Leshem Choshen · @LChoshen
1086 followers · 353 posts · Server sigmoid.social

You edit a model telling it Baiden is the U.S. president
Now you ask:
Who lives in the white house?
What do you think it answers? (hint: not Baiden)

@mega arxiv.org/abs/2307.12976

#nlproc #modelediting #machinelearning

Last updated 2 years ago

Leshem Choshen · @LChoshen
1086 followers · 353 posts · Server sigmoid.social

This entity wasn't in the pretraining😢
Don't cry, little ML researcher

Take new term definitions
Continue them with follow-up sentences
Distill your model (D-KL) to continue the same, without the definition
You know the new terms now
Go prompt them tiger🐯

arxiv.org/abs/2306.09306

#nlproc #modelrecycling #modelediting #machinelearning

Last updated 2 years ago