549,
|
| 1894 |
+
"grad_norm": 1.8559486865997314,
|
| 1895 |
+
"learning_rate": 2.077142857142857e-06,
|
| 1896 |
+
"loss": 0.0142,
|
| 1897 |
+
"step": 6550
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 2.6803913575214025,
|
| 1901 |
+
"grad_norm": 1.0747308731079102,
|
| 1902 |
+
"learning_rate": 2.0414285714285716e-06,
|
| 1903 |
+
"loss": 0.0105,
|
| 1904 |
+
"step": 6575
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 2.6905829596412554,
|
| 1908 |
+
"grad_norm": 0.8565563559532166,
|
| 1909 |
+
"learning_rate": 2.0057142857142857e-06,
|
| 1910 |
+
"loss": 0.0144,
|
| 1911 |
+
"step": 6600
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 2.7007745617611087,
|
| 1915 |
+
"grad_norm": 1.1864107847213745,
|
| 1916 |
+
"learning_rate": 1.97e-06,
|
| 1917 |
+
"loss": 0.013,
|
| 1918 |
+
"step": 6625
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 2.710966163880962,
|
| 1922 |
+
"grad_norm": 0.6643468141555786,
|
| 1923 |
+
"learning_rate": 1.9342857142857143e-06,
|
| 1924 |
+
"loss": 0.0149,
|
| 1925 |
+
"step": 6650
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 2.7211577660008155,
|
| 1929 |
+
"grad_norm": 0.6889024376869202,
|
| 1930 |
+
"learning_rate": 1.8985714285714288e-06,
|
| 1931 |
+
"loss": 0.0115,
|
| 1932 |
+
"step": 6675
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 2.7313493681206684,
|
| 1936 |
+
"grad_norm": 1.5234169960021973,
|
| 1937 |
+
"learning_rate": 1.8628571428571429e-06,
|
| 1938 |
+
"loss": 0.0139,
|
| 1939 |
+
"step": 6700
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 2.741540970240522,
|
| 1943 |
+
"grad_norm": 1.095461368560791,
|
| 1944 |
+
"learning_rate": 1.8271428571428574e-06,
|
| 1945 |
+
"loss": 0.0132,
|
| 1946 |
+
"step": 6725
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.751732572360375,
|
| 1950 |
+
"grad_norm": 1.7404385805130005,
|
| 1951 |
+
"learning_rate": 1.7914285714285715e-06,
|
| 1952 |
+
"loss": 0.0135,
|
| 1953 |
+
"step": 6750
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 2.761924174480228,
|
| 1957 |
+
"grad_norm": 0.8274975419044495,
|
| 1958 |
+
"learning_rate": 1.755714285714286e-06,
|
| 1959 |
+
"loss": 0.013,
|
| 1960 |
+
"step": 6775
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 2.7721157766000815,
|
| 1964 |
+
"grad_norm": 0.8858354091644287,
|
| 1965 |
+
"learning_rate": 1.72e-06,
|
| 1966 |
+
"loss": 0.0118,
|
| 1967 |
+
"step": 6800
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 2.782307378719935,
|
| 1971 |
+
"grad_norm": 0.5157521963119507,
|
| 1972 |
+
"learning_rate": 1.6842857142857143e-06,
|
| 1973 |
+
"loss": 0.0107,
|
| 1974 |
+
"step": 6825
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.792498980839788,
|
| 1978 |
+
"grad_norm": 1.2508779764175415,
|
| 1979 |
+
"learning_rate": 1.6485714285714289e-06,
|
| 1980 |
+
"loss": 0.015,
|
| 1981 |
+
"step": 6850
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 2.802690582959641,
|
| 1985 |
+
"grad_norm": 1.2929726839065552,
|
| 1986 |
+
"learning_rate": 1.612857142857143e-06,
|
| 1987 |
+
"loss": 0.0125,
|
| 1988 |
+
"step": 6875
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 2.8128821850794945,
|
| 1992 |
+
"grad_norm": 0.8893064856529236,
|
| 1993 |
+
"learning_rate": 1.5771428571428574e-06,
|
| 1994 |
+
"loss": 0.0122,
|
| 1995 |
+
"step": 6900
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 2.823073787199348,
|
| 1999 |
+
"grad_norm": 1.2606889009475708,
|
| 2000 |
+
"learning_rate": 1.5414285714285715e-06,
|
| 2001 |
+
"loss": 0.0137,
|
| 2002 |
+
"step": 6925
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 2.833265389319201,
|
| 2006 |
+
"grad_norm": 1.410904884338379,
|
| 2007 |
+
"learning_rate": 1.5057142857142858e-06,
|
| 2008 |
+
"loss": 0.0121,
|
| 2009 |
+
"step": 6950
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 2.843456991439054,
|
| 2013 |
+
"grad_norm": 0.792271077632904,
|
| 2014 |
+
"learning_rate": 1.4700000000000001e-06,
|
| 2015 |
+
"loss": 0.0135,
|
| 2016 |
+
"step": 6975
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 2.8536485935589075,
|
| 2020 |
+
"grad_norm": 1.0926941633224487,
|
| 2021 |
+
"learning_rate": 1.4342857142857144e-06,
|
| 2022 |
+
"loss": 0.0117,
|
| 2023 |
+
"step": 7000
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 2.8536485935589075,
|
| 2027 |
+
"eval_loss": 0.022013096138834953,
|
| 2028 |
+
"eval_runtime": 2391.195,
|
| 2029 |
+
"eval_samples_per_second": 4.264,
|
| 2030 |
+
"eval_steps_per_second": 0.133,
|
| 2031 |
+
"eval_wer": 2.5110128090281947,
|
| 2032 |
+
"step": 7000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 2.863840195678761,
|
| 2036 |
+
"grad_norm": 1.1174569129943848,
|
| 2037 |
+
"learning_rate": 1.3985714285714285e-06,
|
| 2038 |
+
"loss": 0.0123,
|
| 2039 |
+
"step": 7025
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 2.874031797798614,
|
| 2043 |
+
"grad_norm": 1.3422982692718506,
|
| 2044 |
+
"learning_rate": 1.362857142857143e-06,
|
| 2045 |
+
"loss": 0.0155,
|
| 2046 |
+
"step": 7050
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 2.884223399918467,
|
| 2050 |
+
"grad_norm": 1.1771271228790283,
|
| 2051 |
+
"learning_rate": 1.327142857142857e-06,
|
| 2052 |
+
"loss": 0.014,
|
| 2053 |
+
"step": 7075
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 2.8944150020383206,
|
| 2057 |
+
"grad_norm": 0.8138777017593384,
|
| 2058 |
+
"learning_rate": 1.2914285714285716e-06,
|
| 2059 |
+
"loss": 0.0159,
|
| 2060 |
+
"step": 7100
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 2.9046066041581735,
|
| 2064 |
+
"grad_norm": 0.964419960975647,
|
| 2065 |
+
"learning_rate": 1.2557142857142859e-06,
|
| 2066 |
+
"loss": 0.0119,
|
| 2067 |
+
"step": 7125
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 2.914798206278027,
|
| 2071 |
+
"grad_norm": 1.0425853729248047,
|
| 2072 |
+
"learning_rate": 1.2200000000000002e-06,
|
| 2073 |
+
"loss": 0.0144,
|
| 2074 |
+
"step": 7150
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 2.9249898083978803,
|
| 2078 |
+
"grad_norm": 1.2048903703689575,
|
| 2079 |
+
"learning_rate": 1.1842857142857143e-06,
|
| 2080 |
+
"loss": 0.0112,
|
| 2081 |
+
"step": 7175
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 2.9351814105177336,
|
| 2085 |
+
"grad_norm": 0.9228710532188416,
|
| 2086 |
+
"learning_rate": 1.1485714285714286e-06,
|
| 2087 |
+
"loss": 0.0121,
|
| 2088 |
+
"step": 7200
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 2.9453730126375866,
|
| 2092 |
+
"grad_norm": 0.35819801688194275,
|
| 2093 |
+
"learning_rate": 1.1128571428571429e-06,
|
| 2094 |
+
"loss": 0.0103,
|
| 2095 |
+
"step": 7225
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 2.95556461475744,
|
| 2099 |
+
"grad_norm": 0.8136561512947083,
|
| 2100 |
+
"learning_rate": 1.0771428571428574e-06,
|
| 2101 |
+
"loss": 0.0129,
|
| 2102 |
+
"step": 7250
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 2.965756216877293,
|
| 2106 |
+
"grad_norm": 1.0411508083343506,
|
| 2107 |
+
"learning_rate": 1.0414285714285717e-06,
|
| 2108 |
+
"loss": 0.0104,
|
| 2109 |
+
"step": 7275
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 2.975947818997146,
|
| 2113 |
+
"grad_norm": 0.7591568231582642,
|
| 2114 |
+
"learning_rate": 1.0057142857142857e-06,
|
| 2115 |
+
"loss": 0.0104,
|
| 2116 |
+
"step": 7300
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 2.9861394211169996,
|
| 2120 |
+
"grad_norm": 0.9218117594718933,
|
| 2121 |
+
"learning_rate": 9.7e-07,
|
| 2122 |
+
"loss": 0.0123,
|
| 2123 |
+
"step": 7325
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 2.996331023236853,
|
| 2127 |
+
"grad_norm": 1.3467174768447876,
|
| 2128 |
+
"learning_rate": 9.342857142857144e-07,
|
| 2129 |
+
"loss": 0.01,
|
| 2130 |
+
"step": 7350
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 3.006522625356706,
|
| 2134 |
+
"grad_norm": 0.14992065727710724,
|
| 2135 |
+
"learning_rate": 8.985714285714286e-07,
|
| 2136 |
+
"loss": 0.0068,
|
| 2137 |
+
"step": 7375
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 3.0167142274765593,
|
| 2141 |
+
"grad_norm": 0.44808146357536316,
|
| 2142 |
+
"learning_rate": 8.628571428571429e-07,
|
| 2143 |
+
"loss": 0.0044,
|
| 2144 |
+
"step": 7400
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.0269058295964126,
|
| 2148 |
+
"grad_norm": 0.4159485101699829,
|
| 2149 |
+
"learning_rate": 8.271428571428572e-07,
|
| 2150 |
+
"loss": 0.0045,
|
| 2151 |
+
"step": 7425
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.037097431716266,
|
| 2155 |
+
"grad_norm": 0.5730934143066406,
|
| 2156 |
+
"learning_rate": 7.914285714285715e-07,
|
| 2157 |
+
"loss": 0.0041,
|
| 2158 |
+
"step": 7450
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 3.047289033836119,
|
| 2162 |
+
"grad_norm": 0.2851293683052063,
|
| 2163 |
+
"learning_rate": 7.557142857142857e-07,
|
| 2164 |
+
"loss": 0.0048,
|
| 2165 |
+
"step": 7475
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 3.0574806359559723,
|
| 2169 |
+
"grad_norm": 0.3623310625553131,
|
| 2170 |
+
"learning_rate": 7.2e-07,
|
| 2171 |
+
"loss": 0.0054,
|
| 2172 |
+
"step": 7500
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 3.0676722380758257,
|
| 2176 |
+
"grad_norm": 0.1989881545305252,
|
| 2177 |
+
"learning_rate": 6.842857142857143e-07,
|
| 2178 |
+
"loss": 0.004,
|
| 2179 |
+
"step": 7525
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 3.0778638401956786,
|
| 2183 |
+
"grad_norm": 0.24523906409740448,
|
| 2184 |
+
"learning_rate": 6.485714285714287e-07,
|
| 2185 |
+
"loss": 0.0052,
|
| 2186 |
+
"step": 7550
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 3.088055442315532,
|
| 2190 |
+
"grad_norm": 0.09773947298526764,
|
| 2191 |
+
"learning_rate": 6.128571428571429e-07,
|
| 2192 |
+
"loss": 0.0037,
|
| 2193 |
+
"step": 7575
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 3.0982470444353853,
|
| 2197 |
+
"grad_norm": 0.8551767468452454,
|
| 2198 |
+
"learning_rate": 5.771428571428572e-07,
|
| 2199 |
+
"loss": 0.0041,
|
| 2200 |
+
"step": 7600
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 3.1084386465552383,
|
| 2204 |
+
"grad_norm": 0.32709449529647827,
|
| 2205 |
+
"learning_rate": 5.414285714285715e-07,
|
| 2206 |
+
"loss": 0.0046,
|
| 2207 |
+
"step": 7625
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 3.1186302486750916,
|
| 2211 |
+
"grad_norm": 0.22505348920822144,
|
| 2212 |
+
"learning_rate": 5.057142857142858e-07,
|
| 2213 |
+
"loss": 0.0039,
|
| 2214 |
+
"step": 7650
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 3.128821850794945,
|
| 2218 |
+
"grad_norm": 0.6920948028564453,
|
| 2219 |
+
"learning_rate": 4.7000000000000005e-07,
|
| 2220 |
+
"loss": 0.005,
|
| 2221 |
+
"step": 7675
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.1390134529147984,
|
| 2225 |
+
"grad_norm": 0.37104272842407227,
|
| 2226 |
+
"learning_rate": 4.342857142857143e-07,
|
| 2227 |
+
"loss": 0.0041,
|
| 2228 |
+
"step": 7700
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 3.1492050550346513,
|
| 2232 |
+
"grad_norm": 0.5752152800559998,
|
| 2233 |
+
"learning_rate": 3.985714285714286e-07,
|
| 2234 |
+
"loss": 0.0046,
|
| 2235 |
+
"step": 7725
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 3.1593966571545047,
|
| 2239 |
+
"grad_norm": 0.40613701939582825,
|
| 2240 |
+
"learning_rate": 3.6285714285714283e-07,
|
| 2241 |
+
"loss": 0.0041,
|
| 2242 |
+
"step": 7750
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 3.169588259274358,
|
| 2246 |
+
"grad_norm": 0.4888840317726135,
|
| 2247 |
+
"learning_rate": 3.271428571428572e-07,
|
| 2248 |
+
"loss": 0.0047,
|
| 2249 |
+
"step": 7775
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 3.179779861394211,
|
| 2253 |
+
"grad_norm": 0.22719787061214447,
|
| 2254 |
+
"learning_rate": 2.914285714285715e-07,
|
| 2255 |
+
"loss": 0.0039,
|
| 2256 |
+
"step": 7800
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.1899714635140644,
|
| 2260 |
+
"grad_norm": 0.3767295479774475,
|
| 2261 |
+
"learning_rate": 2.557142857142857e-07,
|
| 2262 |
+
"loss": 0.0047,
|
| 2263 |
+
"step": 7825
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 3.2001630656339177,
|
| 2267 |
+
"grad_norm": 0.6686317324638367,
|
| 2268 |
+
"learning_rate": 2.2e-07,
|
| 2269 |
+
"loss": 0.0041,
|
| 2270 |
+
"step": 7850
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 3.210354667753771,
|
| 2274 |
+
"grad_norm": 0.40455150604248047,
|
| 2275 |
+
"learning_rate": 1.842857142857143e-07,
|
| 2276 |
+
"loss": 0.0058,
|
| 2277 |
+
"step": 7875
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 3.220546269873624,
|
| 2281 |
+
"grad_norm": 0.2686769962310791,
|
| 2282 |
+
"learning_rate": 1.4857142857142857e-07,
|
| 2283 |
+
"loss": 0.0046,
|
| 2284 |
+
"step": 7900
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 3.2307378719934774,
|
| 2288 |
+
"grad_norm": 0.17891818284988403,
|
| 2289 |
+
"learning_rate": 1.1285714285714287e-07,
|
| 2290 |
+
"loss": 0.0038,
|
| 2291 |
+
"step": 7925
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 3.2409294741133308,
|
| 2295 |
+
"grad_norm": 0.32351842522621155,
|
| 2296 |
+
"learning_rate": 7.714285714285715e-08,
|
| 2297 |
+
"loss": 0.0036,
|
| 2298 |
+
"step": 7950
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.2511210762331837,
|
| 2302 |
+
"grad_norm": 0.2572202682495117,
|
| 2303 |
+
"learning_rate": 4.1428571428571426e-08,
|
| 2304 |
+
"loss": 0.0036,
|
| 2305 |
+
"step": 7975
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 3.261312678353037,
|
| 2309 |
+
"grad_norm": 0.1762164831161499,
|
| 2310 |
+
"learning_rate": 5.714285714285715e-09,
|
| 2311 |
+
"loss": 0.0044,
|
| 2312 |
+
"step": 8000
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 3.261312678353037,
|
| 2316 |
+
"eval_loss": 0.02037939429283142,
|
| 2317 |
+
"eval_runtime": 2392.6668,
|
| 2318 |
+
"eval_samples_per_second": 4.261,
|
| 2319 |
+
"eval_steps_per_second": 0.133,
|
| 2320 |
+
"eval_wer": 2.2782892974889872,
|
| 2321 |
+
"step": 8000
|
| 2322 |
+
}
|
| 2323 |
+
],
|
| 2324 |
+
"logging_steps": 25,
|
| 2325 |
+
"max_steps": 8000,
|
| 2326 |
+
"num_input_tokens_seen": 0,
|
| 2327 |
+
"num_train_epochs": 4,
|
| 2328 |
+
"save_steps": 1000,
|
| 2329 |
+
"stateful_callbacks": {
|
| 2330 |
+
"TrainerControl": {
|
| 2331 |
+
"args": {
|
| 2332 |
+
"should_epoch_stop": false,
|
| 2333 |
+
"should_evaluate": false,
|
| 2334 |
+
"should_log": false,
|
| 2335 |
+
"should_save": true,
|
| 2336 |
+
"should_training_stop": true
|
| 2337 |
+
},
|
| 2338 |
+
"attributes": {}
|
| 2339 |
+
}
|
| 2340 |
+
},
|
| 2341 |
+
"total_flos": 2.6124729999261696e+20,
|
| 2342 |
+
"train_batch_size": 32,
|
| 2343 |
+
"trial_name": null,
|
| 2344 |
+
"trial_params": null
|
| 2345 |
+
}
|