Nouha Dziri And if you want to find more papers from #NeurIPS23 we have a GPT (we embedded all of ˜4K papers) that can take just a picture like this and do the work for you! chat.openai.com/g/g-roTFoEAkP-…

Nouha Dziri @ ICLR 2024 lorenzotural.com/artificial-int…

The closest paper I came across on this fundamental question:

arxiv.org/abs/2311.00059
by Peter "@ICLR" West Ximing Lu Faeze Brahman Nouha Dziri Linjie (Lindsey) Li and many more

(thanks for sending it my way Oscar Mañas)

🥰Excited to share that I will be joining AI2 Allen Institute for AI MOSAIC this September as a predoctoral young investigator!! So excited to continue working with amazing Yejin Choi Nouha Dziri Liwei Jiang Kavel Rao and can't wait to collaborate with others!

Come see my good friend Nouha Dziri @ ICLR 2024 great #NeurIPS2023 paper in Hall B #431 happening now! Paper on the limits of transformers

A hint that current LLMs are memorizing, not reasoning.

LLMs like ChatGPT *seem* to reason to solve challenging compositional tasks—but don’t actually do so, as the solutions can be readily derived from the training data, Nouha Dziri et al find; performance will rapidly decay with increased task complexity:
buff.ly/3NJoQTL

Come listen to Linlu Qiu @ ICLR2024🇦🇹 talking about the puzzling behavior of LLMs in inductive reasoning and iterative hypothesis refinement. Hall A8-9 #ICLR2024 🔥

👀👀👀

at #ICCV2023 I'll present CHAMPAGNE introducing 1) YTDialogue18M: a large-scale (18M) dataset of video-based dialogues, 2) CHAMPAGNE: a large generative model of conversations that learned from videos.
Foyer Sud #139 Thur 2:30-4:30
Jack Hessel Nouha Dziri @ ICLR 2024 Yejin Choi Youngjae Yu

Extremely metal name for a paper all else being equal

The evidence is accumulating.

Finally some rigorous investigation of LLMs reasoning capabilities

Seungju Han Allen Institute for AI MOSAIC Yejin Choi Nouha Dziri @ ICLR 2024 Liwei Jiang Kavel Rao Congrats 🥳

Finally some solid proof to show that LLMs do NOT really reason. Great work from Nouha Dziri

Seungju Han Allen Institute for AI MOSAIC Yejin Choi Nouha Dziri Liwei Jiang Kavel Rao congrats!

Chen Cecilia Liu Jonas Pfeiffer Google DeepMind Ivan Vulić CambridgeLTL Iryna Gurevych Hendrik Schuff Jiahui Geng Heinz Koeppl Preslav Nakov MBZUAI TU Darmstadt Fajri Koto Tim Baldwin Computer Science, TU Darmstadt hessian.AI »Elastic Weight Removal for Faithful and Abstractive Dialogue Generation« by Nico Daheim, Nouha Dziri @ ICLR 2024 (@allen_ai), Mrinmaya Sachan (@ETH_en), Iryna Gurevych (@UKPLab) and Edoardo Ponti (@EdinburghNLP, The University of Edinburgh)

📑 arxiv.org/abs/2303.17574

#NAACL2024 #NLProc

Good to explore and test limits..

Jack Hessel Nouha Dziri @ ICLR 2024 Yejin Choi Youngjae Yu Codes: github.com/wade3han/champ…
Dataset: seungjuhan.me/champagne/
Paper: arxiv.org/abs/2303.09713