Commit
•
2729feb
1
Parent(s):
7509a05
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +2022 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2022 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8526459553649132,
|
| 3 |
+
"best_model_checkpoint": "stability_weights_dora_norm_swap_v2/checkpoint-1820",
|
| 4 |
+
"epoch": 10.996978851963746,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1820,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06042296072507553,
|
| 13 |
+
"grad_norm": 42.27948760986328,
|
| 14 |
+
"learning_rate": 2.5e-06,
|
| 15 |
+
"loss": 3.9615,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.12084592145015106,
|
| 20 |
+
"grad_norm": 45.654972076416016,
|
| 21 |
+
"learning_rate": 5e-06,
|
| 22 |
+
"loss": 2.8167,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.18126888217522658,
|
| 27 |
+
"grad_norm": 33.64224624633789,
|
| 28 |
+
"learning_rate": 7.5e-06,
|
| 29 |
+
"loss": 3.9832,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.24169184290030213,
|
| 34 |
+
"grad_norm": 38.766605377197266,
|
| 35 |
+
"learning_rate": 1e-05,
|
| 36 |
+
"loss": 3.483,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.3021148036253776,
|
| 41 |
+
"grad_norm": 36.45051574707031,
|
| 42 |
+
"learning_rate": 1.25e-05,
|
| 43 |
+
"loss": 2.6488,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.36253776435045315,
|
| 48 |
+
"grad_norm": 31.120370864868164,
|
| 49 |
+
"learning_rate": 1.5e-05,
|
| 50 |
+
"loss": 2.7824,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.4229607250755287,
|
| 55 |
+
"grad_norm": 40.57133865356445,
|
| 56 |
+
"learning_rate": 1.75e-05,
|
| 57 |
+
"loss": 3.4959,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.48338368580060426,
|
| 62 |
+
"grad_norm": 39.07503890991211,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 2.7043,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.5438066465256798,
|
| 69 |
+
"grad_norm": 27.892555236816406,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 2.3916,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.6042296072507553,
|
| 76 |
+
"grad_norm": 35.324249267578125,
|
| 77 |
+
"learning_rate": 2.5e-05,
|
| 78 |
+
"loss": 3.5084,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.6646525679758308,
|
| 83 |
+
"grad_norm": 34.17033386230469,
|
| 84 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 85 |
+
"loss": 2.15,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.7250755287009063,
|
| 90 |
+
"grad_norm": 28.871646881103516,
|
| 91 |
+
"learning_rate": 3e-05,
|
| 92 |
+
"loss": 2.4973,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7854984894259819,
|
| 97 |
+
"grad_norm": 34.328651428222656,
|
| 98 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 99 |
+
"loss": 2.2425,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.8459214501510574,
|
| 104 |
+
"grad_norm": 27.76066017150879,
|
| 105 |
+
"learning_rate": 3.5e-05,
|
| 106 |
+
"loss": 2.4867,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.9063444108761329,
|
| 111 |
+
"grad_norm": 46.40857696533203,
|
| 112 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 113 |
+
"loss": 2.3429,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.9667673716012085,
|
| 118 |
+
"grad_norm": 24.273012161254883,
|
| 119 |
+
"learning_rate": 4e-05,
|
| 120 |
+
"loss": 1.8543,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.9969788519637462,
|
| 125 |
+
"eval_validation_loss": 2.0505075454711914,
|
| 126 |
+
"eval_validation_mae": 1.01314377784729,
|
| 127 |
+
"eval_validation_mse": 2.0505075454711914,
|
| 128 |
+
"eval_validation_pearson": 0.46795711140864427,
|
| 129 |
+
"eval_validation_rmse": 1.4319593906402588,
|
| 130 |
+
"eval_validation_runtime": 129.1022,
|
| 131 |
+
"eval_validation_samples_per_second": 2.649,
|
| 132 |
+
"eval_validation_spearman": 0.5000603914960909,
|
| 133 |
+
"eval_validation_steps_per_second": 2.649,
|
| 134 |
+
"step": 165
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.9969788519637462,
|
| 138 |
+
"eval_test_loss": 2.0498456954956055,
|
| 139 |
+
"eval_test_mae": 1.0129708051681519,
|
| 140 |
+
"eval_test_mse": 2.0498456954956055,
|
| 141 |
+
"eval_test_pearson": 0.4678271674806638,
|
| 142 |
+
"eval_test_rmse": 1.4317282438278198,
|
| 143 |
+
"eval_test_runtime": 128.8868,
|
| 144 |
+
"eval_test_samples_per_second": 2.653,
|
| 145 |
+
"eval_test_spearman": 0.49953701661422967,
|
| 146 |
+
"eval_test_steps_per_second": 2.653,
|
| 147 |
+
"step": 165
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.9969788519637462,
|
| 151 |
+
"eval_myoglobin_loss": 1.1084952354431152,
|
| 152 |
+
"eval_myoglobin_mae": 0.7779242396354675,
|
| 153 |
+
"eval_myoglobin_mse": 1.1084952354431152,
|
| 154 |
+
"eval_myoglobin_pearson": 0.3155295627311716,
|
| 155 |
+
"eval_myoglobin_rmse": 1.0528509616851807,
|
| 156 |
+
"eval_myoglobin_runtime": 50.3681,
|
| 157 |
+
"eval_myoglobin_samples_per_second": 2.66,
|
| 158 |
+
"eval_myoglobin_spearman": 0.3150316299942073,
|
| 159 |
+
"eval_myoglobin_steps_per_second": 2.66,
|
| 160 |
+
"step": 165
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.9969788519637462,
|
| 164 |
+
"eval_myoglobin_r_loss": 1.1083894968032837,
|
| 165 |
+
"eval_myoglobin_r_mae": 0.778092622756958,
|
| 166 |
+
"eval_myoglobin_r_mse": 1.1083894968032837,
|
| 167 |
+
"eval_myoglobin_r_pearson": 0.315758311809077,
|
| 168 |
+
"eval_myoglobin_r_rmse": 1.0528007745742798,
|
| 169 |
+
"eval_myoglobin_r_runtime": 50.3077,
|
| 170 |
+
"eval_myoglobin_r_samples_per_second": 2.664,
|
| 171 |
+
"eval_myoglobin_r_spearman": 0.3167998663343996,
|
| 172 |
+
"eval_myoglobin_r_steps_per_second": 2.664,
|
| 173 |
+
"step": 165
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.9969788519637462,
|
| 177 |
+
"eval_p53_loss": 4.773194789886475,
|
| 178 |
+
"eval_p53_mae": 1.6836440563201904,
|
| 179 |
+
"eval_p53_mse": 4.773194789886475,
|
| 180 |
+
"eval_p53_pearson": 0.260329486693081,
|
| 181 |
+
"eval_p53_rmse": 2.1847641468048096,
|
| 182 |
+
"eval_p53_runtime": 17.5157,
|
| 183 |
+
"eval_p53_samples_per_second": 2.398,
|
| 184 |
+
"eval_p53_spearman": 0.1950488231413393,
|
| 185 |
+
"eval_p53_steps_per_second": 2.398,
|
| 186 |
+
"step": 165
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.027190332326284,
|
| 190 |
+
"grad_norm": 25.500028610229492,
|
| 191 |
+
"learning_rate": 4.25e-05,
|
| 192 |
+
"loss": 2.1163,
|
| 193 |
+
"step": 170
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.0876132930513596,
|
| 197 |
+
"grad_norm": 31.47431755065918,
|
| 198 |
+
"learning_rate": 4.5e-05,
|
| 199 |
+
"loss": 1.7539,
|
| 200 |
+
"step": 180
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.148036253776435,
|
| 204 |
+
"grad_norm": 22.619182586669922,
|
| 205 |
+
"learning_rate": 4.75e-05,
|
| 206 |
+
"loss": 1.4633,
|
| 207 |
+
"step": 190
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.2084592145015105,
|
| 211 |
+
"grad_norm": 38.80492401123047,
|
| 212 |
+
"learning_rate": 5e-05,
|
| 213 |
+
"loss": 1.4391,
|
| 214 |
+
"step": 200
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.2688821752265862,
|
| 218 |
+
"grad_norm": 22.946945190429688,
|
| 219 |
+
"learning_rate": 5.25e-05,
|
| 220 |
+
"loss": 1.3334,
|
| 221 |
+
"step": 210
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.3293051359516617,
|
| 225 |
+
"grad_norm": 14.94211483001709,
|
| 226 |
+
"learning_rate": 5.500000000000001e-05,
|
| 227 |
+
"loss": 1.1311,
|
| 228 |
+
"step": 220
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.3897280966767371,
|
| 232 |
+
"grad_norm": 27.56308364868164,
|
| 233 |
+
"learning_rate": 5.7499999999999995e-05,
|
| 234 |
+
"loss": 1.6216,
|
| 235 |
+
"step": 230
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.4501510574018126,
|
| 239 |
+
"grad_norm": 19.250564575195312,
|
| 240 |
+
"learning_rate": 6e-05,
|
| 241 |
+
"loss": 1.273,
|
| 242 |
+
"step": 240
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.510574018126888,
|
| 246 |
+
"grad_norm": 19.437816619873047,
|
| 247 |
+
"learning_rate": 6.25e-05,
|
| 248 |
+
"loss": 0.9644,
|
| 249 |
+
"step": 250
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.5709969788519638,
|
| 253 |
+
"grad_norm": 31.529216766357422,
|
| 254 |
+
"learning_rate": 6.500000000000001e-05,
|
| 255 |
+
"loss": 1.7949,
|
| 256 |
+
"step": 260
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.6314199395770392,
|
| 260 |
+
"grad_norm": 32.55619430541992,
|
| 261 |
+
"learning_rate": 6.750000000000001e-05,
|
| 262 |
+
"loss": 1.241,
|
| 263 |
+
"step": 270
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.691842900302115,
|
| 267 |
+
"grad_norm": 26.349998474121094,
|
| 268 |
+
"learning_rate": 7e-05,
|
| 269 |
+
"loss": 1.3735,
|
| 270 |
+
"step": 280
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.7522658610271904,
|
| 274 |
+
"grad_norm": 29.46245002746582,
|
| 275 |
+
"learning_rate": 7.25e-05,
|
| 276 |
+
"loss": 1.4623,
|
| 277 |
+
"step": 290
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 1.8126888217522659,
|
| 281 |
+
"grad_norm": 14.214956283569336,
|
| 282 |
+
"learning_rate": 7.500000000000001e-05,
|
| 283 |
+
"loss": 0.8406,
|
| 284 |
+
"step": 300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.8731117824773413,
|
| 288 |
+
"grad_norm": 16.597463607788086,
|
| 289 |
+
"learning_rate": 7.75e-05,
|
| 290 |
+
"loss": 0.9432,
|
| 291 |
+
"step": 310
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 1.9335347432024168,
|
| 295 |
+
"grad_norm": 17.1890869140625,
|
| 296 |
+
"learning_rate": 8e-05,
|
| 297 |
+
"loss": 1.195,
|
| 298 |
+
"step": 320
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 1.9939577039274925,
|
| 302 |
+
"grad_norm": 17.546321868896484,
|
| 303 |
+
"learning_rate": 8.25e-05,
|
| 304 |
+
"loss": 1.0398,
|
| 305 |
+
"step": 330
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 2.0,
|
| 309 |
+
"eval_validation_loss": 0.9950127005577087,
|
| 310 |
+
"eval_validation_mae": 0.6999762654304504,
|
| 311 |
+
"eval_validation_mse": 0.9950127005577087,
|
| 312 |
+
"eval_validation_pearson": 0.7727618943702055,
|
| 313 |
+
"eval_validation_rmse": 0.9975032210350037,
|
| 314 |
+
"eval_validation_runtime": 129.1954,
|
| 315 |
+
"eval_validation_samples_per_second": 2.647,
|
| 316 |
+
"eval_validation_spearman": 0.7928617789227591,
|
| 317 |
+
"eval_validation_steps_per_second": 2.647,
|
| 318 |
+
"step": 331
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 2.0,
|
| 322 |
+
"eval_test_loss": 0.9929149150848389,
|
| 323 |
+
"eval_test_mae": 0.7004790306091309,
|
| 324 |
+
"eval_test_mse": 0.9929149150848389,
|
| 325 |
+
"eval_test_pearson": 0.7726932487807971,
|
| 326 |
+
"eval_test_rmse": 0.9964511394500732,
|
| 327 |
+
"eval_test_runtime": 129.0193,
|
| 328 |
+
"eval_test_samples_per_second": 2.651,
|
| 329 |
+
"eval_test_spearman": 0.7933636966348652,
|
| 330 |
+
"eval_test_steps_per_second": 2.651,
|
| 331 |
+
"step": 331
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 2.0,
|
| 335 |
+
"eval_myoglobin_loss": 0.934596061706543,
|
| 336 |
+
"eval_myoglobin_mae": 0.7355605959892273,
|
| 337 |
+
"eval_myoglobin_mse": 0.934596061706543,
|
| 338 |
+
"eval_myoglobin_pearson": 0.5740624022673728,
|
| 339 |
+
"eval_myoglobin_rmse": 0.9667450785636902,
|
| 340 |
+
"eval_myoglobin_runtime": 50.363,
|
| 341 |
+
"eval_myoglobin_samples_per_second": 2.661,
|
| 342 |
+
"eval_myoglobin_spearman": 0.5940538377065167,
|
| 343 |
+
"eval_myoglobin_steps_per_second": 2.661,
|
| 344 |
+
"step": 331
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.0,
|
| 348 |
+
"eval_myoglobin_r_loss": 0.9417540431022644,
|
| 349 |
+
"eval_myoglobin_r_mae": 0.7385311722755432,
|
| 350 |
+
"eval_myoglobin_r_mse": 0.9417540431022644,
|
| 351 |
+
"eval_myoglobin_r_pearson": 0.5734044991540941,
|
| 352 |
+
"eval_myoglobin_r_rmse": 0.970440149307251,
|
| 353 |
+
"eval_myoglobin_r_runtime": 50.3941,
|
| 354 |
+
"eval_myoglobin_r_samples_per_second": 2.659,
|
| 355 |
+
"eval_myoglobin_r_spearman": 0.5919040214508811,
|
| 356 |
+
"eval_myoglobin_r_steps_per_second": 2.659,
|
| 357 |
+
"step": 331
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 2.0,
|
| 361 |
+
"eval_p53_loss": 4.290196895599365,
|
| 362 |
+
"eval_p53_mae": 1.622589349746704,
|
| 363 |
+
"eval_p53_mse": 4.290196895599365,
|
| 364 |
+
"eval_p53_pearson": 0.29006131482310593,
|
| 365 |
+
"eval_p53_rmse": 2.0712790489196777,
|
| 366 |
+
"eval_p53_runtime": 17.523,
|
| 367 |
+
"eval_p53_samples_per_second": 2.397,
|
| 368 |
+
"eval_p53_spearman": 0.26692597566579634,
|
| 369 |
+
"eval_p53_steps_per_second": 2.397,
|
| 370 |
+
"step": 331
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.054380664652568,
|
| 374 |
+
"grad_norm": 13.338666915893555,
|
| 375 |
+
"learning_rate": 8.5e-05,
|
| 376 |
+
"loss": 0.6917,
|
| 377 |
+
"step": 340
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.1148036253776437,
|
| 381 |
+
"grad_norm": 21.178325653076172,
|
| 382 |
+
"learning_rate": 8.75e-05,
|
| 383 |
+
"loss": 0.8003,
|
| 384 |
+
"step": 350
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.175226586102719,
|
| 388 |
+
"grad_norm": 14.95538330078125,
|
| 389 |
+
"learning_rate": 9e-05,
|
| 390 |
+
"loss": 0.6155,
|
| 391 |
+
"step": 360
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.2356495468277946,
|
| 395 |
+
"grad_norm": 10.555205345153809,
|
| 396 |
+
"learning_rate": 9.250000000000001e-05,
|
| 397 |
+
"loss": 0.6111,
|
| 398 |
+
"step": 370
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.29607250755287,
|
| 402 |
+
"grad_norm": 11.569594383239746,
|
| 403 |
+
"learning_rate": 9.5e-05,
|
| 404 |
+
"loss": 0.8681,
|
| 405 |
+
"step": 380
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.3564954682779455,
|
| 409 |
+
"grad_norm": 18.517148971557617,
|
| 410 |
+
"learning_rate": 9.75e-05,
|
| 411 |
+
"loss": 0.6754,
|
| 412 |
+
"step": 390
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 2.416918429003021,
|
| 416 |
+
"grad_norm": 21.002437591552734,
|
| 417 |
+
"learning_rate": 0.0001,
|
| 418 |
+
"loss": 0.5359,
|
| 419 |
+
"step": 400
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 2.477341389728097,
|
| 423 |
+
"grad_norm": 17.249919891357422,
|
| 424 |
+
"learning_rate": 9.999706613915566e-05,
|
| 425 |
+
"loss": 0.8332,
|
| 426 |
+
"step": 410
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 2.5377643504531724,
|
| 430 |
+
"grad_norm": 10.25106143951416,
|
| 431 |
+
"learning_rate": 9.998826490092421e-05,
|
| 432 |
+
"loss": 0.496,
|
| 433 |
+
"step": 420
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 2.598187311178248,
|
| 437 |
+
"grad_norm": 18.843320846557617,
|
| 438 |
+
"learning_rate": 9.997359731816998e-05,
|
| 439 |
+
"loss": 0.7244,
|
| 440 |
+
"step": 430
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.6586102719033233,
|
| 444 |
+
"grad_norm": 11.991676330566406,
|
| 445 |
+
"learning_rate": 9.995306511219885e-05,
|
| 446 |
+
"loss": 0.5112,
|
| 447 |
+
"step": 440
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 2.719033232628399,
|
| 451 |
+
"grad_norm": 21.79170036315918,
|
| 452 |
+
"learning_rate": 9.992667069255619e-05,
|
| 453 |
+
"loss": 0.5307,
|
| 454 |
+
"step": 450
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 2.7794561933534743,
|
| 458 |
+
"grad_norm": 9.106186866760254,
|
| 459 |
+
"learning_rate": 9.989441715674422e-05,
|
| 460 |
+
"loss": 0.4371,
|
| 461 |
+
"step": 460
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 2.8398791540785497,
|
| 465 |
+
"grad_norm": 10.968169212341309,
|
| 466 |
+
"learning_rate": 9.985630828985835e-05,
|
| 467 |
+
"loss": 0.5922,
|
| 468 |
+
"step": 470
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 2.900302114803625,
|
| 472 |
+
"grad_norm": 14.699440002441406,
|
| 473 |
+
"learning_rate": 9.981234856414307e-05,
|
| 474 |
+
"loss": 0.5858,
|
| 475 |
+
"step": 480
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 2.9607250755287007,
|
| 479 |
+
"grad_norm": 13.304734230041504,
|
| 480 |
+
"learning_rate": 9.97625431384671e-05,
|
| 481 |
+
"loss": 0.6854,
|
| 482 |
+
"step": 490
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.996978851963746,
|
| 486 |
+
"eval_validation_loss": 0.8451957702636719,
|
| 487 |
+
"eval_validation_mae": 0.6374915242195129,
|
| 488 |
+
"eval_validation_mse": 0.8451957702636719,
|
| 489 |
+
"eval_validation_pearson": 0.8118986310850005,
|
| 490 |
+
"eval_validation_rmse": 0.9193453192710876,
|
| 491 |
+
"eval_validation_runtime": 128.9657,
|
| 492 |
+
"eval_validation_samples_per_second": 2.652,
|
| 493 |
+
"eval_validation_spearman": 0.8191585157836645,
|
| 494 |
+
"eval_validation_steps_per_second": 2.652,
|
| 495 |
+
"step": 496
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 2.996978851963746,
|
| 499 |
+
"eval_test_loss": 0.8352708220481873,
|
| 500 |
+
"eval_test_mae": 0.6319109201431274,
|
| 501 |
+
"eval_test_mse": 0.8352708220481873,
|
| 502 |
+
"eval_test_pearson": 0.8117305964358366,
|
| 503 |
+
"eval_test_rmse": 0.9139315485954285,
|
| 504 |
+
"eval_test_runtime": 129.0007,
|
| 505 |
+
"eval_test_samples_per_second": 2.651,
|
| 506 |
+
"eval_test_spearman": 0.8192730040390717,
|
| 507 |
+
"eval_test_steps_per_second": 2.651,
|
| 508 |
+
"step": 496
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.996978851963746,
|
| 512 |
+
"eval_myoglobin_loss": 0.8053471446037292,
|
| 513 |
+
"eval_myoglobin_mae": 0.6675698757171631,
|
| 514 |
+
"eval_myoglobin_mse": 0.8053471446037292,
|
| 515 |
+
"eval_myoglobin_pearson": 0.6029393659750895,
|
| 516 |
+
"eval_myoglobin_rmse": 0.8974113464355469,
|
| 517 |
+
"eval_myoglobin_runtime": 50.3201,
|
| 518 |
+
"eval_myoglobin_samples_per_second": 2.663,
|
| 519 |
+
"eval_myoglobin_spearman": 0.6248470874841663,
|
| 520 |
+
"eval_myoglobin_steps_per_second": 2.663,
|
| 521 |
+
"step": 496
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 2.996978851963746,
|
| 525 |
+
"eval_myoglobin_r_loss": 0.8195508122444153,
|
| 526 |
+
"eval_myoglobin_r_mae": 0.6755225658416748,
|
| 527 |
+
"eval_myoglobin_r_mse": 0.8195508122444153,
|
| 528 |
+
"eval_myoglobin_r_pearson": 0.6011248022884956,
|
| 529 |
+
"eval_myoglobin_r_rmse": 0.9052904844284058,
|
| 530 |
+
"eval_myoglobin_r_runtime": 50.3351,
|
| 531 |
+
"eval_myoglobin_r_samples_per_second": 2.662,
|
| 532 |
+
"eval_myoglobin_r_spearman": 0.6238021071928538,
|
| 533 |
+
"eval_myoglobin_r_steps_per_second": 2.662,
|
| 534 |
+
"step": 496
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.996978851963746,
|
| 538 |
+
"eval_p53_loss": 4.1966938972473145,
|
| 539 |
+
"eval_p53_mae": 1.6126927137374878,
|
| 540 |
+
"eval_p53_mse": 4.1966938972473145,
|
| 541 |
+
"eval_p53_pearson": 0.3033285839483142,
|
| 542 |
+
"eval_p53_rmse": 2.0485832691192627,
|
| 543 |
+
"eval_p53_runtime": 17.5229,
|
| 544 |
+
"eval_p53_samples_per_second": 2.397,
|
| 545 |
+
"eval_p53_spearman": 0.2764069529435432,
|
| 546 |
+
"eval_p53_steps_per_second": 2.397,
|
| 547 |
+
"step": 496
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.0211480362537766,
|
| 551 |
+
"grad_norm": 7.5384063720703125,
|
| 552 |
+
"learning_rate": 9.970689785771798e-05,
|
| 553 |
+
"loss": 0.4402,
|
| 554 |
+
"step": 500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 3.081570996978852,
|
| 558 |
+
"grad_norm": 13.131170272827148,
|
| 559 |
+
"learning_rate": 9.964541925211612e-05,
|
| 560 |
+
"loss": 0.3957,
|
| 561 |
+
"step": 510
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 3.1419939577039275,
|
| 565 |
+
"grad_norm": 5.908936977386475,
|
| 566 |
+
"learning_rate": 9.957811453644847e-05,
|
| 567 |
+
"loss": 0.2294,
|
| 568 |
+
"step": 520
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 3.202416918429003,
|
| 572 |
+
"grad_norm": 7.813998699188232,
|
| 573 |
+
"learning_rate": 9.950499160922183e-05,
|
| 574 |
+
"loss": 0.2424,
|
| 575 |
+
"step": 530
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 3.2628398791540785,
|
| 579 |
+
"grad_norm": 9.31921100616455,
|
| 580 |
+
"learning_rate": 9.942605905173592e-05,
|
| 581 |
+
"loss": 0.3864,
|
| 582 |
+
"step": 540
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 3.323262839879154,
|
| 586 |
+
"grad_norm": 6.702413558959961,
|
| 587 |
+
"learning_rate": 9.934132612707632e-05,
|
| 588 |
+
"loss": 0.3761,
|
| 589 |
+
"step": 550
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 3.38368580060423,
|
| 593 |
+
"grad_norm": 9.235360145568848,
|
| 594 |
+
"learning_rate": 9.925080277902743e-05,
|
| 595 |
+
"loss": 0.2577,
|
| 596 |
+
"step": 560
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.4441087613293053,
|
| 600 |
+
"grad_norm": 10.142770767211914,
|
| 601 |
+
"learning_rate": 9.91544996309055e-05,
|
| 602 |
+
"loss": 0.322,
|
| 603 |
+
"step": 570
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 3.504531722054381,
|
| 607 |
+
"grad_norm": 12.021828651428223,
|
| 608 |
+
"learning_rate": 9.905242798431196e-05,
|
| 609 |
+
"loss": 0.2585,
|
| 610 |
+
"step": 580
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.5649546827794563,
|
| 614 |
+
"grad_norm": 7.120593070983887,
|
| 615 |
+
"learning_rate": 9.894459981780711e-05,
|
| 616 |
+
"loss": 0.3731,
|
| 617 |
+
"step": 590
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 3.6253776435045317,
|
| 621 |
+
"grad_norm": 9.525044441223145,
|
| 622 |
+
"learning_rate": 9.883102778550434e-05,
|
| 623 |
+
"loss": 0.3116,
|
| 624 |
+
"step": 600
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 3.685800604229607,
|
| 628 |
+
"grad_norm": 13.72534465789795,
|
| 629 |
+
"learning_rate": 9.871172521558523e-05,
|
| 630 |
+
"loss": 0.3361,
|
| 631 |
+
"step": 610
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 3.7462235649546827,
|
| 635 |
+
"grad_norm": 12.432317733764648,
|
| 636 |
+
"learning_rate": 9.858670610873528e-05,
|
| 637 |
+
"loss": 0.2497,
|
| 638 |
+
"step": 620
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.806646525679758,
|
| 642 |
+
"grad_norm": 7.066479206085205,
|
| 643 |
+
"learning_rate": 9.845598513650103e-05,
|
| 644 |
+
"loss": 0.3698,
|
| 645 |
+
"step": 630
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 3.8670694864048336,
|
| 649 |
+
"grad_norm": 9.628741264343262,
|
| 650 |
+
"learning_rate": 9.831957763956813e-05,
|
| 651 |
+
"loss": 0.3154,
|
| 652 |
+
"step": 640
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 3.9274924471299095,
|
| 656 |
+
"grad_norm": 11.99616813659668,
|
| 657 |
+
"learning_rate": 9.817749962596115e-05,
|
| 658 |
+
"loss": 0.3181,
|
| 659 |
+
"step": 650
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 3.987915407854985,
|
| 663 |
+
"grad_norm": 9.52004623413086,
|
| 664 |
+
"learning_rate": 9.802976776916494e-05,
|
| 665 |
+
"loss": 0.2454,
|
| 666 |
+
"step": 660
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 4.0,
|
| 670 |
+
"eval_validation_loss": 0.7796794772148132,
|
| 671 |
+
"eval_validation_mae": 0.6025121808052063,
|
| 672 |
+
"eval_validation_mse": 0.7796794772148132,
|
| 673 |
+
"eval_validation_pearson": 0.8275993492481604,
|
| 674 |
+
"eval_validation_rmse": 0.8829945921897888,
|
| 675 |
+
"eval_validation_runtime": 128.9701,
|
| 676 |
+
"eval_validation_samples_per_second": 2.652,
|
| 677 |
+
"eval_validation_spearman": 0.8297559568405268,
|
| 678 |
+
"eval_validation_steps_per_second": 2.652,
|
| 679 |
+
"step": 662
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 4.0,
|
| 683 |
+
"eval_test_loss": 0.7812950015068054,
|
| 684 |
+
"eval_test_mae": 0.6039361953735352,
|
| 685 |
+
"eval_test_mse": 0.7812950015068054,
|
| 686 |
+
"eval_test_pearson": 0.8269227759203943,
|
| 687 |
+
"eval_test_rmse": 0.8839089274406433,
|
| 688 |
+
"eval_test_runtime": 128.9749,
|
| 689 |
+
"eval_test_samples_per_second": 2.652,
|
| 690 |
+
"eval_test_spearman": 0.8285234450057767,
|
| 691 |
+
"eval_test_steps_per_second": 2.652,
|
| 692 |
+
"step": 662
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 4.0,
|
| 696 |
+
"eval_myoglobin_loss": 0.8277855515480042,
|
| 697 |
+
"eval_myoglobin_mae": 0.670007050037384,
|
| 698 |
+
"eval_myoglobin_mse": 0.8277855515480042,
|
| 699 |
+
"eval_myoglobin_pearson": 0.6006203565841588,
|
| 700 |
+
"eval_myoglobin_rmse": 0.9098272323608398,
|
| 701 |
+
"eval_myoglobin_runtime": 50.3879,
|
| 702 |
+
"eval_myoglobin_samples_per_second": 2.659,
|
| 703 |
+
"eval_myoglobin_spearman": 0.6452977757628354,
|
| 704 |
+
"eval_myoglobin_steps_per_second": 2.659,
|
| 705 |
+
"step": 662
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 4.0,
|
| 709 |
+
"eval_myoglobin_r_loss": 0.8345361948013306,
|
| 710 |
+
"eval_myoglobin_r_mae": 0.6731795072555542,
|
| 711 |
+
"eval_myoglobin_r_mse": 0.8345361948013306,
|
| 712 |
+
"eval_myoglobin_r_pearson": 0.5986034577274495,
|
| 713 |
+
"eval_myoglobin_r_rmse": 0.9135295152664185,
|
| 714 |
+
"eval_myoglobin_r_runtime": 50.4043,
|
| 715 |
+
"eval_myoglobin_r_samples_per_second": 2.659,
|
| 716 |
+
"eval_myoglobin_r_spearman": 0.6437390342781807,
|
| 717 |
+
"eval_myoglobin_r_steps_per_second": 2.659,
|
| 718 |
+
"step": 662
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 4.0,
|
| 722 |
+
"eval_p53_loss": 3.8773045539855957,
|
| 723 |
+
"eval_p53_mae": 1.5222440958023071,
|
| 724 |
+
"eval_p53_mse": 3.8773045539855957,
|
| 725 |
+
"eval_p53_pearson": 0.3431499426951632,
|
| 726 |
+
"eval_p53_rmse": 1.9690872430801392,
|
| 727 |
+
"eval_p53_runtime": 17.5208,
|
| 728 |
+
"eval_p53_samples_per_second": 2.397,
|
| 729 |
+
"eval_p53_spearman": 0.3029050689249383,
|
| 730 |
+
"eval_p53_steps_per_second": 2.397,
|
| 731 |
+
"step": 662
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 4.04833836858006,
|
| 735 |
+
"grad_norm": 7.301741123199463,
|
| 736 |
+
"learning_rate": 9.787639940616788e-05,
|
| 737 |
+
"loss": 0.1719,
|
| 738 |
+
"step": 670
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 4.108761329305136,
|
| 742 |
+
"grad_norm": 7.330774784088135,
|
| 743 |
+
"learning_rate": 9.771741253542741e-05,
|
| 744 |
+
"loss": 0.2217,
|
| 745 |
+
"step": 680
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 4.169184290030212,
|
| 749 |
+
"grad_norm": 4.552008152008057,
|
| 750 |
+
"learning_rate": 9.755282581475769e-05,
|
| 751 |
+
"loss": 0.2123,
|
| 752 |
+
"step": 690
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.229607250755287,
|
| 756 |
+
"grad_norm": 4.030465126037598,
|
| 757 |
+
"learning_rate": 9.738265855914013e-05,
|
| 758 |
+
"loss": 0.1504,
|
| 759 |
+
"step": 700
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 4.290030211480363,
|
| 763 |
+
"grad_norm": 13.193011283874512,
|
| 764 |
+
"learning_rate": 9.720693073845667e-05,
|
| 765 |
+
"loss": 0.1922,
|
| 766 |
+
"step": 710
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 4.350453172205438,
|
| 770 |
+
"grad_norm": 7.253832817077637,
|
| 771 |
+
"learning_rate": 9.70256629751462e-05,
|
| 772 |
+
"loss": 0.1163,
|
| 773 |
+
"step": 720
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 4.410876132930514,
|
| 777 |
+
"grad_norm": 4.736353874206543,
|
| 778 |
+
"learning_rate": 9.683887654178445e-05,
|
| 779 |
+
"loss": 0.1391,
|
| 780 |
+
"step": 730
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 4.471299093655589,
|
| 784 |
+
"grad_norm": 8.901742935180664,
|
| 785 |
+
"learning_rate": 9.664659335858755e-05,
|
| 786 |
+
"loss": 0.1764,
|
| 787 |
+
"step": 740
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 4.531722054380665,
|
| 791 |
+
"grad_norm": 6.86702299118042,
|
| 792 |
+
"learning_rate": 9.644883599083958e-05,
|
| 793 |
+
"loss": 0.2754,
|
| 794 |
+
"step": 750
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.59214501510574,
|
| 798 |
+
"grad_norm": 7.574690818786621,
|
| 799 |
+
"learning_rate": 9.624562764624445e-05,
|
| 800 |
+
"loss": 0.1909,
|
| 801 |
+
"step": 760
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 4.652567975830816,
|
| 805 |
+
"grad_norm": 7.513609409332275,
|
| 806 |
+
"learning_rate": 9.603699217220239e-05,
|
| 807 |
+
"loss": 0.1752,
|
| 808 |
+
"step": 770
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 4.712990936555891,
|
| 812 |
+
"grad_norm": 6.482759475708008,
|
| 813 |
+
"learning_rate": 9.582295405301131e-05,
|
| 814 |
+
"loss": 0.1733,
|
| 815 |
+
"step": 780
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 4.7734138972809665,
|
| 819 |
+
"grad_norm": 12.048788070678711,
|
| 820 |
+
"learning_rate": 9.56035384069935e-05,
|
| 821 |
+
"loss": 0.1986,
|
| 822 |
+
"step": 790
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 4.833836858006042,
|
| 826 |
+
"grad_norm": 10.056185722351074,
|
| 827 |
+
"learning_rate": 9.537877098354786e-05,
|
| 828 |
+
"loss": 0.1798,
|
| 829 |
+
"step": 800
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 4.8942598187311175,
|
| 833 |
+
"grad_norm": 11.082551956176758,
|
| 834 |
+
"learning_rate": 9.514867816012809e-05,
|
| 835 |
+
"loss": 0.2064,
|
| 836 |
+
"step": 810
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.954682779456194,
|
| 840 |
+
"grad_norm": 11.846305847167969,
|
| 841 |
+
"learning_rate": 9.491328693914722e-05,
|
| 842 |
+
"loss": 0.1868,
|
| 843 |
+
"step": 820
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 4.996978851963746,
|
| 847 |
+
"eval_validation_loss": 0.7762272953987122,
|
| 848 |
+
"eval_validation_mae": 0.5832256078720093,
|
| 849 |
+
"eval_validation_mse": 0.7762272953987122,
|
| 850 |
+
"eval_validation_pearson": 0.8296356628068153,
|
| 851 |
+
"eval_validation_rmse": 0.8810376524925232,
|
| 852 |
+
"eval_validation_runtime": 129.148,
|
| 853 |
+
"eval_validation_samples_per_second": 2.648,
|
| 854 |
+
"eval_validation_spearman": 0.8313329837924658,
|
| 855 |
+
"eval_validation_steps_per_second": 2.648,
|
| 856 |
+
"step": 827
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 4.996978851963746,
|
| 860 |
+
"eval_test_loss": 0.7746825814247131,
|
| 861 |
+
"eval_test_mae": 0.5824557542800903,
|
| 862 |
+
"eval_test_mse": 0.7746825814247131,
|
| 863 |
+
"eval_test_pearson": 0.8289941215708723,
|
| 864 |
+
"eval_test_rmse": 0.8801605701446533,
|
| 865 |
+
"eval_test_runtime": 129.0007,
|
| 866 |
+
"eval_test_samples_per_second": 2.651,
|
| 867 |
+
"eval_test_spearman": 0.83056037563114,
|
| 868 |
+
"eval_test_steps_per_second": 2.651,
|
| 869 |
+
"step": 827
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 4.996978851963746,
|
| 873 |
+
"eval_myoglobin_loss": 0.7784543037414551,
|
| 874 |
+
"eval_myoglobin_mae": 0.6274142265319824,
|
| 875 |
+
"eval_myoglobin_mse": 0.7784543037414551,
|
| 876 |
+
"eval_myoglobin_pearson": 0.6119697266512958,
|
| 877 |
+
"eval_myoglobin_rmse": 0.8823005557060242,
|
| 878 |
+
"eval_myoglobin_runtime": 50.3439,
|
| 879 |
+
"eval_myoglobin_samples_per_second": 2.662,
|
| 880 |
+
"eval_myoglobin_spearman": 0.6659828987607965,
|
| 881 |
+
"eval_myoglobin_steps_per_second": 2.662,
|
| 882 |
+
"step": 827
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 4.996978851963746,
|
| 886 |
+
"eval_myoglobin_r_loss": 0.7818257212638855,
|
| 887 |
+
"eval_myoglobin_r_mae": 0.6293807625770569,
|
| 888 |
+
"eval_myoglobin_r_mse": 0.7818257212638855,
|
| 889 |
+
"eval_myoglobin_r_pearson": 0.6113029738762634,
|
| 890 |
+
"eval_myoglobin_r_rmse": 0.8842090964317322,
|
| 891 |
+
"eval_myoglobin_r_runtime": 50.3816,
|
| 892 |
+
"eval_myoglobin_r_samples_per_second": 2.66,
|
| 893 |
+
"eval_myoglobin_r_spearman": 0.6645264107175352,
|
| 894 |
+
"eval_myoglobin_r_steps_per_second": 2.66,
|
| 895 |
+
"step": 827
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 4.996978851963746,
|
| 899 |
+
"eval_p53_loss": 3.965853452682495,
|
| 900 |
+
"eval_p53_mae": 1.5425353050231934,
|
| 901 |
+
"eval_p53_mse": 3.965853452682495,
|
| 902 |
+
"eval_p53_pearson": 0.33295329636894166,
|
| 903 |
+
"eval_p53_rmse": 1.9914450645446777,
|
| 904 |
+
"eval_p53_runtime": 17.5238,
|
| 905 |
+
"eval_p53_samples_per_second": 2.397,
|
| 906 |
+
"eval_p53_spearman": 0.2773793608694659,
|
| 907 |
+
"eval_p53_steps_per_second": 2.397,
|
| 908 |
+
"step": 827
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 5.015105740181269,
|
| 912 |
+
"grad_norm": 6.842067241668701,
|
| 913 |
+
"learning_rate": 9.467262494480869e-05,
|
| 914 |
+
"loss": 0.1314,
|
| 915 |
+
"step": 830
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 5.075528700906345,
|
| 919 |
+
"grad_norm": 3.0206170082092285,
|
| 920 |
+
"learning_rate": 9.442672041986457e-05,
|
| 921 |
+
"loss": 0.077,
|
| 922 |
+
"step": 840
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 5.13595166163142,
|
| 926 |
+
"grad_norm": 4.4932355880737305,
|
| 927 |
+
"learning_rate": 9.417560222230115e-05,
|
| 928 |
+
"loss": 0.124,
|
| 929 |
+
"step": 850
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 5.196374622356496,
|
| 933 |
+
"grad_norm": 6.419702053070068,
|
| 934 |
+
"learning_rate": 9.391929982195232e-05,
|
| 935 |
+
"loss": 0.0903,
|
| 936 |
+
"step": 860
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 5.256797583081571,
|
| 940 |
+
"grad_norm": 5.199291229248047,
|
| 941 |
+
"learning_rate": 9.365784329704115e-05,
|
| 942 |
+
"loss": 0.117,
|
| 943 |
+
"step": 870
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 5.317220543806647,
|
| 947 |
+
"grad_norm": 6.824092864990234,
|
| 948 |
+
"learning_rate": 9.339126333065007e-05,
|
| 949 |
+
"loss": 0.1264,
|
| 950 |
+
"step": 880
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.377643504531722,
|
| 954 |
+
"grad_norm": 7.177854061126709,
|
| 955 |
+
"learning_rate": 9.31195912071201e-05,
|
| 956 |
+
"loss": 0.1153,
|
| 957 |
+
"step": 890
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 5.438066465256798,
|
| 961 |
+
"grad_norm": 6.354623317718506,
|
| 962 |
+
"learning_rate": 9.284285880837946e-05,
|
| 963 |
+
"loss": 0.1371,
|
| 964 |
+
"step": 900
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 5.498489425981873,
|
| 968 |
+
"grad_norm": 7.266890525817871,
|
| 969 |
+
"learning_rate": 9.256109861020213e-05,
|
| 970 |
+
"loss": 0.1253,
|
| 971 |
+
"step": 910
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 5.5589123867069485,
|
| 975 |
+
"grad_norm": 5.081086158752441,
|
| 976 |
+
"learning_rate": 9.22743436783966e-05,
|
| 977 |
+
"loss": 0.0986,
|
| 978 |
+
"step": 920
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 5.619335347432024,
|
| 982 |
+
"grad_norm": 4.2226762771606445,
|
| 983 |
+
"learning_rate": 9.198262766492554e-05,
|
| 984 |
+
"loss": 0.0786,
|
| 985 |
+
"step": 930
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 5.6797583081570995,
|
| 989 |
+
"grad_norm": 5.185201168060303,
|
| 990 |
+
"learning_rate": 9.168598480395651e-05,
|
| 991 |
+
"loss": 0.1014,
|
| 992 |
+
"step": 940
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 5.740181268882175,
|
| 996 |
+
"grad_norm": 8.60638427734375,
|
| 997 |
+
"learning_rate": 9.138444990784453e-05,
|
| 998 |
+
"loss": 0.1176,
|
| 999 |
+
"step": 950
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 5.80060422960725,
|
| 1003 |
+
"grad_norm": 7.529942989349365,
|
| 1004 |
+
"learning_rate": 9.107805836304658e-05,
|
| 1005 |
+
"loss": 0.1253,
|
| 1006 |
+
"step": 960
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 5.861027190332326,
|
| 1010 |
+
"grad_norm": 13.042976379394531,
|
| 1011 |
+
"learning_rate": 9.076684612596891e-05,
|
| 1012 |
+
"loss": 0.1317,
|
| 1013 |
+
"step": 970
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 5.921450151057401,
|
| 1017 |
+
"grad_norm": 6.046395778656006,
|
| 1018 |
+
"learning_rate": 9.045084971874738e-05,
|
| 1019 |
+
"loss": 0.137,
|
| 1020 |
+
"step": 980
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 5.981873111782478,
|
| 1024 |
+
"grad_norm": 7.686984062194824,
|
| 1025 |
+
"learning_rate": 9.013010622496144e-05,
|
| 1026 |
+
"loss": 0.116,
|
| 1027 |
+
"step": 990
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 6.0,
|
| 1031 |
+
"eval_validation_loss": 0.7437558770179749,
|
| 1032 |
+
"eval_validation_mae": 0.560340940952301,
|
| 1033 |
+
"eval_validation_mse": 0.7437558770179749,
|
| 1034 |
+
"eval_validation_pearson": 0.8386764926773793,
|
| 1035 |
+
"eval_validation_rmse": 0.8624128103256226,
|
| 1036 |
+
"eval_validation_runtime": 128.9359,
|
| 1037 |
+
"eval_validation_samples_per_second": 2.652,
|
| 1038 |
+
"eval_validation_spearman": 0.8401724374307825,
|
| 1039 |
+
"eval_validation_steps_per_second": 2.652,
|
| 1040 |
+
"step": 993
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 6.0,
|
| 1044 |
+
"eval_test_loss": 0.7406001091003418,
|
| 1045 |
+
"eval_test_mae": 0.558407723903656,
|
| 1046 |
+
"eval_test_mse": 0.7406001091003418,
|
| 1047 |
+
"eval_test_pearson": 0.8379222565005193,
|
| 1048 |
+
"eval_test_rmse": 0.8605812788009644,
|
| 1049 |
+
"eval_test_runtime": 128.8915,
|
| 1050 |
+
"eval_test_samples_per_second": 2.653,
|
| 1051 |
+
"eval_test_spearman": 0.8392344740172174,
|
| 1052 |
+
"eval_test_steps_per_second": 2.653,
|
| 1053 |
+
"step": 993
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 6.0,
|
| 1057 |
+
"eval_myoglobin_loss": 0.7151281833648682,
|
| 1058 |
+
"eval_myoglobin_mae": 0.6059358716011047,
|
| 1059 |
+
"eval_myoglobin_mse": 0.7151281833648682,
|
| 1060 |
+
"eval_myoglobin_pearson": 0.6404604400648282,
|
| 1061 |
+
"eval_myoglobin_rmse": 0.8456525206565857,
|
| 1062 |
+
"eval_myoglobin_runtime": 50.3332,
|
| 1063 |
+
"eval_myoglobin_samples_per_second": 2.662,
|
| 1064 |
+
"eval_myoglobin_spearman": 0.695120141585149,
|
| 1065 |
+
"eval_myoglobin_steps_per_second": 2.662,
|
| 1066 |
+
"step": 993
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 6.0,
|
| 1070 |
+
"eval_myoglobin_r_loss": 0.718505322933197,
|
| 1071 |
+
"eval_myoglobin_r_mae": 0.6069437861442566,
|
| 1072 |
+
"eval_myoglobin_r_mse": 0.718505322933197,
|
| 1073 |
+
"eval_myoglobin_r_pearson": 0.6396679973590835,
|
| 1074 |
+
"eval_myoglobin_r_rmse": 0.847646951675415,
|
| 1075 |
+
"eval_myoglobin_r_runtime": 50.3333,
|
| 1076 |
+
"eval_myoglobin_r_samples_per_second": 2.662,
|
| 1077 |
+
"eval_myoglobin_r_spearman": 0.6941175590622191,
|
| 1078 |
+
"eval_myoglobin_r_steps_per_second": 2.662,
|
| 1079 |
+
"step": 993
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 6.0,
|
| 1083 |
+
"eval_p53_loss": 3.9644815921783447,
|
| 1084 |
+
"eval_p53_mae": 1.5399214029312134,
|
| 1085 |
+
"eval_p53_mse": 3.9644815921783447,
|
| 1086 |
+
"eval_p53_pearson": 0.3275408804183826,
|
| 1087 |
+
"eval_p53_rmse": 1.991100549697876,
|
| 1088 |
+
"eval_p53_runtime": 17.5252,
|
| 1089 |
+
"eval_p53_samples_per_second": 2.397,
|
| 1090 |
+
"eval_p53_spearman": 0.2655483977707391,
|
| 1091 |
+
"eval_p53_steps_per_second": 2.397,
|
| 1092 |
+
"step": 993
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 6.042296072507553,
|
| 1096 |
+
"grad_norm": 4.583357334136963,
|
| 1097 |
+
"learning_rate": 8.980465328528219e-05,
|
| 1098 |
+
"loss": 0.0736,
|
| 1099 |
+
"step": 1000
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 6.102719033232629,
|
| 1103 |
+
"grad_norm": 4.502609729766846,
|
| 1104 |
+
"learning_rate": 8.94745290930551e-05,
|
| 1105 |
+
"loss": 0.0453,
|
| 1106 |
+
"step": 1010
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 6.163141993957704,
|
| 1110 |
+
"grad_norm": 4.533963203430176,
|
| 1111 |
+
"learning_rate": 8.913977238981778e-05,
|
| 1112 |
+
"loss": 0.0728,
|
| 1113 |
+
"step": 1020
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 6.22356495468278,
|
| 1117 |
+
"grad_norm": 3.250356912612915,
|
| 1118 |
+
"learning_rate": 8.880042246075365e-05,
|
| 1119 |
+
"loss": 0.0644,
|
| 1120 |
+
"step": 1030
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 6.283987915407855,
|
| 1124 |
+
"grad_norm": 5.3714470863342285,
|
| 1125 |
+
"learning_rate": 8.845651913008145e-05,
|
| 1126 |
+
"loss": 0.0654,
|
| 1127 |
+
"step": 1040
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 6.3444108761329305,
|
| 1131 |
+
"grad_norm": 4.90172815322876,
|
| 1132 |
+
"learning_rate": 8.810810275638183e-05,
|
| 1133 |
+
"loss": 0.0976,
|
| 1134 |
+
"step": 1050
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 6.404833836858006,
|
| 1138 |
+
"grad_norm": 4.008552551269531,
|
| 1139 |
+
"learning_rate": 8.775521422786104e-05,
|
| 1140 |
+
"loss": 0.0821,
|
| 1141 |
+
"step": 1060
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 6.4652567975830815,
|
| 1145 |
+
"grad_norm": 6.362269878387451,
|
| 1146 |
+
"learning_rate": 8.739789495755253e-05,
|
| 1147 |
+
"loss": 0.0668,
|
| 1148 |
+
"step": 1070
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 6.525679758308157,
|
| 1152 |
+
"grad_norm": 3.8847382068634033,
|
| 1153 |
+
"learning_rate": 8.703618687845696e-05,
|
| 1154 |
+
"loss": 0.0964,
|
| 1155 |
+
"step": 1080
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 6.586102719033232,
|
| 1159 |
+
"grad_norm": 6.632302761077881,
|
| 1160 |
+
"learning_rate": 8.667013243862113e-05,
|
| 1161 |
+
"loss": 0.0588,
|
| 1162 |
+
"step": 1090
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 6.646525679758308,
|
| 1166 |
+
"grad_norm": 4.578704833984375,
|
| 1167 |
+
"learning_rate": 8.629977459615655e-05,
|
| 1168 |
+
"loss": 0.1196,
|
| 1169 |
+
"step": 1100
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 6.706948640483383,
|
| 1173 |
+
"grad_norm": 3.4470386505126953,
|
| 1174 |
+
"learning_rate": 8.592515681419813e-05,
|
| 1175 |
+
"loss": 0.099,
|
| 1176 |
+
"step": 1110
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 6.76737160120846,
|
| 1180 |
+
"grad_norm": 5.420199871063232,
|
| 1181 |
+
"learning_rate": 8.554632305580354e-05,
|
| 1182 |
+
"loss": 0.08,
|
| 1183 |
+
"step": 1120
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 6.827794561933535,
|
| 1187 |
+
"grad_norm": 2.7681386470794678,
|
| 1188 |
+
"learning_rate": 8.5163317778794e-05,
|
| 1189 |
+
"loss": 0.0618,
|
| 1190 |
+
"step": 1130
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.888217522658611,
|
| 1194 |
+
"grad_norm": 6.834138870239258,
|
| 1195 |
+
"learning_rate": 8.477618593053693e-05,
|
| 1196 |
+
"loss": 0.1016,
|
| 1197 |
+
"step": 1140
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 6.948640483383686,
|
| 1201 |
+
"grad_norm": 3.5948920249938965,
|
| 1202 |
+
"learning_rate": 8.438497294267117e-05,
|
| 1203 |
+
"loss": 0.0986,
|
| 1204 |
+
"step": 1150
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 6.996978851963746,
|
| 1208 |
+
"eval_validation_loss": 0.7183637619018555,
|
| 1209 |
+
"eval_validation_mae": 0.546740710735321,
|
| 1210 |
+
"eval_validation_mse": 0.7183637619018555,
|
| 1211 |
+
"eval_validation_pearson": 0.8431698872855691,
|
| 1212 |
+
"eval_validation_rmse": 0.8475634455680847,
|
| 1213 |
+
"eval_validation_runtime": 129.0329,
|
| 1214 |
+
"eval_validation_samples_per_second": 2.65,
|
| 1215 |
+
"eval_validation_spearman": 0.8431269246509225,
|
| 1216 |
+
"eval_validation_steps_per_second": 2.65,
|
| 1217 |
+
"step": 1158
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 6.996978851963746,
|
| 1221 |
+
"eval_test_loss": 0.7154301404953003,
|
| 1222 |
+
"eval_test_mae": 0.5448740720748901,
|
| 1223 |
+
"eval_test_mse": 0.7154301404953003,
|
| 1224 |
+
"eval_test_pearson": 0.8424761229464366,
|
| 1225 |
+
"eval_test_rmse": 0.845831036567688,
|
| 1226 |
+
"eval_test_runtime": 128.9049,
|
| 1227 |
+
"eval_test_samples_per_second": 2.653,
|
| 1228 |
+
"eval_test_spearman": 0.8418909616629952,
|
| 1229 |
+
"eval_test_steps_per_second": 2.653,
|
| 1230 |
+
"step": 1158
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 6.996978851963746,
|
| 1234 |
+
"eval_myoglobin_loss": 0.7281521558761597,
|
| 1235 |
+
"eval_myoglobin_mae": 0.6114689707756042,
|
| 1236 |
+
"eval_myoglobin_mse": 0.7281521558761597,
|
| 1237 |
+
"eval_myoglobin_pearson": 0.6346894753809238,
|
| 1238 |
+
"eval_myoglobin_rmse": 0.8533183336257935,
|
| 1239 |
+
"eval_myoglobin_runtime": 50.3288,
|
| 1240 |
+
"eval_myoglobin_samples_per_second": 2.662,
|
| 1241 |
+
"eval_myoglobin_spearman": 0.6774278022377248,
|
| 1242 |
+
"eval_myoglobin_steps_per_second": 2.662,
|
| 1243 |
+
"step": 1158
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 6.996978851963746,
|
| 1247 |
+
"eval_myoglobin_r_loss": 0.7321227192878723,
|
| 1248 |
+
"eval_myoglobin_r_mae": 0.6134014129638672,
|
| 1249 |
+
"eval_myoglobin_r_mse": 0.7321227192878723,
|
| 1250 |
+
"eval_myoglobin_r_pearson": 0.633906508155178,
|
| 1251 |
+
"eval_myoglobin_r_rmse": 0.8556417226791382,
|
| 1252 |
+
"eval_myoglobin_r_runtime": 50.3016,
|
| 1253 |
+
"eval_myoglobin_r_samples_per_second": 2.664,
|
| 1254 |
+
"eval_myoglobin_r_spearman": 0.6763329422189034,
|
| 1255 |
+
"eval_myoglobin_r_steps_per_second": 2.664,
|
| 1256 |
+
"step": 1158
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.996978851963746,
|
| 1260 |
+
"eval_p53_loss": 4.01906156539917,
|
| 1261 |
+
"eval_p53_mae": 1.5535447597503662,
|
| 1262 |
+
"eval_p53_mse": 4.01906156539917,
|
| 1263 |
+
"eval_p53_pearson": 0.3308597233844921,
|
| 1264 |
+
"eval_p53_rmse": 2.0047597885131836,
|
| 1265 |
+
"eval_p53_runtime": 17.5174,
|
| 1266 |
+
"eval_p53_samples_per_second": 2.398,
|
| 1267 |
+
"eval_p53_spearman": 0.3005550831039583,
|
| 1268 |
+
"eval_p53_steps_per_second": 2.398,
|
| 1269 |
+
"step": 1158
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 7.009063444108762,
|
| 1273 |
+
"grad_norm": 1.746163010597229,
|
| 1274 |
+
"learning_rate": 8.39897247257754e-05,
|
| 1275 |
+
"loss": 0.0779,
|
| 1276 |
+
"step": 1160
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 7.069486404833837,
|
| 1280 |
+
"grad_norm": 3.740064859390259,
|
| 1281 |
+
"learning_rate": 8.359048766398031e-05,
|
| 1282 |
+
"loss": 0.0478,
|
| 1283 |
+
"step": 1170
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.1299093655589125,
|
| 1287 |
+
"grad_norm": 3.6287150382995605,
|
| 1288 |
+
"learning_rate": 8.318730860952522e-05,
|
| 1289 |
+
"loss": 0.0989,
|
| 1290 |
+
"step": 1180
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 7.190332326283988,
|
| 1294 |
+
"grad_norm": 5.2572784423828125,
|
| 1295 |
+
"learning_rate": 8.278023487725982e-05,
|
| 1296 |
+
"loss": 0.0487,
|
| 1297 |
+
"step": 1190
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 7.2507552870090635,
|
| 1301 |
+
"grad_norm": 6.087055206298828,
|
| 1302 |
+
"learning_rate": 8.236931423909138e-05,
|
| 1303 |
+
"loss": 0.0552,
|
| 1304 |
+
"step": 1200
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 7.311178247734139,
|
| 1308 |
+
"grad_norm": 4.305581092834473,
|
| 1309 |
+
"learning_rate": 8.19545949183788e-05,
|
| 1310 |
+
"loss": 0.0553,
|
| 1311 |
+
"step": 1210
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 7.371601208459214,
|
| 1315 |
+
"grad_norm": 4.056112289428711,
|
| 1316 |
+
"learning_rate": 8.153612558427311e-05,
|
| 1317 |
+
"loss": 0.0463,
|
| 1318 |
+
"step": 1220
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 7.43202416918429,
|
| 1322 |
+
"grad_norm": 2.302788734436035,
|
| 1323 |
+
"learning_rate": 8.111395534600603e-05,
|
| 1324 |
+
"loss": 0.0461,
|
| 1325 |
+
"step": 1230
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.492447129909365,
|
| 1329 |
+
"grad_norm": 4.409842491149902,
|
| 1330 |
+
"learning_rate": 8.068813374712688e-05,
|
| 1331 |
+
"loss": 0.0692,
|
| 1332 |
+
"step": 1240
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 7.552870090634441,
|
| 1336 |
+
"grad_norm": 5.059557914733887,
|
| 1337 |
+
"learning_rate": 8.025871075968828e-05,
|
| 1338 |
+
"loss": 0.0462,
|
| 1339 |
+
"step": 1250
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 7.613293051359516,
|
| 1343 |
+
"grad_norm": 3.1334400177001953,
|
| 1344 |
+
"learning_rate": 7.982573677838172e-05,
|
| 1345 |
+
"loss": 0.0729,
|
| 1346 |
+
"step": 1260
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 7.673716012084592,
|
| 1350 |
+
"grad_norm": 5.167521953582764,
|
| 1351 |
+
"learning_rate": 7.938926261462366e-05,
|
| 1352 |
+
"loss": 0.0728,
|
| 1353 |
+
"step": 1270
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 7.734138972809667,
|
| 1357 |
+
"grad_norm": 7.676636219024658,
|
| 1358 |
+
"learning_rate": 7.894933949059245e-05,
|
| 1359 |
+
"loss": 0.0484,
|
| 1360 |
+
"step": 1280
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 7.794561933534744,
|
| 1364 |
+
"grad_norm": 6.20352029800415,
|
| 1365 |
+
"learning_rate": 7.850601903321716e-05,
|
| 1366 |
+
"loss": 0.067,
|
| 1367 |
+
"step": 1290
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 7.854984894259819,
|
| 1371 |
+
"grad_norm": 5.005100727081299,
|
| 1372 |
+
"learning_rate": 7.805935326811912e-05,
|
| 1373 |
+
"loss": 0.0519,
|
| 1374 |
+
"step": 1300
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 7.9154078549848945,
|
| 1378 |
+
"grad_norm": 3.7345407009124756,
|
| 1379 |
+
"learning_rate": 7.760939461350623e-05,
|
| 1380 |
+
"loss": 0.0603,
|
| 1381 |
+
"step": 1310
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 7.97583081570997,
|
| 1385 |
+
"grad_norm": 4.5358781814575195,
|
| 1386 |
+
"learning_rate": 7.715619587402164e-05,
|
| 1387 |
+
"loss": 0.0591,
|
| 1388 |
+
"step": 1320
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 8.0,
|
| 1392 |
+
"eval_validation_loss": 0.7083848118782043,
|
| 1393 |
+
"eval_validation_mae": 0.5314592719078064,
|
| 1394 |
+
"eval_validation_mse": 0.7083848118782043,
|
| 1395 |
+
"eval_validation_pearson": 0.8431805287577845,
|
| 1396 |
+
"eval_validation_rmse": 0.841655969619751,
|
| 1397 |
+
"eval_validation_runtime": 129.0006,
|
| 1398 |
+
"eval_validation_samples_per_second": 2.651,
|
| 1399 |
+
"eval_validation_spearman": 0.8425645367331428,
|
| 1400 |
+
"eval_validation_steps_per_second": 2.651,
|
| 1401 |
+
"step": 1324
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 8.0,
|
| 1405 |
+
"eval_test_loss": 0.7102228999137878,
|
| 1406 |
+
"eval_test_mae": 0.5320845246315002,
|
| 1407 |
+
"eval_test_mse": 0.7102228999137878,
|
| 1408 |
+
"eval_test_pearson": 0.8422782344343887,
|
| 1409 |
+
"eval_test_rmse": 0.8427472114562988,
|
| 1410 |
+
"eval_test_runtime": 129.0009,
|
| 1411 |
+
"eval_test_samples_per_second": 2.651,
|
| 1412 |
+
"eval_test_spearman": 0.841548097097326,
|
| 1413 |
+
"eval_test_steps_per_second": 2.651,
|
| 1414 |
+
"step": 1324
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 8.0,
|
| 1418 |
+
"eval_myoglobin_loss": 0.7520159482955933,
|
| 1419 |
+
"eval_myoglobin_mae": 0.6083745956420898,
|
| 1420 |
+
"eval_myoglobin_mse": 0.7520159482955933,
|
| 1421 |
+
"eval_myoglobin_pearson": 0.6194230091693953,
|
| 1422 |
+
"eval_myoglobin_rmse": 0.8671885132789612,
|
| 1423 |
+
"eval_myoglobin_runtime": 50.3253,
|
| 1424 |
+
"eval_myoglobin_samples_per_second": 2.663,
|
| 1425 |
+
"eval_myoglobin_spearman": 0.6743252831866682,
|
| 1426 |
+
"eval_myoglobin_steps_per_second": 2.663,
|
| 1427 |
+
"step": 1324
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 8.0,
|
| 1431 |
+
"eval_myoglobin_r_loss": 0.753574013710022,
|
| 1432 |
+
"eval_myoglobin_r_mae": 0.6102104783058167,
|
| 1433 |
+
"eval_myoglobin_r_mse": 0.753574013710022,
|
| 1434 |
+
"eval_myoglobin_r_pearson": 0.6190984211925752,
|
| 1435 |
+
"eval_myoglobin_r_rmse": 0.8680863976478577,
|
| 1436 |
+
"eval_myoglobin_r_runtime": 50.3474,
|
| 1437 |
+
"eval_myoglobin_r_samples_per_second": 2.662,
|
| 1438 |
+
"eval_myoglobin_r_spearman": 0.6751208648404359,
|
| 1439 |
+
"eval_myoglobin_r_steps_per_second": 2.662,
|
| 1440 |
+
"step": 1324
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 8.0,
|
| 1444 |
+
"eval_p53_loss": 3.998952627182007,
|
| 1445 |
+
"eval_p53_mae": 1.5620994567871094,
|
| 1446 |
+
"eval_p53_mse": 3.998952627182007,
|
| 1447 |
+
"eval_p53_pearson": 0.32915902114473283,
|
| 1448 |
+
"eval_p53_rmse": 1.999738097190857,
|
| 1449 |
+
"eval_p53_runtime": 17.5293,
|
| 1450 |
+
"eval_p53_samples_per_second": 2.396,
|
| 1451 |
+
"eval_p53_spearman": 0.2918034117706535,
|
| 1452 |
+
"eval_p53_steps_per_second": 2.396,
|
| 1453 |
+
"step": 1324
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 8.036253776435045,
|
| 1457 |
+
"grad_norm": 2.709134101867676,
|
| 1458 |
+
"learning_rate": 7.669981023454682e-05,
|
| 1459 |
+
"loss": 0.0448,
|
| 1460 |
+
"step": 1330
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 8.09667673716012,
|
| 1464 |
+
"grad_norm": 3.791492223739624,
|
| 1465 |
+
"learning_rate": 7.624029125396004e-05,
|
| 1466 |
+
"loss": 0.0381,
|
| 1467 |
+
"step": 1340
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 8.157099697885196,
|
| 1471 |
+
"grad_norm": 2.499840259552002,
|
| 1472 |
+
"learning_rate": 7.577769285885109e-05,
|
| 1473 |
+
"loss": 0.0399,
|
| 1474 |
+
"step": 1350
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 8.217522658610273,
|
| 1478 |
+
"grad_norm": 4.001781463623047,
|
| 1479 |
+
"learning_rate": 7.53120693371927e-05,
|
| 1480 |
+
"loss": 0.0473,
|
| 1481 |
+
"step": 1360
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 8.277945619335348,
|
| 1485 |
+
"grad_norm": 4.293058395385742,
|
| 1486 |
+
"learning_rate": 7.484347533196961e-05,
|
| 1487 |
+
"loss": 0.0423,
|
| 1488 |
+
"step": 1370
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 8.338368580060424,
|
| 1492 |
+
"grad_norm": 6.312211990356445,
|
| 1493 |
+
"learning_rate": 7.437196583476596e-05,
|
| 1494 |
+
"loss": 0.0607,
|
| 1495 |
+
"step": 1380
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 8.3987915407855,
|
| 1499 |
+
"grad_norm": 5.174694538116455,
|
| 1500 |
+
"learning_rate": 7.389759617931182e-05,
|
| 1501 |
+
"loss": 0.0513,
|
| 1502 |
+
"step": 1390
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 8.459214501510575,
|
| 1506 |
+
"grad_norm": 3.2399227619171143,
|
| 1507 |
+
"learning_rate": 7.342042203498951e-05,
|
| 1508 |
+
"loss": 0.0264,
|
| 1509 |
+
"step": 1400
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 8.51963746223565,
|
| 1513 |
+
"grad_norm": 3.2468111515045166,
|
| 1514 |
+
"learning_rate": 7.294049940030055e-05,
|
| 1515 |
+
"loss": 0.0397,
|
| 1516 |
+
"step": 1410
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 8.580060422960726,
|
| 1520 |
+
"grad_norm": 4.853147983551025,
|
| 1521 |
+
"learning_rate": 7.245788459629396e-05,
|
| 1522 |
+
"loss": 0.0504,
|
| 1523 |
+
"step": 1420
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 8.640483383685801,
|
| 1527 |
+
"grad_norm": 3.306142568588257,
|
| 1528 |
+
"learning_rate": 7.197263425995682e-05,
|
| 1529 |
+
"loss": 0.0405,
|
| 1530 |
+
"step": 1430
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 8.700906344410877,
|
| 1534 |
+
"grad_norm": 3.277174711227417,
|
| 1535 |
+
"learning_rate": 7.14848053375676e-05,
|
| 1536 |
+
"loss": 0.0382,
|
| 1537 |
+
"step": 1440
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 8.761329305135952,
|
| 1541 |
+
"grad_norm": 3.5067923069000244,
|
| 1542 |
+
"learning_rate": 7.099445507801323e-05,
|
| 1543 |
+
"loss": 0.0402,
|
| 1544 |
+
"step": 1450
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 8.821752265861027,
|
| 1548 |
+
"grad_norm": 4.621323108673096,
|
| 1549 |
+
"learning_rate": 7.05016410260708e-05,
|
| 1550 |
+
"loss": 0.0601,
|
| 1551 |
+
"step": 1460
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 8.882175226586103,
|
| 1555 |
+
"grad_norm": 2.868621587753296,
|
| 1556 |
+
"learning_rate": 7.000642101565434e-05,
|
| 1557 |
+
"loss": 0.0389,
|
| 1558 |
+
"step": 1470
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 8.942598187311178,
|
| 1562 |
+
"grad_norm": 3.935192584991455,
|
| 1563 |
+
"learning_rate": 6.950885316302773e-05,
|
| 1564 |
+
"loss": 0.0809,
|
| 1565 |
+
"step": 1480
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 8.996978851963746,
|
| 1569 |
+
"eval_validation_loss": 0.7205836176872253,
|
| 1570 |
+
"eval_validation_mae": 0.5384864807128906,
|
| 1571 |
+
"eval_validation_mse": 0.7205836176872253,
|
| 1572 |
+
"eval_validation_pearson": 0.8410535584549592,
|
| 1573 |
+
"eval_validation_rmse": 0.8488719463348389,
|
| 1574 |
+
"eval_validation_runtime": 129.1697,
|
| 1575 |
+
"eval_validation_samples_per_second": 2.648,
|
| 1576 |
+
"eval_validation_spearman": 0.8456297609552365,
|
| 1577 |
+
"eval_validation_steps_per_second": 2.648,
|
| 1578 |
+
"step": 1489
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 8.996978851963746,
|
| 1582 |
+
"eval_test_loss": 0.7227678298950195,
|
| 1583 |
+
"eval_test_mae": 0.5399196743965149,
|
| 1584 |
+
"eval_test_mse": 0.7227678298950195,
|
| 1585 |
+
"eval_test_pearson": 0.8400645100430921,
|
| 1586 |
+
"eval_test_rmse": 0.8501575589179993,
|
| 1587 |
+
"eval_test_runtime": 129.1979,
|
| 1588 |
+
"eval_test_samples_per_second": 2.647,
|
| 1589 |
+
"eval_test_spearman": 0.8440572355074423,
|
| 1590 |
+
"eval_test_steps_per_second": 2.647,
|
| 1591 |
+
"step": 1489
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 8.996978851963746,
|
| 1595 |
+
"eval_myoglobin_loss": 0.7287605404853821,
|
| 1596 |
+
"eval_myoglobin_mae": 0.611557126045227,
|
| 1597 |
+
"eval_myoglobin_mse": 0.7287605404853821,
|
| 1598 |
+
"eval_myoglobin_pearson": 0.6309798982049415,
|
| 1599 |
+
"eval_myoglobin_rmse": 0.8536747097969055,
|
| 1600 |
+
"eval_myoglobin_runtime": 50.4254,
|
| 1601 |
+
"eval_myoglobin_samples_per_second": 2.657,
|
| 1602 |
+
"eval_myoglobin_spearman": 0.6759837841263407,
|
| 1603 |
+
"eval_myoglobin_steps_per_second": 2.657,
|
| 1604 |
+
"step": 1489
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 8.996978851963746,
|
| 1608 |
+
"eval_myoglobin_r_loss": 0.7296391725540161,
|
| 1609 |
+
"eval_myoglobin_r_mae": 0.6120913028717041,
|
| 1610 |
+
"eval_myoglobin_r_mse": 0.7296391725540161,
|
| 1611 |
+
"eval_myoglobin_r_pearson": 0.63082406938247,
|
| 1612 |
+
"eval_myoglobin_r_rmse": 0.8541892170906067,
|
| 1613 |
+
"eval_myoglobin_r_runtime": 50.4386,
|
| 1614 |
+
"eval_myoglobin_r_samples_per_second": 2.657,
|
| 1615 |
+
"eval_myoglobin_r_spearman": 0.6778293340441719,
|
| 1616 |
+
"eval_myoglobin_r_steps_per_second": 2.657,
|
| 1617 |
+
"step": 1489
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 8.996978851963746,
|
| 1621 |
+
"eval_p53_loss": 3.9503331184387207,
|
| 1622 |
+
"eval_p53_mae": 1.5360697507858276,
|
| 1623 |
+
"eval_p53_mse": 3.9503331184387207,
|
| 1624 |
+
"eval_p53_pearson": 0.32911657672957517,
|
| 1625 |
+
"eval_p53_rmse": 1.9875445365905762,
|
| 1626 |
+
"eval_p53_runtime": 17.5407,
|
| 1627 |
+
"eval_p53_samples_per_second": 2.394,
|
| 1628 |
+
"eval_p53_spearman": 0.2984481992644589,
|
| 1629 |
+
"eval_p53_steps_per_second": 2.394,
|
| 1630 |
+
"step": 1489
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 9.003021148036254,
|
| 1634 |
+
"grad_norm": 3.8243162631988525,
|
| 1635 |
+
"learning_rate": 6.90089958599846e-05,
|
| 1636 |
+
"loss": 0.0256,
|
| 1637 |
+
"step": 1490
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 9.06344410876133,
|
| 1641 |
+
"grad_norm": 1.7380579710006714,
|
| 1642 |
+
"learning_rate": 6.850690776699573e-05,
|
| 1643 |
+
"loss": 0.0361,
|
| 1644 |
+
"step": 1500
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 9.123867069486405,
|
| 1648 |
+
"grad_norm": 3.18019962310791,
|
| 1649 |
+
"learning_rate": 6.800264780632494e-05,
|
| 1650 |
+
"loss": 0.0318,
|
| 1651 |
+
"step": 1510
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 9.18429003021148,
|
| 1655 |
+
"grad_norm": 3.6475789546966553,
|
| 1656 |
+
"learning_rate": 6.749627515511442e-05,
|
| 1657 |
+
"loss": 0.0308,
|
| 1658 |
+
"step": 1520
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 9.244712990936556,
|
| 1662 |
+
"grad_norm": 2.547883987426758,
|
| 1663 |
+
"learning_rate": 6.698784923843992e-05,
|
| 1664 |
+
"loss": 0.0266,
|
| 1665 |
+
"step": 1530
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 9.305135951661631,
|
| 1669 |
+
"grad_norm": 2.434000015258789,
|
| 1670 |
+
"learning_rate": 6.647742972233703e-05,
|
| 1671 |
+
"loss": 0.0302,
|
| 1672 |
+
"step": 1540
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 9.365558912386707,
|
| 1676 |
+
"grad_norm": 5.521008014678955,
|
| 1677 |
+
"learning_rate": 6.5965076506799e-05,
|
| 1678 |
+
"loss": 0.0358,
|
| 1679 |
+
"step": 1550
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 9.425981873111782,
|
| 1683 |
+
"grad_norm": 3.3292031288146973,
|
| 1684 |
+
"learning_rate": 6.545084971874738e-05,
|
| 1685 |
+
"loss": 0.037,
|
| 1686 |
+
"step": 1560
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 9.486404833836858,
|
| 1690 |
+
"grad_norm": 3.638929605484009,
|
| 1691 |
+
"learning_rate": 6.493480970497569e-05,
|
| 1692 |
+
"loss": 0.0422,
|
| 1693 |
+
"step": 1570
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 9.546827794561933,
|
| 1697 |
+
"grad_norm": 2.6074700355529785,
|
| 1698 |
+
"learning_rate": 6.441701702506754e-05,
|
| 1699 |
+
"loss": 0.0292,
|
| 1700 |
+
"step": 1580
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 9.607250755287009,
|
| 1704 |
+
"grad_norm": 2.0434532165527344,
|
| 1705 |
+
"learning_rate": 6.389753244428972e-05,
|
| 1706 |
+
"loss": 0.021,
|
| 1707 |
+
"step": 1590
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 9.667673716012084,
|
| 1711 |
+
"grad_norm": 4.6780524253845215,
|
| 1712 |
+
"learning_rate": 6.337641692646106e-05,
|
| 1713 |
+
"loss": 0.0366,
|
| 1714 |
+
"step": 1600
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 9.72809667673716,
|
| 1718 |
+
"grad_norm": 3.076948642730713,
|
| 1719 |
+
"learning_rate": 6.285373162679803e-05,
|
| 1720 |
+
"loss": 0.0295,
|
| 1721 |
+
"step": 1610
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 9.788519637462235,
|
| 1725 |
+
"grad_norm": 4.26541805267334,
|
| 1726 |
+
"learning_rate": 6.232953788473811e-05,
|
| 1727 |
+
"loss": 0.0274,
|
| 1728 |
+
"step": 1620
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 9.84894259818731,
|
| 1732 |
+
"grad_norm": 2.9073691368103027,
|
| 1733 |
+
"learning_rate": 6.1803897216741e-05,
|
| 1734 |
+
"loss": 0.0324,
|
| 1735 |
+
"step": 1630
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 9.909365558912386,
|
| 1739 |
+
"grad_norm": 4.8707990646362305,
|
| 1740 |
+
"learning_rate": 6.127687130906972e-05,
|
| 1741 |
+
"loss": 0.0401,
|
| 1742 |
+
"step": 1640
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 9.969788519637461,
|
| 1746 |
+
"grad_norm": 4.892948150634766,
|
| 1747 |
+
"learning_rate": 6.0748522010551215e-05,
|
| 1748 |
+
"loss": 0.0461,
|
| 1749 |
+
"step": 1650
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 10.0,
|
| 1753 |
+
"eval_validation_loss": 0.7096761465072632,
|
| 1754 |
+
"eval_validation_mae": 0.5233484506607056,
|
| 1755 |
+
"eval_validation_mse": 0.7096761465072632,
|
| 1756 |
+
"eval_validation_pearson": 0.8435451256537725,
|
| 1757 |
+
"eval_validation_rmse": 0.8424227833747864,
|
| 1758 |
+
"eval_validation_runtime": 128.988,
|
| 1759 |
+
"eval_validation_samples_per_second": 2.651,
|
| 1760 |
+
"eval_validation_spearman": 0.8466052369033387,
|
| 1761 |
+
"eval_validation_steps_per_second": 2.651,
|
| 1762 |
+
"step": 1655
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 10.0,
|
| 1766 |
+
"eval_test_loss": 0.7089855074882507,
|
| 1767 |
+
"eval_test_mae": 0.5238877534866333,
|
| 1768 |
+
"eval_test_mse": 0.7089855074882507,
|
| 1769 |
+
"eval_test_pearson": 0.8428611030663262,
|
| 1770 |
+
"eval_test_rmse": 0.8420127630233765,
|
| 1771 |
+
"eval_test_runtime": 128.948,
|
| 1772 |
+
"eval_test_samples_per_second": 2.652,
|
| 1773 |
+
"eval_test_spearman": 0.8455169232513489,
|
| 1774 |
+
"eval_test_steps_per_second": 2.652,
|
| 1775 |
+
"step": 1655
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 10.0,
|
| 1779 |
+
"eval_myoglobin_loss": 0.7252078056335449,
|
| 1780 |
+
"eval_myoglobin_mae": 0.6012104749679565,
|
| 1781 |
+
"eval_myoglobin_mse": 0.7252078056335449,
|
| 1782 |
+
"eval_myoglobin_pearson": 0.636367763746648,
|
| 1783 |
+
"eval_myoglobin_rmse": 0.8515913486480713,
|
| 1784 |
+
"eval_myoglobin_runtime": 50.3593,
|
| 1785 |
+
"eval_myoglobin_samples_per_second": 2.661,
|
| 1786 |
+
"eval_myoglobin_spearman": 0.6746395254699745,
|
| 1787 |
+
"eval_myoglobin_steps_per_second": 2.661,
|
| 1788 |
+
"step": 1655
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 10.0,
|
| 1792 |
+
"eval_myoglobin_r_loss": 0.726817786693573,
|
| 1793 |
+
"eval_myoglobin_r_mae": 0.6012217402458191,
|
| 1794 |
+
"eval_myoglobin_r_mse": 0.726817786693573,
|
| 1795 |
+
"eval_myoglobin_r_pearson": 0.6361235811248207,
|
| 1796 |
+
"eval_myoglobin_r_rmse": 0.8525360822677612,
|
| 1797 |
+
"eval_myoglobin_r_runtime": 50.3696,
|
| 1798 |
+
"eval_myoglobin_r_samples_per_second": 2.66,
|
| 1799 |
+
"eval_myoglobin_r_spearman": 0.6742305117044012,
|
| 1800 |
+
"eval_myoglobin_r_steps_per_second": 2.66,
|
| 1801 |
+
"step": 1655
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 10.0,
|
| 1805 |
+
"eval_p53_loss": 3.99676775932312,
|
| 1806 |
+
"eval_p53_mae": 1.5769025087356567,
|
| 1807 |
+
"eval_p53_mse": 3.99676775932312,
|
| 1808 |
+
"eval_p53_pearson": 0.3403924251906574,
|
| 1809 |
+
"eval_p53_rmse": 1.9991917610168457,
|
| 1810 |
+
"eval_p53_runtime": 17.5283,
|
| 1811 |
+
"eval_p53_samples_per_second": 2.396,
|
| 1812 |
+
"eval_p53_spearman": 0.2857258622336363,
|
| 1813 |
+
"eval_p53_steps_per_second": 2.396,
|
| 1814 |
+
"step": 1655
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 10.030211480362539,
|
| 1818 |
+
"grad_norm": 2.2076847553253174,
|
| 1819 |
+
"learning_rate": 6.021891132531825e-05,
|
| 1820 |
+
"loss": 0.0307,
|
| 1821 |
+
"step": 1660
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 10.090634441087614,
|
| 1825 |
+
"grad_norm": 2.394089937210083,
|
| 1826 |
+
"learning_rate": 5.9688101405532925e-05,
|
| 1827 |
+
"loss": 0.0187,
|
| 1828 |
+
"step": 1670
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 10.15105740181269,
|
| 1832 |
+
"grad_norm": 2.8219094276428223,
|
| 1833 |
+
"learning_rate": 5.9156154544092815e-05,
|
| 1834 |
+
"loss": 0.0259,
|
| 1835 |
+
"step": 1680
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 10.211480362537765,
|
| 1839 |
+
"grad_norm": 2.308112144470215,
|
| 1840 |
+
"learning_rate": 5.862313316732063e-05,
|
| 1841 |
+
"loss": 0.027,
|
| 1842 |
+
"step": 1690
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 10.27190332326284,
|
| 1846 |
+
"grad_norm": 2.3153975009918213,
|
| 1847 |
+
"learning_rate": 5.808909982763825e-05,
|
| 1848 |
+
"loss": 0.0238,
|
| 1849 |
+
"step": 1700
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 10.332326283987916,
|
| 1853 |
+
"grad_norm": 3.1248135566711426,
|
| 1854 |
+
"learning_rate": 5.7554117196225846e-05,
|
| 1855 |
+
"loss": 0.0228,
|
| 1856 |
+
"step": 1710
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 10.392749244712991,
|
| 1860 |
+
"grad_norm": 3.748237371444702,
|
| 1861 |
+
"learning_rate": 5.701824805566722e-05,
|
| 1862 |
+
"loss": 0.024,
|
| 1863 |
+
"step": 1720
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 10.453172205438067,
|
| 1867 |
+
"grad_norm": 3.601746082305908,
|
| 1868 |
+
"learning_rate": 5.6481555292581946e-05,
|
| 1869 |
+
"loss": 0.0227,
|
| 1870 |
+
"step": 1730
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 10.513595166163142,
|
| 1874 |
+
"grad_norm": 2.709296703338623,
|
| 1875 |
+
"learning_rate": 5.5944101890245324e-05,
|
| 1876 |
+
"loss": 0.0174,
|
| 1877 |
+
"step": 1740
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 10.574018126888218,
|
| 1881 |
+
"grad_norm": 2.704521656036377,
|
| 1882 |
+
"learning_rate": 5.540595092119709e-05,
|
| 1883 |
+
"loss": 0.0315,
|
| 1884 |
+
"step": 1750
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 10.634441087613293,
|
| 1888 |
+
"grad_norm": 3.2573423385620117,
|
| 1889 |
+
"learning_rate": 5.486716553983951e-05,
|
| 1890 |
+
"loss": 0.0287,
|
| 1891 |
+
"step": 1760
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 10.694864048338369,
|
| 1895 |
+
"grad_norm": 4.1254682540893555,
|
| 1896 |
+
"learning_rate": 5.432780897502589e-05,
|
| 1897 |
+
"loss": 0.0311,
|
| 1898 |
+
"step": 1770
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 10.755287009063444,
|
| 1902 |
+
"grad_norm": 4.715117454528809,
|
| 1903 |
+
"learning_rate": 5.378794452264053e-05,
|
| 1904 |
+
"loss": 0.022,
|
| 1905 |
+
"step": 1780
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 10.81570996978852,
|
| 1909 |
+
"grad_norm": 2.1660892963409424,
|
| 1910 |
+
"learning_rate": 5.324763553817054e-05,
|
| 1911 |
+
"loss": 0.0211,
|
| 1912 |
+
"step": 1790
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 10.876132930513595,
|
| 1916 |
+
"grad_norm": 3.1040217876434326,
|
| 1917 |
+
"learning_rate": 5.270694542927088e-05,
|
| 1918 |
+
"loss": 0.0347,
|
| 1919 |
+
"step": 1800
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 10.93655589123867,
|
| 1923 |
+
"grad_norm": 2.742941379547119,
|
| 1924 |
+
"learning_rate": 5.216593764832311e-05,
|
| 1925 |
+
"loss": 0.0226,
|
| 1926 |
+
"step": 1810
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 10.996978851963746,
|
| 1930 |
+
"grad_norm": 3.172571897506714,
|
| 1931 |
+
"learning_rate": 5.162467568498903e-05,
|
| 1932 |
+
"loss": 0.025,
|
| 1933 |
+
"step": 1820
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 10.996978851963746,
|
| 1937 |
+
"eval_validation_loss": 0.6895685195922852,
|
| 1938 |
+
"eval_validation_mae": 0.5120286345481873,
|
| 1939 |
+
"eval_validation_mse": 0.6895685195922852,
|
| 1940 |
+
"eval_validation_pearson": 0.8489549431017167,
|
| 1941 |
+
"eval_validation_rmse": 0.8304026126861572,
|
| 1942 |
+
"eval_validation_runtime": 128.9581,
|
| 1943 |
+
"eval_validation_samples_per_second": 2.652,
|
| 1944 |
+
"eval_validation_spearman": 0.8526459553649132,
|
| 1945 |
+
"eval_validation_steps_per_second": 2.652,
|
| 1946 |
+
"step": 1820
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 10.996978851963746,
|
| 1950 |
+
"eval_test_loss": 0.6883087158203125,
|
| 1951 |
+
"eval_test_mae": 0.5119835734367371,
|
| 1952 |
+
"eval_test_mse": 0.6883087158203125,
|
| 1953 |
+
"eval_test_pearson": 0.848377152226476,
|
| 1954 |
+
"eval_test_rmse": 0.829643726348877,
|
| 1955 |
+
"eval_test_runtime": 128.9938,
|
| 1956 |
+
"eval_test_samples_per_second": 2.651,
|
| 1957 |
+
"eval_test_spearman": 0.8513768312964548,
|
| 1958 |
+
"eval_test_steps_per_second": 2.651,
|
| 1959 |
+
"step": 1820
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 10.996978851963746,
|
| 1963 |
+
"eval_myoglobin_loss": 0.7179906368255615,
|
| 1964 |
+
"eval_myoglobin_mae": 0.5888025760650635,
|
| 1965 |
+
"eval_myoglobin_mse": 0.7179906368255615,
|
| 1966 |
+
"eval_myoglobin_pearson": 0.6391641538657495,
|
| 1967 |
+
"eval_myoglobin_rmse": 0.8473432660102844,
|
| 1968 |
+
"eval_myoglobin_runtime": 50.3591,
|
| 1969 |
+
"eval_myoglobin_samples_per_second": 2.661,
|
| 1970 |
+
"eval_myoglobin_spearman": 0.6862166102248016,
|
| 1971 |
+
"eval_myoglobin_steps_per_second": 2.661,
|
| 1972 |
+
"step": 1820
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 10.996978851963746,
|
| 1976 |
+
"eval_myoglobin_r_loss": 0.7204333543777466,
|
| 1977 |
+
"eval_myoglobin_r_mae": 0.5895242691040039,
|
| 1978 |
+
"eval_myoglobin_r_mse": 0.7204333543777466,
|
| 1979 |
+
"eval_myoglobin_r_pearson": 0.6383291059933072,
|
| 1980 |
+
"eval_myoglobin_r_rmse": 0.8487834334373474,
|
| 1981 |
+
"eval_myoglobin_r_runtime": 50.3628,
|
| 1982 |
+
"eval_myoglobin_r_samples_per_second": 2.661,
|
| 1983 |
+
"eval_myoglobin_r_spearman": 0.683303634138279,
|
| 1984 |
+
"eval_myoglobin_r_steps_per_second": 2.661,
|
| 1985 |
+
"step": 1820
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 10.996978851963746,
|
| 1989 |
+
"eval_p53_loss": 4.048507213592529,
|
| 1990 |
+
"eval_p53_mae": 1.5837998390197754,
|
| 1991 |
+
"eval_p53_mse": 4.048507213592529,
|
| 1992 |
+
"eval_p53_pearson": 0.3178191790357232,
|
| 1993 |
+
"eval_p53_rmse": 2.0120902061462402,
|
| 1994 |
+
"eval_p53_runtime": 17.5292,
|
| 1995 |
+
"eval_p53_samples_per_second": 2.396,
|
| 1996 |
+
"eval_p53_spearman": 0.2873465421101742,
|
| 1997 |
+
"eval_p53_steps_per_second": 2.396,
|
| 1998 |
+
"step": 1820
|
| 1999 |
+
}
|
| 2000 |
+
],
|
| 2001 |
+
"logging_steps": 10,
|
| 2002 |
+
"max_steps": 3300,
|
| 2003 |
+
"num_input_tokens_seen": 0,
|
| 2004 |
+
"num_train_epochs": 20,
|
| 2005 |
+
"save_steps": 500,
|
| 2006 |
+
"stateful_callbacks": {
|
| 2007 |
+
"TrainerControl": {
|
| 2008 |
+
"args": {
|
| 2009 |
+
"should_epoch_stop": false,
|
| 2010 |
+
"should_evaluate": false,
|
| 2011 |
+
"should_log": false,
|
| 2012 |
+
"should_save": true,
|
| 2013 |
+
"should_training_stop": false
|
| 2014 |
+
},
|
| 2015 |
+
"attributes": {}
|
| 2016 |
+
}
|
| 2017 |
+
},
|
| 2018 |
+
"total_flos": 0.0,
|
| 2019 |
+
"train_batch_size": 1,
|
| 2020 |
+
"trial_name": null,
|
| 2021 |
+
"trial_params": null
|
| 2022 |
+
}
|