Training checkpoint at step 6000
Browse files- trainer_state.json +2194 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2194 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 6000,
|
| 3 |
+
"best_metric": 2.4190170764923096,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-6000",
|
| 5 |
+
"epoch": 0.12,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 6000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0005,
|
| 14 |
+
"grad_norm": 27.027176292677446,
|
| 15 |
+
"learning_rate": 4.8e-08,
|
| 16 |
+
"loss": 3.52,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.001,
|
| 21 |
+
"grad_norm": 22.883614597253285,
|
| 22 |
+
"learning_rate": 9.8e-08,
|
| 23 |
+
"loss": 3.4361,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0015,
|
| 28 |
+
"grad_norm": 14.88008652186332,
|
| 29 |
+
"learning_rate": 1.4800000000000003e-07,
|
| 30 |
+
"loss": 3.2752,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.002,
|
| 35 |
+
"grad_norm": 8.829920836438578,
|
| 36 |
+
"learning_rate": 1.9800000000000003e-07,
|
| 37 |
+
"loss": 3.073,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.002,
|
| 42 |
+
"eval_loss": 2.8928089141845703,
|
| 43 |
+
"eval_runtime": 31.5789,
|
| 44 |
+
"eval_samples_per_second": 3.23,
|
| 45 |
+
"eval_steps_per_second": 1.615,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.0025,
|
| 50 |
+
"grad_norm": 6.672581323543055,
|
| 51 |
+
"learning_rate": 2.48e-07,
|
| 52 |
+
"loss": 2.8787,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.003,
|
| 57 |
+
"grad_norm": 3.485187933164644,
|
| 58 |
+
"learning_rate": 2.9800000000000005e-07,
|
| 59 |
+
"loss": 2.7569,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.0035,
|
| 64 |
+
"grad_norm": 1.6514027733962566,
|
| 65 |
+
"learning_rate": 3.48e-07,
|
| 66 |
+
"loss": 2.683,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.004,
|
| 71 |
+
"grad_norm": 1.714322054077562,
|
| 72 |
+
"learning_rate": 3.9800000000000004e-07,
|
| 73 |
+
"loss": 2.6417,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.004,
|
| 78 |
+
"eval_loss": 2.608551263809204,
|
| 79 |
+
"eval_runtime": 31.7434,
|
| 80 |
+
"eval_samples_per_second": 3.213,
|
| 81 |
+
"eval_steps_per_second": 1.607,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.0045,
|
| 86 |
+
"grad_norm": 1.1166252829937406,
|
| 87 |
+
"learning_rate": 4.4800000000000004e-07,
|
| 88 |
+
"loss": 2.6075,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.005,
|
| 93 |
+
"grad_norm": 1.2360541139925998,
|
| 94 |
+
"learning_rate": 4.98e-07,
|
| 95 |
+
"loss": 2.5833,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.0055,
|
| 100 |
+
"grad_norm": 1.1186934925325145,
|
| 101 |
+
"learning_rate": 5.480000000000001e-07,
|
| 102 |
+
"loss": 2.568,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.006,
|
| 107 |
+
"grad_norm": 2.2165517261683907,
|
| 108 |
+
"learning_rate": 5.98e-07,
|
| 109 |
+
"loss": 2.5488,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.006,
|
| 114 |
+
"eval_loss": 2.532663345336914,
|
| 115 |
+
"eval_runtime": 31.7717,
|
| 116 |
+
"eval_samples_per_second": 3.21,
|
| 117 |
+
"eval_steps_per_second": 1.605,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.0065,
|
| 122 |
+
"grad_norm": 1.9955553189401838,
|
| 123 |
+
"learning_rate": 6.48e-07,
|
| 124 |
+
"loss": 2.5186,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.007,
|
| 129 |
+
"grad_norm": 1.7134269827298882,
|
| 130 |
+
"learning_rate": 6.98e-07,
|
| 131 |
+
"loss": 2.5133,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.0075,
|
| 136 |
+
"grad_norm": 4.086994695184575,
|
| 137 |
+
"learning_rate": 7.480000000000001e-07,
|
| 138 |
+
"loss": 2.4979,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.008,
|
| 143 |
+
"grad_norm": 2.2539165526987732,
|
| 144 |
+
"learning_rate": 7.98e-07,
|
| 145 |
+
"loss": 2.49,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.008,
|
| 150 |
+
"eval_loss": 2.4952430725097656,
|
| 151 |
+
"eval_runtime": 31.9652,
|
| 152 |
+
"eval_samples_per_second": 3.191,
|
| 153 |
+
"eval_steps_per_second": 1.595,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.0085,
|
| 158 |
+
"grad_norm": 1.138897058010547,
|
| 159 |
+
"learning_rate": 8.480000000000001e-07,
|
| 160 |
+
"loss": 2.4748,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.009,
|
| 165 |
+
"grad_norm": 1.0112216946364496,
|
| 166 |
+
"learning_rate": 8.980000000000001e-07,
|
| 167 |
+
"loss": 2.4801,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.0095,
|
| 172 |
+
"grad_norm": 1.3243191157122005,
|
| 173 |
+
"learning_rate": 9.480000000000001e-07,
|
| 174 |
+
"loss": 2.4699,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.01,
|
| 179 |
+
"grad_norm": 1.2276747327077127,
|
| 180 |
+
"learning_rate": 9.98e-07,
|
| 181 |
+
"loss": 2.468,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.01,
|
| 186 |
+
"eval_loss": 2.4748668670654297,
|
| 187 |
+
"eval_runtime": 31.7813,
|
| 188 |
+
"eval_samples_per_second": 3.209,
|
| 189 |
+
"eval_steps_per_second": 1.605,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.0105,
|
| 194 |
+
"grad_norm": 0.845856364918703,
|
| 195 |
+
"learning_rate": 1.0480000000000002e-06,
|
| 196 |
+
"loss": 2.4738,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.011,
|
| 201 |
+
"grad_norm": 1.3677643157822397,
|
| 202 |
+
"learning_rate": 1.0980000000000001e-06,
|
| 203 |
+
"loss": 2.4535,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.0115,
|
| 208 |
+
"grad_norm": 2.919464896391848,
|
| 209 |
+
"learning_rate": 1.148e-06,
|
| 210 |
+
"loss": 2.4558,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.012,
|
| 215 |
+
"grad_norm": 0.9435018771336037,
|
| 216 |
+
"learning_rate": 1.1980000000000002e-06,
|
| 217 |
+
"loss": 2.4568,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.012,
|
| 222 |
+
"eval_loss": 2.4655494689941406,
|
| 223 |
+
"eval_runtime": 31.7457,
|
| 224 |
+
"eval_samples_per_second": 3.213,
|
| 225 |
+
"eval_steps_per_second": 1.607,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.0125,
|
| 230 |
+
"grad_norm": 0.844314601352543,
|
| 231 |
+
"learning_rate": 1.248e-06,
|
| 232 |
+
"loss": 2.4493,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.013,
|
| 237 |
+
"grad_norm": 0.8266693044311944,
|
| 238 |
+
"learning_rate": 1.2980000000000001e-06,
|
| 239 |
+
"loss": 2.4491,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.0135,
|
| 244 |
+
"grad_norm": 0.9456226537014805,
|
| 245 |
+
"learning_rate": 1.348e-06,
|
| 246 |
+
"loss": 2.4538,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.014,
|
| 251 |
+
"grad_norm": 1.241067240172021,
|
| 252 |
+
"learning_rate": 1.3980000000000002e-06,
|
| 253 |
+
"loss": 2.441,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.014,
|
| 258 |
+
"eval_loss": 2.459726572036743,
|
| 259 |
+
"eval_runtime": 31.7996,
|
| 260 |
+
"eval_samples_per_second": 3.208,
|
| 261 |
+
"eval_steps_per_second": 1.604,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.0145,
|
| 266 |
+
"grad_norm": 0.8214981637560076,
|
| 267 |
+
"learning_rate": 1.4480000000000002e-06,
|
| 268 |
+
"loss": 2.4375,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.015,
|
| 273 |
+
"grad_norm": 0.8463041725741063,
|
| 274 |
+
"learning_rate": 1.498e-06,
|
| 275 |
+
"loss": 2.4476,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.0155,
|
| 280 |
+
"grad_norm": 1.0459233803315569,
|
| 281 |
+
"learning_rate": 1.548e-06,
|
| 282 |
+
"loss": 2.4388,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.016,
|
| 287 |
+
"grad_norm": 0.7899668512736558,
|
| 288 |
+
"learning_rate": 1.5980000000000002e-06,
|
| 289 |
+
"loss": 2.4376,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.016,
|
| 294 |
+
"eval_loss": 2.4541866779327393,
|
| 295 |
+
"eval_runtime": 31.8537,
|
| 296 |
+
"eval_samples_per_second": 3.202,
|
| 297 |
+
"eval_steps_per_second": 1.601,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.0165,
|
| 302 |
+
"grad_norm": 0.8397014905084252,
|
| 303 |
+
"learning_rate": 1.6480000000000001e-06,
|
| 304 |
+
"loss": 2.436,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.017,
|
| 309 |
+
"grad_norm": 0.7623848831497283,
|
| 310 |
+
"learning_rate": 1.6980000000000003e-06,
|
| 311 |
+
"loss": 2.4384,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.0175,
|
| 316 |
+
"grad_norm": 0.7990535915346776,
|
| 317 |
+
"learning_rate": 1.7480000000000002e-06,
|
| 318 |
+
"loss": 2.4388,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.018,
|
| 323 |
+
"grad_norm": 1.1027343926443682,
|
| 324 |
+
"learning_rate": 1.798e-06,
|
| 325 |
+
"loss": 2.4195,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.018,
|
| 330 |
+
"eval_loss": 2.4497900009155273,
|
| 331 |
+
"eval_runtime": 32.04,
|
| 332 |
+
"eval_samples_per_second": 3.184,
|
| 333 |
+
"eval_steps_per_second": 1.592,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.0185,
|
| 338 |
+
"grad_norm": 1.0518607606934676,
|
| 339 |
+
"learning_rate": 1.8480000000000001e-06,
|
| 340 |
+
"loss": 2.441,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.019,
|
| 345 |
+
"grad_norm": 0.7969899064558551,
|
| 346 |
+
"learning_rate": 1.898e-06,
|
| 347 |
+
"loss": 2.4416,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.0195,
|
| 352 |
+
"grad_norm": 0.6779464500616844,
|
| 353 |
+
"learning_rate": 1.9480000000000002e-06,
|
| 354 |
+
"loss": 2.4397,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.02,
|
| 359 |
+
"grad_norm": 0.8662904314628106,
|
| 360 |
+
"learning_rate": 1.998e-06,
|
| 361 |
+
"loss": 2.4316,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.02,
|
| 366 |
+
"eval_loss": 2.4468765258789062,
|
| 367 |
+
"eval_runtime": 31.891,
|
| 368 |
+
"eval_samples_per_second": 3.198,
|
| 369 |
+
"eval_steps_per_second": 1.599,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.0205,
|
| 374 |
+
"grad_norm": 0.6931713924838875,
|
| 375 |
+
"learning_rate": 2.048e-06,
|
| 376 |
+
"loss": 2.4456,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.021,
|
| 381 |
+
"grad_norm": 0.6887441871643851,
|
| 382 |
+
"learning_rate": 2.098e-06,
|
| 383 |
+
"loss": 2.4253,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.0215,
|
| 388 |
+
"grad_norm": 0.7500338911423412,
|
| 389 |
+
"learning_rate": 2.148e-06,
|
| 390 |
+
"loss": 2.431,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.022,
|
| 395 |
+
"grad_norm": 0.7458051760406093,
|
| 396 |
+
"learning_rate": 2.198e-06,
|
| 397 |
+
"loss": 2.4164,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.022,
|
| 402 |
+
"eval_loss": 2.4442293643951416,
|
| 403 |
+
"eval_runtime": 31.8697,
|
| 404 |
+
"eval_samples_per_second": 3.201,
|
| 405 |
+
"eval_steps_per_second": 1.6,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.0225,
|
| 410 |
+
"grad_norm": 0.8345425864188605,
|
| 411 |
+
"learning_rate": 2.2480000000000003e-06,
|
| 412 |
+
"loss": 2.4241,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.023,
|
| 417 |
+
"grad_norm": 0.6997049438769294,
|
| 418 |
+
"learning_rate": 2.2980000000000003e-06,
|
| 419 |
+
"loss": 2.43,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.0235,
|
| 424 |
+
"grad_norm": 0.7476759709197881,
|
| 425 |
+
"learning_rate": 2.3480000000000002e-06,
|
| 426 |
+
"loss": 2.4342,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.024,
|
| 431 |
+
"grad_norm": 0.6735584083816767,
|
| 432 |
+
"learning_rate": 2.398e-06,
|
| 433 |
+
"loss": 2.4274,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.024,
|
| 438 |
+
"eval_loss": 2.4423961639404297,
|
| 439 |
+
"eval_runtime": 31.6272,
|
| 440 |
+
"eval_samples_per_second": 3.225,
|
| 441 |
+
"eval_steps_per_second": 1.613,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.0245,
|
| 446 |
+
"grad_norm": 0.7414830106555006,
|
| 447 |
+
"learning_rate": 2.448e-06,
|
| 448 |
+
"loss": 2.4363,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 0.025,
|
| 453 |
+
"grad_norm": 0.7852755880662065,
|
| 454 |
+
"learning_rate": 2.498e-06,
|
| 455 |
+
"loss": 2.4356,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.0255,
|
| 460 |
+
"grad_norm": 0.6550676975591231,
|
| 461 |
+
"learning_rate": 2.5480000000000004e-06,
|
| 462 |
+
"loss": 2.4219,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 0.026,
|
| 467 |
+
"grad_norm": 0.6708503716821785,
|
| 468 |
+
"learning_rate": 2.598e-06,
|
| 469 |
+
"loss": 2.4442,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.026,
|
| 474 |
+
"eval_loss": 2.440678358078003,
|
| 475 |
+
"eval_runtime": 31.7661,
|
| 476 |
+
"eval_samples_per_second": 3.211,
|
| 477 |
+
"eval_steps_per_second": 1.605,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.0265,
|
| 482 |
+
"grad_norm": 0.6923805904104993,
|
| 483 |
+
"learning_rate": 2.648e-06,
|
| 484 |
+
"loss": 2.4317,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.027,
|
| 489 |
+
"grad_norm": 0.6600109660858106,
|
| 490 |
+
"learning_rate": 2.6980000000000003e-06,
|
| 491 |
+
"loss": 2.432,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.0275,
|
| 496 |
+
"grad_norm": 0.841715383150229,
|
| 497 |
+
"learning_rate": 2.748e-06,
|
| 498 |
+
"loss": 2.4196,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.028,
|
| 503 |
+
"grad_norm": 0.6392005959511108,
|
| 504 |
+
"learning_rate": 2.798e-06,
|
| 505 |
+
"loss": 2.4274,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.028,
|
| 510 |
+
"eval_loss": 2.439229726791382,
|
| 511 |
+
"eval_runtime": 32.0465,
|
| 512 |
+
"eval_samples_per_second": 3.183,
|
| 513 |
+
"eval_steps_per_second": 1.591,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.0285,
|
| 518 |
+
"grad_norm": 0.6653339947473879,
|
| 519 |
+
"learning_rate": 2.848e-06,
|
| 520 |
+
"loss": 2.4209,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.029,
|
| 525 |
+
"grad_norm": 0.6607591145573396,
|
| 526 |
+
"learning_rate": 2.8980000000000005e-06,
|
| 527 |
+
"loss": 2.4111,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.0295,
|
| 532 |
+
"grad_norm": 0.6492342012137399,
|
| 533 |
+
"learning_rate": 2.9480000000000004e-06,
|
| 534 |
+
"loss": 2.4319,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.03,
|
| 539 |
+
"grad_norm": 0.6418256237639189,
|
| 540 |
+
"learning_rate": 2.9980000000000003e-06,
|
| 541 |
+
"loss": 2.4257,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.03,
|
| 546 |
+
"eval_loss": 2.4380884170532227,
|
| 547 |
+
"eval_runtime": 32.1017,
|
| 548 |
+
"eval_samples_per_second": 3.177,
|
| 549 |
+
"eval_steps_per_second": 1.589,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.0305,
|
| 554 |
+
"grad_norm": 0.719808061901716,
|
| 555 |
+
"learning_rate": 3.0480000000000003e-06,
|
| 556 |
+
"loss": 2.4305,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.031,
|
| 561 |
+
"grad_norm": 0.6138892760464039,
|
| 562 |
+
"learning_rate": 3.0980000000000007e-06,
|
| 563 |
+
"loss": 2.4253,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.0315,
|
| 568 |
+
"grad_norm": 0.7179717159222389,
|
| 569 |
+
"learning_rate": 3.1480000000000006e-06,
|
| 570 |
+
"loss": 2.4286,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.032,
|
| 575 |
+
"grad_norm": 0.6337699388954209,
|
| 576 |
+
"learning_rate": 3.198e-06,
|
| 577 |
+
"loss": 2.4281,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.032,
|
| 582 |
+
"eval_loss": 2.4367759227752686,
|
| 583 |
+
"eval_runtime": 32.1865,
|
| 584 |
+
"eval_samples_per_second": 3.169,
|
| 585 |
+
"eval_steps_per_second": 1.585,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.0325,
|
| 590 |
+
"grad_norm": 0.6399383081078225,
|
| 591 |
+
"learning_rate": 3.248e-06,
|
| 592 |
+
"loss": 2.4127,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.033,
|
| 597 |
+
"grad_norm": 0.6239480160142674,
|
| 598 |
+
"learning_rate": 3.298e-06,
|
| 599 |
+
"loss": 2.4271,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.0335,
|
| 604 |
+
"grad_norm": 0.6964721038747086,
|
| 605 |
+
"learning_rate": 3.348e-06,
|
| 606 |
+
"loss": 2.4168,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.034,
|
| 611 |
+
"grad_norm": 0.6246300346633158,
|
| 612 |
+
"learning_rate": 3.3980000000000003e-06,
|
| 613 |
+
"loss": 2.4312,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.034,
|
| 618 |
+
"eval_loss": 2.43576717376709,
|
| 619 |
+
"eval_runtime": 32.35,
|
| 620 |
+
"eval_samples_per_second": 3.153,
|
| 621 |
+
"eval_steps_per_second": 1.577,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.0345,
|
| 626 |
+
"grad_norm": 0.6609046760569887,
|
| 627 |
+
"learning_rate": 3.4480000000000003e-06,
|
| 628 |
+
"loss": 2.4201,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.035,
|
| 633 |
+
"grad_norm": 0.611833218468793,
|
| 634 |
+
"learning_rate": 3.4980000000000002e-06,
|
| 635 |
+
"loss": 2.4248,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.0355,
|
| 640 |
+
"grad_norm": 0.6374610168215615,
|
| 641 |
+
"learning_rate": 3.548e-06,
|
| 642 |
+
"loss": 2.4195,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.036,
|
| 647 |
+
"grad_norm": 0.608911757784224,
|
| 648 |
+
"learning_rate": 3.5980000000000005e-06,
|
| 649 |
+
"loss": 2.4207,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.036,
|
| 654 |
+
"eval_loss": 2.4352190494537354,
|
| 655 |
+
"eval_runtime": 32.4107,
|
| 656 |
+
"eval_samples_per_second": 3.147,
|
| 657 |
+
"eval_steps_per_second": 1.574,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.0365,
|
| 662 |
+
"grad_norm": 0.7277576842118675,
|
| 663 |
+
"learning_rate": 3.6480000000000005e-06,
|
| 664 |
+
"loss": 2.429,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.037,
|
| 669 |
+
"grad_norm": 0.6177267450079238,
|
| 670 |
+
"learning_rate": 3.6980000000000004e-06,
|
| 671 |
+
"loss": 2.4216,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.0375,
|
| 676 |
+
"grad_norm": 0.6909621222715888,
|
| 677 |
+
"learning_rate": 3.7480000000000004e-06,
|
| 678 |
+
"loss": 2.4141,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 0.038,
|
| 683 |
+
"grad_norm": 0.6271064789808471,
|
| 684 |
+
"learning_rate": 3.7980000000000007e-06,
|
| 685 |
+
"loss": 2.4204,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.038,
|
| 690 |
+
"eval_loss": 2.434185743331909,
|
| 691 |
+
"eval_runtime": 32.1923,
|
| 692 |
+
"eval_samples_per_second": 3.168,
|
| 693 |
+
"eval_steps_per_second": 1.584,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.0385,
|
| 698 |
+
"grad_norm": 4.465543129416645,
|
| 699 |
+
"learning_rate": 3.848e-06,
|
| 700 |
+
"loss": 2.4278,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 0.039,
|
| 705 |
+
"grad_norm": 0.59428248175071,
|
| 706 |
+
"learning_rate": 3.898e-06,
|
| 707 |
+
"loss": 2.4231,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.0395,
|
| 712 |
+
"grad_norm": 0.6300066797920092,
|
| 713 |
+
"learning_rate": 3.948e-06,
|
| 714 |
+
"loss": 2.4163,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.04,
|
| 719 |
+
"grad_norm": 0.5995770487766363,
|
| 720 |
+
"learning_rate": 3.9980000000000005e-06,
|
| 721 |
+
"loss": 2.4236,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.04,
|
| 726 |
+
"eval_loss": 2.433772563934326,
|
| 727 |
+
"eval_runtime": 32.062,
|
| 728 |
+
"eval_samples_per_second": 3.181,
|
| 729 |
+
"eval_steps_per_second": 1.591,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.0405,
|
| 734 |
+
"grad_norm": 0.595289417756029,
|
| 735 |
+
"learning_rate": 4.048e-06,
|
| 736 |
+
"loss": 2.424,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.041,
|
| 741 |
+
"grad_norm": 0.6134282240517589,
|
| 742 |
+
"learning_rate": 4.098e-06,
|
| 743 |
+
"loss": 2.4255,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.0415,
|
| 748 |
+
"grad_norm": 0.6629564791128602,
|
| 749 |
+
"learning_rate": 4.148000000000001e-06,
|
| 750 |
+
"loss": 2.4097,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.042,
|
| 755 |
+
"grad_norm": 0.621927005205136,
|
| 756 |
+
"learning_rate": 4.198e-06,
|
| 757 |
+
"loss": 2.4268,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.042,
|
| 762 |
+
"eval_loss": 2.433004379272461,
|
| 763 |
+
"eval_runtime": 32.0064,
|
| 764 |
+
"eval_samples_per_second": 3.187,
|
| 765 |
+
"eval_steps_per_second": 1.593,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.0425,
|
| 770 |
+
"grad_norm": 0.5955395744872489,
|
| 771 |
+
"learning_rate": 4.248000000000001e-06,
|
| 772 |
+
"loss": 2.4134,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.043,
|
| 777 |
+
"grad_norm": 0.630503522814338,
|
| 778 |
+
"learning_rate": 4.298e-06,
|
| 779 |
+
"loss": 2.4195,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.0435,
|
| 784 |
+
"grad_norm": 0.6187515125513555,
|
| 785 |
+
"learning_rate": 4.3480000000000006e-06,
|
| 786 |
+
"loss": 2.4258,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.044,
|
| 791 |
+
"grad_norm": 0.7454395191545767,
|
| 792 |
+
"learning_rate": 4.398000000000001e-06,
|
| 793 |
+
"loss": 2.4226,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.044,
|
| 798 |
+
"eval_loss": 2.4322543144226074,
|
| 799 |
+
"eval_runtime": 31.9813,
|
| 800 |
+
"eval_samples_per_second": 3.189,
|
| 801 |
+
"eval_steps_per_second": 1.595,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 0.0445,
|
| 806 |
+
"grad_norm": 0.6347211303495337,
|
| 807 |
+
"learning_rate": 4.4480000000000004e-06,
|
| 808 |
+
"loss": 2.4191,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.045,
|
| 813 |
+
"grad_norm": 0.6135245446733344,
|
| 814 |
+
"learning_rate": 4.498e-06,
|
| 815 |
+
"loss": 2.4229,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 0.0455,
|
| 820 |
+
"grad_norm": 0.6009500019971098,
|
| 821 |
+
"learning_rate": 4.548e-06,
|
| 822 |
+
"loss": 2.42,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.046,
|
| 827 |
+
"grad_norm": 0.661258489557284,
|
| 828 |
+
"learning_rate": 4.598e-06,
|
| 829 |
+
"loss": 2.4129,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.046,
|
| 834 |
+
"eval_loss": 2.432189464569092,
|
| 835 |
+
"eval_runtime": 31.9429,
|
| 836 |
+
"eval_samples_per_second": 3.193,
|
| 837 |
+
"eval_steps_per_second": 1.597,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.0465,
|
| 842 |
+
"grad_norm": 0.6139592783182132,
|
| 843 |
+
"learning_rate": 4.648e-06,
|
| 844 |
+
"loss": 2.4104,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.047,
|
| 849 |
+
"grad_norm": 0.583220993400474,
|
| 850 |
+
"learning_rate": 4.698000000000001e-06,
|
| 851 |
+
"loss": 2.4244,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 0.0475,
|
| 856 |
+
"grad_norm": 0.6293186545915876,
|
| 857 |
+
"learning_rate": 4.748e-06,
|
| 858 |
+
"loss": 2.4225,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.048,
|
| 863 |
+
"grad_norm": 0.5798657043139257,
|
| 864 |
+
"learning_rate": 4.7980000000000005e-06,
|
| 865 |
+
"loss": 2.4283,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.048,
|
| 870 |
+
"eval_loss": 2.4312729835510254,
|
| 871 |
+
"eval_runtime": 31.7379,
|
| 872 |
+
"eval_samples_per_second": 3.214,
|
| 873 |
+
"eval_steps_per_second": 1.607,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.0485,
|
| 878 |
+
"grad_norm": 0.6301056488676946,
|
| 879 |
+
"learning_rate": 4.848000000000001e-06,
|
| 880 |
+
"loss": 2.4238,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.049,
|
| 885 |
+
"grad_norm": 0.6050753634716797,
|
| 886 |
+
"learning_rate": 4.898e-06,
|
| 887 |
+
"loss": 2.4209,
|
| 888 |
+
"step": 2450
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.0495,
|
| 892 |
+
"grad_norm": 0.5954330421177886,
|
| 893 |
+
"learning_rate": 4.948000000000001e-06,
|
| 894 |
+
"loss": 2.4208,
|
| 895 |
+
"step": 2475
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.05,
|
| 899 |
+
"grad_norm": 0.6115913011006808,
|
| 900 |
+
"learning_rate": 4.998e-06,
|
| 901 |
+
"loss": 2.4199,
|
| 902 |
+
"step": 2500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.05,
|
| 906 |
+
"eval_loss": 2.430593490600586,
|
| 907 |
+
"eval_runtime": 31.7859,
|
| 908 |
+
"eval_samples_per_second": 3.209,
|
| 909 |
+
"eval_steps_per_second": 1.604,
|
| 910 |
+
"step": 2500
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.0505,
|
| 914 |
+
"grad_norm": 0.6088167798442012,
|
| 915 |
+
"learning_rate": 5.048000000000001e-06,
|
| 916 |
+
"loss": 2.4204,
|
| 917 |
+
"step": 2525
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.051,
|
| 921 |
+
"grad_norm": 0.5886456022713933,
|
| 922 |
+
"learning_rate": 5.098000000000001e-06,
|
| 923 |
+
"loss": 2.4233,
|
| 924 |
+
"step": 2550
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.0515,
|
| 928 |
+
"grad_norm": 0.5755814876588983,
|
| 929 |
+
"learning_rate": 5.1480000000000005e-06,
|
| 930 |
+
"loss": 2.414,
|
| 931 |
+
"step": 2575
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 0.052,
|
| 935 |
+
"grad_norm": 0.6101796511458513,
|
| 936 |
+
"learning_rate": 5.198000000000001e-06,
|
| 937 |
+
"loss": 2.4134,
|
| 938 |
+
"step": 2600
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.052,
|
| 942 |
+
"eval_loss": 2.430147886276245,
|
| 943 |
+
"eval_runtime": 31.667,
|
| 944 |
+
"eval_samples_per_second": 3.221,
|
| 945 |
+
"eval_steps_per_second": 1.611,
|
| 946 |
+
"step": 2600
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 0.0525,
|
| 950 |
+
"grad_norm": 0.5829483894700689,
|
| 951 |
+
"learning_rate": 5.248000000000001e-06,
|
| 952 |
+
"loss": 2.4176,
|
| 953 |
+
"step": 2625
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.053,
|
| 957 |
+
"grad_norm": 0.5756679405925968,
|
| 958 |
+
"learning_rate": 5.298000000000001e-06,
|
| 959 |
+
"loss": 2.4196,
|
| 960 |
+
"step": 2650
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.0535,
|
| 964 |
+
"grad_norm": 0.6203149656143291,
|
| 965 |
+
"learning_rate": 5.348000000000001e-06,
|
| 966 |
+
"loss": 2.4128,
|
| 967 |
+
"step": 2675
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.054,
|
| 971 |
+
"grad_norm": 0.6107431848759605,
|
| 972 |
+
"learning_rate": 5.398e-06,
|
| 973 |
+
"loss": 2.4066,
|
| 974 |
+
"step": 2700
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.054,
|
| 978 |
+
"eval_loss": 2.4298744201660156,
|
| 979 |
+
"eval_runtime": 31.8888,
|
| 980 |
+
"eval_samples_per_second": 3.199,
|
| 981 |
+
"eval_steps_per_second": 1.599,
|
| 982 |
+
"step": 2700
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.0545,
|
| 986 |
+
"grad_norm": 0.6313360362618398,
|
| 987 |
+
"learning_rate": 5.448e-06,
|
| 988 |
+
"loss": 2.4116,
|
| 989 |
+
"step": 2725
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.055,
|
| 993 |
+
"grad_norm": 0.7567581764202255,
|
| 994 |
+
"learning_rate": 5.498e-06,
|
| 995 |
+
"loss": 2.4137,
|
| 996 |
+
"step": 2750
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.0555,
|
| 1000 |
+
"grad_norm": 0.5808819096916863,
|
| 1001 |
+
"learning_rate": 5.548e-06,
|
| 1002 |
+
"loss": 2.4261,
|
| 1003 |
+
"step": 2775
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.056,
|
| 1007 |
+
"grad_norm": 0.7401050453151701,
|
| 1008 |
+
"learning_rate": 5.5980000000000004e-06,
|
| 1009 |
+
"loss": 2.4102,
|
| 1010 |
+
"step": 2800
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.056,
|
| 1014 |
+
"eval_loss": 2.429075002670288,
|
| 1015 |
+
"eval_runtime": 31.9187,
|
| 1016 |
+
"eval_samples_per_second": 3.196,
|
| 1017 |
+
"eval_steps_per_second": 1.598,
|
| 1018 |
+
"step": 2800
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.0565,
|
| 1022 |
+
"grad_norm": 0.6100412128745759,
|
| 1023 |
+
"learning_rate": 5.648e-06,
|
| 1024 |
+
"loss": 2.4205,
|
| 1025 |
+
"step": 2825
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.057,
|
| 1029 |
+
"grad_norm": 0.6038298357908357,
|
| 1030 |
+
"learning_rate": 5.698e-06,
|
| 1031 |
+
"loss": 2.4104,
|
| 1032 |
+
"step": 2850
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.0575,
|
| 1036 |
+
"grad_norm": 0.6294303689076208,
|
| 1037 |
+
"learning_rate": 5.748e-06,
|
| 1038 |
+
"loss": 2.4101,
|
| 1039 |
+
"step": 2875
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.058,
|
| 1043 |
+
"grad_norm": 0.6000316496044382,
|
| 1044 |
+
"learning_rate": 5.798e-06,
|
| 1045 |
+
"loss": 2.4116,
|
| 1046 |
+
"step": 2900
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.058,
|
| 1050 |
+
"eval_loss": 2.428636074066162,
|
| 1051 |
+
"eval_runtime": 31.9776,
|
| 1052 |
+
"eval_samples_per_second": 3.19,
|
| 1053 |
+
"eval_steps_per_second": 1.595,
|
| 1054 |
+
"step": 2900
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.0585,
|
| 1058 |
+
"grad_norm": 0.6662370599985865,
|
| 1059 |
+
"learning_rate": 5.848000000000001e-06,
|
| 1060 |
+
"loss": 2.4271,
|
| 1061 |
+
"step": 2925
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 0.059,
|
| 1065 |
+
"grad_norm": 0.6065686333783092,
|
| 1066 |
+
"learning_rate": 5.898e-06,
|
| 1067 |
+
"loss": 2.4141,
|
| 1068 |
+
"step": 2950
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.0595,
|
| 1072 |
+
"grad_norm": 0.5896191268179571,
|
| 1073 |
+
"learning_rate": 5.9480000000000005e-06,
|
| 1074 |
+
"loss": 2.4194,
|
| 1075 |
+
"step": 2975
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.06,
|
| 1079 |
+
"grad_norm": 0.5984986372167933,
|
| 1080 |
+
"learning_rate": 5.998000000000001e-06,
|
| 1081 |
+
"loss": 2.4107,
|
| 1082 |
+
"step": 3000
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.06,
|
| 1086 |
+
"eval_loss": 2.428344488143921,
|
| 1087 |
+
"eval_runtime": 31.827,
|
| 1088 |
+
"eval_samples_per_second": 3.205,
|
| 1089 |
+
"eval_steps_per_second": 1.602,
|
| 1090 |
+
"step": 3000
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 0.0605,
|
| 1094 |
+
"grad_norm": 0.6057904687423932,
|
| 1095 |
+
"learning_rate": 6.048e-06,
|
| 1096 |
+
"loss": 2.4231,
|
| 1097 |
+
"step": 3025
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 0.061,
|
| 1101 |
+
"grad_norm": 0.5775023699888965,
|
| 1102 |
+
"learning_rate": 6.098000000000001e-06,
|
| 1103 |
+
"loss": 2.4193,
|
| 1104 |
+
"step": 3050
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.0615,
|
| 1108 |
+
"grad_norm": 0.5945486563983137,
|
| 1109 |
+
"learning_rate": 6.148e-06,
|
| 1110 |
+
"loss": 2.4101,
|
| 1111 |
+
"step": 3075
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 0.062,
|
| 1115 |
+
"grad_norm": 0.5893073406656858,
|
| 1116 |
+
"learning_rate": 6.198000000000001e-06,
|
| 1117 |
+
"loss": 2.41,
|
| 1118 |
+
"step": 3100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.062,
|
| 1122 |
+
"eval_loss": 2.4278364181518555,
|
| 1123 |
+
"eval_runtime": 31.4582,
|
| 1124 |
+
"eval_samples_per_second": 3.242,
|
| 1125 |
+
"eval_steps_per_second": 1.621,
|
| 1126 |
+
"step": 3100
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 0.0625,
|
| 1130 |
+
"grad_norm": 0.6413551002827471,
|
| 1131 |
+
"learning_rate": 6.248000000000001e-06,
|
| 1132 |
+
"loss": 2.4155,
|
| 1133 |
+
"step": 3125
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 0.063,
|
| 1137 |
+
"grad_norm": 0.5799664342522566,
|
| 1138 |
+
"learning_rate": 6.2980000000000005e-06,
|
| 1139 |
+
"loss": 2.409,
|
| 1140 |
+
"step": 3150
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.0635,
|
| 1144 |
+
"grad_norm": 0.5811811320062699,
|
| 1145 |
+
"learning_rate": 6.348000000000001e-06,
|
| 1146 |
+
"loss": 2.4103,
|
| 1147 |
+
"step": 3175
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.064,
|
| 1151 |
+
"grad_norm": 1.7009375984265656,
|
| 1152 |
+
"learning_rate": 6.398000000000001e-06,
|
| 1153 |
+
"loss": 2.4063,
|
| 1154 |
+
"step": 3200
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.064,
|
| 1158 |
+
"eval_loss": 2.4270801544189453,
|
| 1159 |
+
"eval_runtime": 31.5638,
|
| 1160 |
+
"eval_samples_per_second": 3.232,
|
| 1161 |
+
"eval_steps_per_second": 1.616,
|
| 1162 |
+
"step": 3200
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 0.0645,
|
| 1166 |
+
"grad_norm": 0.5922661228031734,
|
| 1167 |
+
"learning_rate": 6.448000000000001e-06,
|
| 1168 |
+
"loss": 2.4146,
|
| 1169 |
+
"step": 3225
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 0.065,
|
| 1173 |
+
"grad_norm": 0.6108654698148237,
|
| 1174 |
+
"learning_rate": 6.498000000000001e-06,
|
| 1175 |
+
"loss": 2.4202,
|
| 1176 |
+
"step": 3250
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 0.0655,
|
| 1180 |
+
"grad_norm": 0.5882408729466215,
|
| 1181 |
+
"learning_rate": 6.548000000000001e-06,
|
| 1182 |
+
"loss": 2.4226,
|
| 1183 |
+
"step": 3275
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 0.066,
|
| 1187 |
+
"grad_norm": 0.6095634937429834,
|
| 1188 |
+
"learning_rate": 6.598000000000001e-06,
|
| 1189 |
+
"loss": 2.4175,
|
| 1190 |
+
"step": 3300
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.066,
|
| 1194 |
+
"eval_loss": 2.4271743297576904,
|
| 1195 |
+
"eval_runtime": 31.605,
|
| 1196 |
+
"eval_samples_per_second": 3.227,
|
| 1197 |
+
"eval_steps_per_second": 1.614,
|
| 1198 |
+
"step": 3300
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.0665,
|
| 1202 |
+
"grad_norm": 0.584006486469731,
|
| 1203 |
+
"learning_rate": 6.648e-06,
|
| 1204 |
+
"loss": 2.4183,
|
| 1205 |
+
"step": 3325
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 0.067,
|
| 1209 |
+
"grad_norm": 0.6183114977641251,
|
| 1210 |
+
"learning_rate": 6.698e-06,
|
| 1211 |
+
"loss": 2.4074,
|
| 1212 |
+
"step": 3350
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 0.0675,
|
| 1216 |
+
"grad_norm": 0.6102359150325862,
|
| 1217 |
+
"learning_rate": 6.7480000000000004e-06,
|
| 1218 |
+
"loss": 2.4168,
|
| 1219 |
+
"step": 3375
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 0.068,
|
| 1223 |
+
"grad_norm": 0.6988080460632056,
|
| 1224 |
+
"learning_rate": 6.798e-06,
|
| 1225 |
+
"loss": 2.433,
|
| 1226 |
+
"step": 3400
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.068,
|
| 1230 |
+
"eval_loss": 2.4267990589141846,
|
| 1231 |
+
"eval_runtime": 31.5337,
|
| 1232 |
+
"eval_samples_per_second": 3.235,
|
| 1233 |
+
"eval_steps_per_second": 1.617,
|
| 1234 |
+
"step": 3400
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.0685,
|
| 1238 |
+
"grad_norm": 0.5923385092093629,
|
| 1239 |
+
"learning_rate": 6.848e-06,
|
| 1240 |
+
"loss": 2.4137,
|
| 1241 |
+
"step": 3425
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.069,
|
| 1245 |
+
"grad_norm": 0.5873912274008383,
|
| 1246 |
+
"learning_rate": 6.898e-06,
|
| 1247 |
+
"loss": 2.4183,
|
| 1248 |
+
"step": 3450
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.0695,
|
| 1252 |
+
"grad_norm": 0.5885684717655756,
|
| 1253 |
+
"learning_rate": 6.948e-06,
|
| 1254 |
+
"loss": 2.4282,
|
| 1255 |
+
"step": 3475
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.07,
|
| 1259 |
+
"grad_norm": 0.6026217656458652,
|
| 1260 |
+
"learning_rate": 6.998000000000001e-06,
|
| 1261 |
+
"loss": 2.4234,
|
| 1262 |
+
"step": 3500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.07,
|
| 1266 |
+
"eval_loss": 2.4262564182281494,
|
| 1267 |
+
"eval_runtime": 31.7503,
|
| 1268 |
+
"eval_samples_per_second": 3.213,
|
| 1269 |
+
"eval_steps_per_second": 1.606,
|
| 1270 |
+
"step": 3500
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.0705,
|
| 1274 |
+
"grad_norm": 0.5820881270462898,
|
| 1275 |
+
"learning_rate": 7.048e-06,
|
| 1276 |
+
"loss": 2.413,
|
| 1277 |
+
"step": 3525
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.071,
|
| 1281 |
+
"grad_norm": 0.6178510668793894,
|
| 1282 |
+
"learning_rate": 7.0980000000000005e-06,
|
| 1283 |
+
"loss": 2.3954,
|
| 1284 |
+
"step": 3550
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.0715,
|
| 1288 |
+
"grad_norm": 0.6186160369787075,
|
| 1289 |
+
"learning_rate": 7.148000000000001e-06,
|
| 1290 |
+
"loss": 2.4153,
|
| 1291 |
+
"step": 3575
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.072,
|
| 1295 |
+
"grad_norm": 0.6053079331192983,
|
| 1296 |
+
"learning_rate": 7.198e-06,
|
| 1297 |
+
"loss": 2.4061,
|
| 1298 |
+
"step": 3600
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.072,
|
| 1302 |
+
"eval_loss": 2.4260003566741943,
|
| 1303 |
+
"eval_runtime": 31.4103,
|
| 1304 |
+
"eval_samples_per_second": 3.247,
|
| 1305 |
+
"eval_steps_per_second": 1.624,
|
| 1306 |
+
"step": 3600
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.0725,
|
| 1310 |
+
"grad_norm": 0.6002224672812325,
|
| 1311 |
+
"learning_rate": 7.248000000000001e-06,
|
| 1312 |
+
"loss": 2.4062,
|
| 1313 |
+
"step": 3625
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.073,
|
| 1317 |
+
"grad_norm": 0.616881726200715,
|
| 1318 |
+
"learning_rate": 7.298e-06,
|
| 1319 |
+
"loss": 2.4167,
|
| 1320 |
+
"step": 3650
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 0.0735,
|
| 1324 |
+
"grad_norm": 0.6148731575970318,
|
| 1325 |
+
"learning_rate": 7.348000000000001e-06,
|
| 1326 |
+
"loss": 2.4123,
|
| 1327 |
+
"step": 3675
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.074,
|
| 1331 |
+
"grad_norm": 0.6221338587681139,
|
| 1332 |
+
"learning_rate": 7.398000000000001e-06,
|
| 1333 |
+
"loss": 2.4199,
|
| 1334 |
+
"step": 3700
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.074,
|
| 1338 |
+
"eval_loss": 2.4258594512939453,
|
| 1339 |
+
"eval_runtime": 31.717,
|
| 1340 |
+
"eval_samples_per_second": 3.216,
|
| 1341 |
+
"eval_steps_per_second": 1.608,
|
| 1342 |
+
"step": 3700
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 0.0745,
|
| 1346 |
+
"grad_norm": 0.6024880998969679,
|
| 1347 |
+
"learning_rate": 7.4480000000000005e-06,
|
| 1348 |
+
"loss": 2.4187,
|
| 1349 |
+
"step": 3725
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 0.075,
|
| 1353 |
+
"grad_norm": 0.5998431875234804,
|
| 1354 |
+
"learning_rate": 7.498000000000001e-06,
|
| 1355 |
+
"loss": 2.4045,
|
| 1356 |
+
"step": 3750
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 0.0755,
|
| 1360 |
+
"grad_norm": 0.5963168253580089,
|
| 1361 |
+
"learning_rate": 7.548000000000001e-06,
|
| 1362 |
+
"loss": 2.4161,
|
| 1363 |
+
"step": 3775
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 0.076,
|
| 1367 |
+
"grad_norm": 0.5891194096424622,
|
| 1368 |
+
"learning_rate": 7.598000000000001e-06,
|
| 1369 |
+
"loss": 2.4217,
|
| 1370 |
+
"step": 3800
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.076,
|
| 1374 |
+
"eval_loss": 2.425435781478882,
|
| 1375 |
+
"eval_runtime": 32.0333,
|
| 1376 |
+
"eval_samples_per_second": 3.184,
|
| 1377 |
+
"eval_steps_per_second": 1.592,
|
| 1378 |
+
"step": 3800
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 0.0765,
|
| 1382 |
+
"grad_norm": 0.6220515512248757,
|
| 1383 |
+
"learning_rate": 7.648e-06,
|
| 1384 |
+
"loss": 2.4115,
|
| 1385 |
+
"step": 3825
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.077,
|
| 1389 |
+
"grad_norm": 0.592208980582776,
|
| 1390 |
+
"learning_rate": 7.698000000000002e-06,
|
| 1391 |
+
"loss": 2.4123,
|
| 1392 |
+
"step": 3850
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 0.0775,
|
| 1396 |
+
"grad_norm": 0.6050688229723428,
|
| 1397 |
+
"learning_rate": 7.748000000000001e-06,
|
| 1398 |
+
"loss": 2.4124,
|
| 1399 |
+
"step": 3875
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 0.078,
|
| 1403 |
+
"grad_norm": 0.6128946719272819,
|
| 1404 |
+
"learning_rate": 7.798e-06,
|
| 1405 |
+
"loss": 2.4167,
|
| 1406 |
+
"step": 3900
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.078,
|
| 1410 |
+
"eval_loss": 2.4252073764801025,
|
| 1411 |
+
"eval_runtime": 31.7629,
|
| 1412 |
+
"eval_samples_per_second": 3.211,
|
| 1413 |
+
"eval_steps_per_second": 1.606,
|
| 1414 |
+
"step": 3900
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 0.0785,
|
| 1418 |
+
"grad_norm": 0.6300203936594084,
|
| 1419 |
+
"learning_rate": 7.848000000000002e-06,
|
| 1420 |
+
"loss": 2.4253,
|
| 1421 |
+
"step": 3925
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 0.079,
|
| 1425 |
+
"grad_norm": 0.622492494084331,
|
| 1426 |
+
"learning_rate": 7.898e-06,
|
| 1427 |
+
"loss": 2.4126,
|
| 1428 |
+
"step": 3950
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 0.0795,
|
| 1432 |
+
"grad_norm": 0.6054040520886763,
|
| 1433 |
+
"learning_rate": 7.948e-06,
|
| 1434 |
+
"loss": 2.4082,
|
| 1435 |
+
"step": 3975
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 0.08,
|
| 1439 |
+
"grad_norm": 0.5997365393444213,
|
| 1440 |
+
"learning_rate": 7.998e-06,
|
| 1441 |
+
"loss": 2.4187,
|
| 1442 |
+
"step": 4000
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.08,
|
| 1446 |
+
"eval_loss": 2.4248712062835693,
|
| 1447 |
+
"eval_runtime": 31.7678,
|
| 1448 |
+
"eval_samples_per_second": 3.211,
|
| 1449 |
+
"eval_steps_per_second": 1.605,
|
| 1450 |
+
"step": 4000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 0.0805,
|
| 1454 |
+
"grad_norm": 0.5914805613039377,
|
| 1455 |
+
"learning_rate": 8.048e-06,
|
| 1456 |
+
"loss": 2.4136,
|
| 1457 |
+
"step": 4025
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.081,
|
| 1461 |
+
"grad_norm": 0.6868999656119101,
|
| 1462 |
+
"learning_rate": 8.098000000000001e-06,
|
| 1463 |
+
"loss": 2.4071,
|
| 1464 |
+
"step": 4050
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 0.0815,
|
| 1468 |
+
"grad_norm": 0.6116238023737347,
|
| 1469 |
+
"learning_rate": 8.148e-06,
|
| 1470 |
+
"loss": 2.399,
|
| 1471 |
+
"step": 4075
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 0.082,
|
| 1475 |
+
"grad_norm": 0.6278682082032867,
|
| 1476 |
+
"learning_rate": 8.198e-06,
|
| 1477 |
+
"loss": 2.4147,
|
| 1478 |
+
"step": 4100
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.082,
|
| 1482 |
+
"eval_loss": 2.424673318862915,
|
| 1483 |
+
"eval_runtime": 31.702,
|
| 1484 |
+
"eval_samples_per_second": 3.217,
|
| 1485 |
+
"eval_steps_per_second": 1.609,
|
| 1486 |
+
"step": 4100
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.0825,
|
| 1490 |
+
"grad_norm": 0.652529340562497,
|
| 1491 |
+
"learning_rate": 8.248e-06,
|
| 1492 |
+
"loss": 2.4122,
|
| 1493 |
+
"step": 4125
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.083,
|
| 1497 |
+
"grad_norm": 0.6241764244719189,
|
| 1498 |
+
"learning_rate": 8.298000000000001e-06,
|
| 1499 |
+
"loss": 2.4034,
|
| 1500 |
+
"step": 4150
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.0835,
|
| 1504 |
+
"grad_norm": 0.6093599459247064,
|
| 1505 |
+
"learning_rate": 8.348e-06,
|
| 1506 |
+
"loss": 2.4184,
|
| 1507 |
+
"step": 4175
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.084,
|
| 1511 |
+
"grad_norm": 0.6145457262520279,
|
| 1512 |
+
"learning_rate": 8.398e-06,
|
| 1513 |
+
"loss": 2.4099,
|
| 1514 |
+
"step": 4200
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.084,
|
| 1518 |
+
"eval_loss": 2.424262046813965,
|
| 1519 |
+
"eval_runtime": 31.7126,
|
| 1520 |
+
"eval_samples_per_second": 3.216,
|
| 1521 |
+
"eval_steps_per_second": 1.608,
|
| 1522 |
+
"step": 4200
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.0845,
|
| 1526 |
+
"grad_norm": 0.6094287468338311,
|
| 1527 |
+
"learning_rate": 8.448000000000001e-06,
|
| 1528 |
+
"loss": 2.413,
|
| 1529 |
+
"step": 4225
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.085,
|
| 1533 |
+
"grad_norm": 0.6138052906293812,
|
| 1534 |
+
"learning_rate": 8.498e-06,
|
| 1535 |
+
"loss": 2.3935,
|
| 1536 |
+
"step": 4250
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.0855,
|
| 1540 |
+
"grad_norm": 0.6122465571930669,
|
| 1541 |
+
"learning_rate": 8.548e-06,
|
| 1542 |
+
"loss": 2.4061,
|
| 1543 |
+
"step": 4275
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.086,
|
| 1547 |
+
"grad_norm": 0.612830490698143,
|
| 1548 |
+
"learning_rate": 8.598000000000001e-06,
|
| 1549 |
+
"loss": 2.4112,
|
| 1550 |
+
"step": 4300
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.086,
|
| 1554 |
+
"eval_loss": 2.4238767623901367,
|
| 1555 |
+
"eval_runtime": 31.7292,
|
| 1556 |
+
"eval_samples_per_second": 3.215,
|
| 1557 |
+
"eval_steps_per_second": 1.607,
|
| 1558 |
+
"step": 4300
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.0865,
|
| 1562 |
+
"grad_norm": 0.628133619898939,
|
| 1563 |
+
"learning_rate": 8.648000000000001e-06,
|
| 1564 |
+
"loss": 2.4046,
|
| 1565 |
+
"step": 4325
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 0.087,
|
| 1569 |
+
"grad_norm": 0.6496528950628708,
|
| 1570 |
+
"learning_rate": 8.698e-06,
|
| 1571 |
+
"loss": 2.4068,
|
| 1572 |
+
"step": 4350
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 0.0875,
|
| 1576 |
+
"grad_norm": 0.5799286999894695,
|
| 1577 |
+
"learning_rate": 8.748000000000002e-06,
|
| 1578 |
+
"loss": 2.4072,
|
| 1579 |
+
"step": 4375
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 0.088,
|
| 1583 |
+
"grad_norm": 0.5910425054287555,
|
| 1584 |
+
"learning_rate": 8.798000000000001e-06,
|
| 1585 |
+
"loss": 2.3926,
|
| 1586 |
+
"step": 4400
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.088,
|
| 1590 |
+
"eval_loss": 2.4238674640655518,
|
| 1591 |
+
"eval_runtime": 31.7606,
|
| 1592 |
+
"eval_samples_per_second": 3.212,
|
| 1593 |
+
"eval_steps_per_second": 1.606,
|
| 1594 |
+
"step": 4400
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 0.0885,
|
| 1598 |
+
"grad_norm": 0.6159620367072861,
|
| 1599 |
+
"learning_rate": 8.848e-06,
|
| 1600 |
+
"loss": 2.4115,
|
| 1601 |
+
"step": 4425
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 0.089,
|
| 1605 |
+
"grad_norm": 0.6972746637095123,
|
| 1606 |
+
"learning_rate": 8.898000000000002e-06,
|
| 1607 |
+
"loss": 2.4105,
|
| 1608 |
+
"step": 4450
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 0.0895,
|
| 1612 |
+
"grad_norm": 0.585353172093314,
|
| 1613 |
+
"learning_rate": 8.948000000000001e-06,
|
| 1614 |
+
"loss": 2.4198,
|
| 1615 |
+
"step": 4475
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 0.09,
|
| 1619 |
+
"grad_norm": 0.6059468344943013,
|
| 1620 |
+
"learning_rate": 8.998000000000001e-06,
|
| 1621 |
+
"loss": 2.4069,
|
| 1622 |
+
"step": 4500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.09,
|
| 1626 |
+
"eval_loss": 2.42350435256958,
|
| 1627 |
+
"eval_runtime": 31.6869,
|
| 1628 |
+
"eval_samples_per_second": 3.219,
|
| 1629 |
+
"eval_steps_per_second": 1.609,
|
| 1630 |
+
"step": 4500
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 0.0905,
|
| 1634 |
+
"grad_norm": 0.6015924987371338,
|
| 1635 |
+
"learning_rate": 9.048e-06,
|
| 1636 |
+
"loss": 2.4081,
|
| 1637 |
+
"step": 4525
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 0.091,
|
| 1641 |
+
"grad_norm": 0.6006000726208087,
|
| 1642 |
+
"learning_rate": 9.098000000000002e-06,
|
| 1643 |
+
"loss": 2.4079,
|
| 1644 |
+
"step": 4550
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"epoch": 0.0915,
|
| 1648 |
+
"grad_norm": 0.6334216081429662,
|
| 1649 |
+
"learning_rate": 9.148e-06,
|
| 1650 |
+
"loss": 2.4021,
|
| 1651 |
+
"step": 4575
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 0.092,
|
| 1655 |
+
"grad_norm": 0.618758486975248,
|
| 1656 |
+
"learning_rate": 9.198e-06,
|
| 1657 |
+
"loss": 2.4191,
|
| 1658 |
+
"step": 4600
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.092,
|
| 1662 |
+
"eval_loss": 2.42366361618042,
|
| 1663 |
+
"eval_runtime": 31.7351,
|
| 1664 |
+
"eval_samples_per_second": 3.214,
|
| 1665 |
+
"eval_steps_per_second": 1.607,
|
| 1666 |
+
"step": 4600
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 0.0925,
|
| 1670 |
+
"grad_norm": 0.5982185393268022,
|
| 1671 |
+
"learning_rate": 9.248e-06,
|
| 1672 |
+
"loss": 2.4131,
|
| 1673 |
+
"step": 4625
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 0.093,
|
| 1677 |
+
"grad_norm": 0.5778256378393931,
|
| 1678 |
+
"learning_rate": 9.298e-06,
|
| 1679 |
+
"loss": 2.4105,
|
| 1680 |
+
"step": 4650
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 0.0935,
|
| 1684 |
+
"grad_norm": 0.5892823966497687,
|
| 1685 |
+
"learning_rate": 9.348000000000001e-06,
|
| 1686 |
+
"loss": 2.4146,
|
| 1687 |
+
"step": 4675
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 0.094,
|
| 1691 |
+
"grad_norm": 0.6000897787974973,
|
| 1692 |
+
"learning_rate": 9.398e-06,
|
| 1693 |
+
"loss": 2.4141,
|
| 1694 |
+
"step": 4700
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.094,
|
| 1698 |
+
"eval_loss": 2.4225125312805176,
|
| 1699 |
+
"eval_runtime": 31.7008,
|
| 1700 |
+
"eval_samples_per_second": 3.218,
|
| 1701 |
+
"eval_steps_per_second": 1.609,
|
| 1702 |
+
"step": 4700
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 0.0945,
|
| 1706 |
+
"grad_norm": 0.6317324097500899,
|
| 1707 |
+
"learning_rate": 9.448e-06,
|
| 1708 |
+
"loss": 2.4157,
|
| 1709 |
+
"step": 4725
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.095,
|
| 1713 |
+
"grad_norm": 0.6157270042215848,
|
| 1714 |
+
"learning_rate": 9.498000000000001e-06,
|
| 1715 |
+
"loss": 2.4091,
|
| 1716 |
+
"step": 4750
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 0.0955,
|
| 1720 |
+
"grad_norm": 0.5753740107095965,
|
| 1721 |
+
"learning_rate": 9.548e-06,
|
| 1722 |
+
"loss": 2.4142,
|
| 1723 |
+
"step": 4775
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 0.096,
|
| 1727 |
+
"grad_norm": 0.6173977503240126,
|
| 1728 |
+
"learning_rate": 9.598e-06,
|
| 1729 |
+
"loss": 2.4083,
|
| 1730 |
+
"step": 4800
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.096,
|
| 1734 |
+
"eval_loss": 2.422691583633423,
|
| 1735 |
+
"eval_runtime": 31.4709,
|
| 1736 |
+
"eval_samples_per_second": 3.241,
|
| 1737 |
+
"eval_steps_per_second": 1.621,
|
| 1738 |
+
"step": 4800
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.0965,
|
| 1742 |
+
"grad_norm": 0.5942953368600239,
|
| 1743 |
+
"learning_rate": 9.648000000000001e-06,
|
| 1744 |
+
"loss": 2.4087,
|
| 1745 |
+
"step": 4825
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.097,
|
| 1749 |
+
"grad_norm": 0.6555799317672051,
|
| 1750 |
+
"learning_rate": 9.698000000000001e-06,
|
| 1751 |
+
"loss": 2.4014,
|
| 1752 |
+
"step": 4850
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.0975,
|
| 1756 |
+
"grad_norm": 0.5757950367748221,
|
| 1757 |
+
"learning_rate": 9.748e-06,
|
| 1758 |
+
"loss": 2.4068,
|
| 1759 |
+
"step": 4875
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.098,
|
| 1763 |
+
"grad_norm": 0.632774385045014,
|
| 1764 |
+
"learning_rate": 9.798e-06,
|
| 1765 |
+
"loss": 2.4087,
|
| 1766 |
+
"step": 4900
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.098,
|
| 1770 |
+
"eval_loss": 2.4220755100250244,
|
| 1771 |
+
"eval_runtime": 31.4352,
|
| 1772 |
+
"eval_samples_per_second": 3.245,
|
| 1773 |
+
"eval_steps_per_second": 1.622,
|
| 1774 |
+
"step": 4900
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.0985,
|
| 1778 |
+
"grad_norm": 0.5781361622989438,
|
| 1779 |
+
"learning_rate": 9.848000000000001e-06,
|
| 1780 |
+
"loss": 2.4143,
|
| 1781 |
+
"step": 4925
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.099,
|
| 1785 |
+
"grad_norm": 0.6262568188074606,
|
| 1786 |
+
"learning_rate": 9.898e-06,
|
| 1787 |
+
"loss": 2.4142,
|
| 1788 |
+
"step": 4950
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.0995,
|
| 1792 |
+
"grad_norm": 0.6349024994263993,
|
| 1793 |
+
"learning_rate": 9.948e-06,
|
| 1794 |
+
"loss": 2.4086,
|
| 1795 |
+
"step": 4975
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.1,
|
| 1799 |
+
"grad_norm": 0.5902257687086163,
|
| 1800 |
+
"learning_rate": 9.998000000000002e-06,
|
| 1801 |
+
"loss": 2.4075,
|
| 1802 |
+
"step": 5000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.1,
|
| 1806 |
+
"eval_loss": 2.4221627712249756,
|
| 1807 |
+
"eval_runtime": 31.4547,
|
| 1808 |
+
"eval_samples_per_second": 3.243,
|
| 1809 |
+
"eval_steps_per_second": 1.621,
|
| 1810 |
+
"step": 5000
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.1005,
|
| 1814 |
+
"grad_norm": 0.6096554216132576,
|
| 1815 |
+
"learning_rate": 9.994666666666668e-06,
|
| 1816 |
+
"loss": 2.4056,
|
| 1817 |
+
"step": 5025
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 0.101,
|
| 1821 |
+
"grad_norm": 0.6157713116203616,
|
| 1822 |
+
"learning_rate": 9.989111111111111e-06,
|
| 1823 |
+
"loss": 2.4104,
|
| 1824 |
+
"step": 5050
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.1015,
|
| 1828 |
+
"grad_norm": 0.6100961136574927,
|
| 1829 |
+
"learning_rate": 9.983555555555556e-06,
|
| 1830 |
+
"loss": 2.4041,
|
| 1831 |
+
"step": 5075
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.102,
|
| 1835 |
+
"grad_norm": 0.5965243725355741,
|
| 1836 |
+
"learning_rate": 9.978000000000002e-06,
|
| 1837 |
+
"loss": 2.406,
|
| 1838 |
+
"step": 5100
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.102,
|
| 1842 |
+
"eval_loss": 2.4214208126068115,
|
| 1843 |
+
"eval_runtime": 31.4633,
|
| 1844 |
+
"eval_samples_per_second": 3.242,
|
| 1845 |
+
"eval_steps_per_second": 1.621,
|
| 1846 |
+
"step": 5100
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.1025,
|
| 1850 |
+
"grad_norm": 0.7288147495415569,
|
| 1851 |
+
"learning_rate": 9.972444444444445e-06,
|
| 1852 |
+
"loss": 2.419,
|
| 1853 |
+
"step": 5125
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 0.103,
|
| 1857 |
+
"grad_norm": 0.6027052437896476,
|
| 1858 |
+
"learning_rate": 9.966888888888889e-06,
|
| 1859 |
+
"loss": 2.4149,
|
| 1860 |
+
"step": 5150
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 0.1035,
|
| 1864 |
+
"grad_norm": 0.6351514057651396,
|
| 1865 |
+
"learning_rate": 9.961333333333334e-06,
|
| 1866 |
+
"loss": 2.4053,
|
| 1867 |
+
"step": 5175
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.104,
|
| 1871 |
+
"grad_norm": 0.5912339833990681,
|
| 1872 |
+
"learning_rate": 9.95577777777778e-06,
|
| 1873 |
+
"loss": 2.4099,
|
| 1874 |
+
"step": 5200
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.104,
|
| 1878 |
+
"eval_loss": 2.4213571548461914,
|
| 1879 |
+
"eval_runtime": 31.7689,
|
| 1880 |
+
"eval_samples_per_second": 3.211,
|
| 1881 |
+
"eval_steps_per_second": 1.605,
|
| 1882 |
+
"step": 5200
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 0.1045,
|
| 1886 |
+
"grad_norm": 0.6252419519280321,
|
| 1887 |
+
"learning_rate": 9.950222222222223e-06,
|
| 1888 |
+
"loss": 2.4044,
|
| 1889 |
+
"step": 5225
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 0.105,
|
| 1893 |
+
"grad_norm": 0.5932871252062307,
|
| 1894 |
+
"learning_rate": 9.944666666666668e-06,
|
| 1895 |
+
"loss": 2.4041,
|
| 1896 |
+
"step": 5250
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 0.1055,
|
| 1900 |
+
"grad_norm": 0.6265014889786313,
|
| 1901 |
+
"learning_rate": 9.939111111111112e-06,
|
| 1902 |
+
"loss": 2.4121,
|
| 1903 |
+
"step": 5275
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 0.106,
|
| 1907 |
+
"grad_norm": 0.5586876350334784,
|
| 1908 |
+
"learning_rate": 9.933555555555557e-06,
|
| 1909 |
+
"loss": 2.4005,
|
| 1910 |
+
"step": 5300
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.106,
|
| 1914 |
+
"eval_loss": 2.4209611415863037,
|
| 1915 |
+
"eval_runtime": 31.4697,
|
| 1916 |
+
"eval_samples_per_second": 3.241,
|
| 1917 |
+
"eval_steps_per_second": 1.621,
|
| 1918 |
+
"step": 5300
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.1065,
|
| 1922 |
+
"grad_norm": 0.6208578145519013,
|
| 1923 |
+
"learning_rate": 9.928e-06,
|
| 1924 |
+
"loss": 2.4095,
|
| 1925 |
+
"step": 5325
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 0.107,
|
| 1929 |
+
"grad_norm": 0.5761711209442947,
|
| 1930 |
+
"learning_rate": 9.922444444444446e-06,
|
| 1931 |
+
"loss": 2.411,
|
| 1932 |
+
"step": 5350
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 0.1075,
|
| 1936 |
+
"grad_norm": 0.6259961321288001,
|
| 1937 |
+
"learning_rate": 9.91688888888889e-06,
|
| 1938 |
+
"loss": 2.4062,
|
| 1939 |
+
"step": 5375
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 0.108,
|
| 1943 |
+
"grad_norm": 0.6636296843455429,
|
| 1944 |
+
"learning_rate": 9.911333333333335e-06,
|
| 1945 |
+
"loss": 2.411,
|
| 1946 |
+
"step": 5400
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.108,
|
| 1950 |
+
"eval_loss": 2.420535087585449,
|
| 1951 |
+
"eval_runtime": 31.4447,
|
| 1952 |
+
"eval_samples_per_second": 3.244,
|
| 1953 |
+
"eval_steps_per_second": 1.622,
|
| 1954 |
+
"step": 5400
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"epoch": 0.1085,
|
| 1958 |
+
"grad_norm": 0.5977322049971575,
|
| 1959 |
+
"learning_rate": 9.905777777777778e-06,
|
| 1960 |
+
"loss": 2.4073,
|
| 1961 |
+
"step": 5425
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 0.109,
|
| 1965 |
+
"grad_norm": 0.605286836273461,
|
| 1966 |
+
"learning_rate": 9.900222222222223e-06,
|
| 1967 |
+
"loss": 2.4023,
|
| 1968 |
+
"step": 5450
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 0.1095,
|
| 1972 |
+
"grad_norm": 0.6244785501127309,
|
| 1973 |
+
"learning_rate": 9.894666666666669e-06,
|
| 1974 |
+
"loss": 2.4084,
|
| 1975 |
+
"step": 5475
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 0.11,
|
| 1979 |
+
"grad_norm": 0.6135442380195029,
|
| 1980 |
+
"learning_rate": 9.889111111111112e-06,
|
| 1981 |
+
"loss": 2.4068,
|
| 1982 |
+
"step": 5500
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.11,
|
| 1986 |
+
"eval_loss": 2.4201102256774902,
|
| 1987 |
+
"eval_runtime": 31.806,
|
| 1988 |
+
"eval_samples_per_second": 3.207,
|
| 1989 |
+
"eval_steps_per_second": 1.603,
|
| 1990 |
+
"step": 5500
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.1105,
|
| 1994 |
+
"grad_norm": 0.6260082123047037,
|
| 1995 |
+
"learning_rate": 9.883555555555556e-06,
|
| 1996 |
+
"loss": 2.4053,
|
| 1997 |
+
"step": 5525
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.111,
|
| 2001 |
+
"grad_norm": 0.5956336151974914,
|
| 2002 |
+
"learning_rate": 9.878000000000001e-06,
|
| 2003 |
+
"loss": 2.4152,
|
| 2004 |
+
"step": 5550
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.1115,
|
| 2008 |
+
"grad_norm": 0.6149620176113736,
|
| 2009 |
+
"learning_rate": 9.872444444444446e-06,
|
| 2010 |
+
"loss": 2.4055,
|
| 2011 |
+
"step": 5575
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.112,
|
| 2015 |
+
"grad_norm": 0.6326092489345128,
|
| 2016 |
+
"learning_rate": 9.86688888888889e-06,
|
| 2017 |
+
"loss": 2.3968,
|
| 2018 |
+
"step": 5600
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.112,
|
| 2022 |
+
"eval_loss": 2.420125722885132,
|
| 2023 |
+
"eval_runtime": 31.8082,
|
| 2024 |
+
"eval_samples_per_second": 3.207,
|
| 2025 |
+
"eval_steps_per_second": 1.603,
|
| 2026 |
+
"step": 5600
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.1125,
|
| 2030 |
+
"grad_norm": 0.6390446494212693,
|
| 2031 |
+
"learning_rate": 9.861333333333333e-06,
|
| 2032 |
+
"loss": 2.4045,
|
| 2033 |
+
"step": 5625
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.113,
|
| 2037 |
+
"grad_norm": 0.6670896967232433,
|
| 2038 |
+
"learning_rate": 9.855777777777779e-06,
|
| 2039 |
+
"loss": 2.4013,
|
| 2040 |
+
"step": 5650
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 0.1135,
|
| 2044 |
+
"grad_norm": 0.6185087617978746,
|
| 2045 |
+
"learning_rate": 9.850222222222224e-06,
|
| 2046 |
+
"loss": 2.4015,
|
| 2047 |
+
"step": 5675
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.114,
|
| 2051 |
+
"grad_norm": 0.6040525454825223,
|
| 2052 |
+
"learning_rate": 9.844666666666667e-06,
|
| 2053 |
+
"loss": 2.4109,
|
| 2054 |
+
"step": 5700
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.114,
|
| 2058 |
+
"eval_loss": 2.419764518737793,
|
| 2059 |
+
"eval_runtime": 31.7256,
|
| 2060 |
+
"eval_samples_per_second": 3.215,
|
| 2061 |
+
"eval_steps_per_second": 1.608,
|
| 2062 |
+
"step": 5700
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.1145,
|
| 2066 |
+
"grad_norm": 0.6010942125132981,
|
| 2067 |
+
"learning_rate": 9.839111111111111e-06,
|
| 2068 |
+
"loss": 2.4092,
|
| 2069 |
+
"step": 5725
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 0.115,
|
| 2073 |
+
"grad_norm": 0.602852118998295,
|
| 2074 |
+
"learning_rate": 9.833555555555556e-06,
|
| 2075 |
+
"loss": 2.414,
|
| 2076 |
+
"step": 5750
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 0.1155,
|
| 2080 |
+
"grad_norm": 0.6189454944937772,
|
| 2081 |
+
"learning_rate": 9.828000000000001e-06,
|
| 2082 |
+
"loss": 2.4112,
|
| 2083 |
+
"step": 5775
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 0.116,
|
| 2087 |
+
"grad_norm": 0.5871735622958322,
|
| 2088 |
+
"learning_rate": 9.822444444444445e-06,
|
| 2089 |
+
"loss": 2.3993,
|
| 2090 |
+
"step": 5800
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.116,
|
| 2094 |
+
"eval_loss": 2.419255495071411,
|
| 2095 |
+
"eval_runtime": 31.7146,
|
| 2096 |
+
"eval_samples_per_second": 3.216,
|
| 2097 |
+
"eval_steps_per_second": 1.608,
|
| 2098 |
+
"step": 5800
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.1165,
|
| 2102 |
+
"grad_norm": 0.5719116548117884,
|
| 2103 |
+
"learning_rate": 9.81688888888889e-06,
|
| 2104 |
+
"loss": 2.4128,
|
| 2105 |
+
"step": 5825
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 0.117,
|
| 2109 |
+
"grad_norm": 0.5855276996729913,
|
| 2110 |
+
"learning_rate": 9.811333333333334e-06,
|
| 2111 |
+
"loss": 2.4127,
|
| 2112 |
+
"step": 5850
|
| 2113 |
+
},
|
| 2114 |
+
{
|
| 2115 |
+
"epoch": 0.1175,
|
| 2116 |
+
"grad_norm": 0.5948413134062237,
|
| 2117 |
+
"learning_rate": 9.805777777777779e-06,
|
| 2118 |
+
"loss": 2.4028,
|
| 2119 |
+
"step": 5875
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 0.118,
|
| 2123 |
+
"grad_norm": 0.6114053718118341,
|
| 2124 |
+
"learning_rate": 9.800222222222223e-06,
|
| 2125 |
+
"loss": 2.4085,
|
| 2126 |
+
"step": 5900
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.118,
|
| 2130 |
+
"eval_loss": 2.4192631244659424,
|
| 2131 |
+
"eval_runtime": 31.8221,
|
| 2132 |
+
"eval_samples_per_second": 3.205,
|
| 2133 |
+
"eval_steps_per_second": 1.603,
|
| 2134 |
+
"step": 5900
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"epoch": 0.1185,
|
| 2138 |
+
"grad_norm": 0.6171839632107143,
|
| 2139 |
+
"learning_rate": 9.794666666666668e-06,
|
| 2140 |
+
"loss": 2.4063,
|
| 2141 |
+
"step": 5925
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.119,
|
| 2145 |
+
"grad_norm": 0.5985426708940325,
|
| 2146 |
+
"learning_rate": 9.789111111111111e-06,
|
| 2147 |
+
"loss": 2.401,
|
| 2148 |
+
"step": 5950
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 0.1195,
|
| 2152 |
+
"grad_norm": 0.6242757087701617,
|
| 2153 |
+
"learning_rate": 9.783555555555557e-06,
|
| 2154 |
+
"loss": 2.3977,
|
| 2155 |
+
"step": 5975
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 0.12,
|
| 2159 |
+
"grad_norm": 0.6472329844420622,
|
| 2160 |
+
"learning_rate": 9.778e-06,
|
| 2161 |
+
"loss": 2.4066,
|
| 2162 |
+
"step": 6000
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.12,
|
| 2166 |
+
"eval_loss": 2.4190170764923096,
|
| 2167 |
+
"eval_runtime": 31.973,
|
| 2168 |
+
"eval_samples_per_second": 3.19,
|
| 2169 |
+
"eval_steps_per_second": 1.595,
|
| 2170 |
+
"step": 6000
|
| 2171 |
+
}
|
| 2172 |
+
],
|
| 2173 |
+
"logging_steps": 25,
|
| 2174 |
+
"max_steps": 50000,
|
| 2175 |
+
"num_input_tokens_seen": 0,
|
| 2176 |
+
"num_train_epochs": 9223372036854775807,
|
| 2177 |
+
"save_steps": 1000,
|
| 2178 |
+
"stateful_callbacks": {
|
| 2179 |
+
"TrainerControl": {
|
| 2180 |
+
"args": {
|
| 2181 |
+
"should_epoch_stop": false,
|
| 2182 |
+
"should_evaluate": false,
|
| 2183 |
+
"should_log": false,
|
| 2184 |
+
"should_save": true,
|
| 2185 |
+
"should_training_stop": false
|
| 2186 |
+
},
|
| 2187 |
+
"attributes": {}
|
| 2188 |
+
}
|
| 2189 |
+
},
|
| 2190 |
+
"total_flos": 1.9099213789963223e+19,
|
| 2191 |
+
"train_batch_size": 1,
|
| 2192 |
+
"trial_name": null,
|
| 2193 |
+
"trial_params": null
|
| 2194 |
+
}
|