#MultimodalAI23 workshop day 😀

We will give an oral presentation at #acl2023 #interpretability track. Ours will be the first to present, 9:00 sharp. A early morning I know! but I promise our work is cool and fun! Plus, there are other five amazing papers from Meta, Amazon, Oxford, Tsinghua, etc

Please come checking out my oral presentation this morning 9:00AM at Metropolitan East, Westin Hotel. It is about a new faithfulness evaluation metric which is fairer and less computing. #ACL2023 #ACL2023NLP

Reviewing for #AAAI2024 and reading reviews for #EMNLP2024 Fun night 🥹

Excited to present my recent work on Inseq and the #ICLR2024 PECoRe interpretability framework for the Sheffield NLP group this afternoon! Many thanks Cass Zhixue for inviting me! 🤗 🐛 Inseq: github.com/inseq-team/ins… 🐑🐑 PECoRe: openreview.net/forum?id=XTHfN…

Happy to share📢our paper on the model explanation of #Multilingual models got accepted in #NAACL2024 main conference. See you, Mexico City⛱️Huge shoutout to Nikos Aletras TLDR: Model is the puppet, tokenizer is the boss behind in multilingual explanations openreview.net/pdf?id=yNadDXe…

Thanks for sharing, and I am recruiting👀📢🧑‍🎓 PhD students!

ReAGent, a model-agnostic feature attribution method, is available in Inseq now, for both decoder-only and encoder-decoder. #ExplanableAI The paper is here: arxiv.org/pdf/2402.00794…

Really enjoyed showcasing research that we've been doing the past few years at Sheffield. The main premise of the lecture was that LLMs are idiot savants with remarkable abilities, performing well under extreme or non-intuitive settings (e.g. pre-training, info loss etc.)

🔍 Lighter is better! 📚 Our latest TACL paper reveals some fascinating aha moments for #LLMs. Check it out! 🚀 #NLProc

Excited to give an oral presentation on our work on model explanations NAACL HLT 2025 (19 Jun, 11:00). Please come and remember take a photo of me, forward me so that I can send it to my mum. She didn’t like the screenshot I sent for last year #ACL see the negative example👇

Please pass my respects to the organizers CambridgeLTL and uni.lu , who seem to have synchronized well with my seminar schedule cycle 😅Not sure about the causality here. Most importantly, welcome to Sheffield!

Hola ☀️the city of Taco 🇲🇽🌽🌵🐲🥑🌮 So glad I have decided to come, Mexico City is so different and interesting. Exciting to see everyone on the site today NAACL HLT 2025

Online disinfo is a big challenge. When building LLMs to detect it, let's, the cs people, learn from professional fact-checkers, and journalists. Join Jake and Carolina Scarton at poster 17 EAMT2024 in Diamond, happening now! #Disinfo #AI #cs2picks arxiv.org/abs/2406.15443

Really happy to talk about LLMs beyond performance at Cardiff NLP today. Thanks for the invitation, I really enjoyed the 2 day workshop, somehow, I found the small-sized workshop benefits communication and discussion more.

📢 Our @ShefNLP Seminar Series is back this semester! Below is the list of our speakers. #NLProc

Really happy to have Ningyu Zhang@ZJU for our Sheffield NLP seminar today. And happy new semester Sheffield Comp Sci everyone.