Загрузка видео...
Не удалось загрузить видео
Spotlight award at ICML 2024. Introducing a new direction for interpretability: Faithfulness Measurable Models. 2x-5x better explanations, more accurate faithfulness metrics, no cost, and optimal explanations. Paper: Hall C 4-9, 24 Jul (11:30 - 13:00).
49,754 просмотров • 2 лет назад •via X (Twitter)
Комментарии: 7

Looking forward to meeting you. Happy to answer any questions, just leave them below. Please also like and retweet. Co-authors: @apsarathchandar and @sivareddyg

Wrong paper link, here is the right one:

congrats!

@sophiexuguo Thanks!

Thanks. Make sure to read the right paper, as I pasted the wrong link. It's

I submitted this paper to AAAI 25. There is a lot of similarities to your work. I will remember to include your work in the camera ready version if accepted!

In this paper that just was accepted to EMNLP 2024, we also trained model with mask tokens (TM) on a different task. It almost didn’t change faithfulness. I think our model was already robust because of a more complex multi-label classification task.

