Mohamed Aghzal

You'll sometimes see the meme that NLP is solved. That's hype, and it's doing harm in the real world. But it's worth thinking about what it'd look like to actually achieve what we're aiming for. (📄 paper, thread 🧵) cims.nyu.edu/~sbowman/bowma…

If you work in the space of LLM explainability, submit your completed/ongoing/recently accepted/under review work to us! Your chance to win awards😍 @XllmReasonPlan

Happy to make the progress with my wonderful students in 2024 (reasoning&planning, LLM interpretability, human-LLM interaction)! Look forward to an exciting and fruitful 2025🎉🎉(Welcome collaborations as always! PhD openings for Fall’25)

[1/6] Mechanistic Interpretability (MI) is an emerging sub-field of interpretability that aims to understand LMs by reverse-engineering its underlying computation. Here we present a comprehensive survey curated specifically as a 𝐠𝐮𝐢𝐝𝐞 𝐟𝐨𝐫 𝐧𝐞𝐰𝐜𝐨𝐦𝐞𝐫𝐬 𝐭𝐨 𝐭𝐡𝐢𝐬…

(1/4) We have been wondering how the promise of LLM-powered Multi-Agents/Characters can benefit critical domains such as Education. Here we present 🚀MathVC🚀, a “virtual classroom” featuring multiple 🤖LLM-simulated student characters🤖 discussing and collaboratively solving…

[📣📧New Dataset on Conversational Email Event Extraction] Check out our work to be presented at EMNLP 2023! Excellent work by my PhD student @salokr_deep and collaborators. See you in Singapore!

(1/n) Do we need the expensive GPT-4 to answer all reasoning questions with different difficulty levels? No! Please follow our new preprint arxiv.org/abs/2310.03094 with @ZiyuYao and Dr. Jie Zhao in Microsoft. We achieve comparable performance but only require 40% of the costs.

🧵🇲🇦 A thread with donation links and fundraisers to help send medical aid, food, clothes and more to Morocco, which has been hit by a deadly earthquake of magnitude 6.8.

📢Registration is now open for the 2023 Machine Translation Marathon in the Americas (MTMA)! This year, MTMA will be at George Mason University on June 5-9, 2023, in Fairfax, VA! Website: mtma23.github.io Register forms.gle/eahQxxQ2fVy6bJ… if you'd like to attend! (1/3)

(1/n) Recent work shows how "(human or AI) feedback" aids LLMs, but where comes the feedback data if you want to explicitly train an LLM to understand it? Our #ACL2023 paper studies 𝘴𝘪𝘮𝘶𝘭𝘢𝘵𝘪𝘯𝘨 human natural-language feedback for error correction in semantic parsing.

I had to travel 26 hours and spend $2000+ to join #ICLR2023 in Rwanda. But people in Africa have to do this every time a conference is held in US. What happens when we make it easier to participate? 1530% higher registrations from Africa. This is important and must continue.

📢 George Mason is excited to host the 10th Mid-Atlantic Student Colloquium on Speech, Language, and Learning (MASC-SLL) on April 22nd, 2023 in Arlington VA! Check out our amazing speakers and stay tuned for updates mascsll.org. *Submission opens today till March 17*.

Birth of a neuron: Time-lapse footage of a stem cell differentiating into a brain cell. This is ultimately where minds and consciousness and thoughts and feelings come from. Credit: Nanolive

Eduard Hovy really hammering it home: NLP is about language and involves humans. It’s not ML on corpora. #acl2022nlp #NLProc