RT @ENachshoni: 🚨 New paper out! 📄.What happens when LLMs & RLMs face conflicting answers to a question? 🤔.They often ignore disagreement a….

RT @mosh_levy: Producing reasoning texts boosts the capabilities of AI models, but do we humans correctly understand these texts? Our lates….

Join millions who have switched to Grok.

RT @AviyaMaimon: 🚨 New paper alert! 🚨.We propose an IQ Test for LLMs — a new way to evaluate models that goes beyond benchmarks and uncover….

RT @AviyaMaimon: We release:.✅ Code.✅ Leaderboard.✅ Skill matrices & tools. Let’s shift to skill‑based evaluation for LLMs!.Full paper here….

RT @biunlp: Congrats @itaimond @Tzuf6 @rtsarfaty !.

@biunlp Check out the paper, demo and code for more details. Collab w/ @obspp18 @lovodkin93 Ido Dagan @biunlp. Paper: # @huggingface: Demo: .Code:

@biunlp We invite researchers to use our benchmark (IGCS-Bench), our generic transfer-learning dataset (GenCS) and our trained SLMs to advance LLM capabilities in extractive content selection!. (5/n).

@biunlp Key finding 2️⃣: For tasks requiring longer selections, LLMs consistently perform better when processing one document at a time instead of the entire set at once. This is not so much the case for tasks with short selections. (4/n)

@biunlp Key finding 1️⃣: Training with a diverse mix of content selection tasks helps boost LLM performance even on new extractive tasks. Generic transfer learning at its best!. (3/n)

@biunlp Motivation: Many NLP tasks require selecting relevant text spans from given source texts. Despite this shared objective, such content selection tasks have traditionally been studied in isolation, each with its own modeling approaches, datasets, and evaluations. (2/n)

🚨 Introducing IGCS, accepted to #TACL!. Instruction Guided Content Selection (IGCS) unifies many tasks such as extractive summarization, evidence retrieval and argument mining under one scheme for selecting extractive spans in given sources. @biunlp.(1/n)

RT @ArieCattan: 🚨 RAG is a popular approach but what happens when the retrieved sources provide conflicting information?🤔. We're excited to….

RT @oriern1: 🧵 New paper at Findings #ACL2025 @aclmeeting!.Not all documents are processed equally well. Some consistently yield poor resul….

RT @lovodkin93: Check out our new paper on highly localized attributions, both in the input and the output!.

RT @hirscheran: 🚨 Introducing LAQuer, accepted to #ACL2025 (main conf)!. LAQuer provides more granular attribution for LLM generations: use….

RT @AlonEirew: Excited to present our system demonstration paper on EventFull — an Event-Event Relation annotation tool — at #NAACL25. Come….

RT @_akhaliq: RefVNLI. Towards Scalable Evaluation of Subject-driven Text-to-image Generation

RT @ShirAshuryTahan: LLMs struggle with tables—but how robust are they really?.🔍 ToRR goes beyond accuracy, testing real-world robustness a….

RT @litalby: 🎉 I'm happy to share that our paper, Make It Count, has been accepted to #CVPR2025!.A huge thanks to my amazing collaborators….