RT @michahu8: 📢 today's scaling laws often don't work for predicting downstream task performance. For some pretraining setups, smooth and p….

RT @ChengleiSi: Are AI scientists already better than human researchers?. We recruited 43 PhD students to spend 3 months executing research….

RT @rico_angell: What causes jailbreaks to transfer between LLMs?. We find that jailbreak strength and model representation similarity pred….

RT @jcyhc_ai: LLMs won’t tell you how to make fake IDs—but will reveal the layouts/materials of IDs and make realistic photos if asked sepa….

RT @natolambert: Nice to see folks studying biases in RLHF / preference tuning all the way down to the datasets. I think many of the biases….

RT @cmalaviya11: Ever wondered what makes language models generate overly verbose, vague, or sycophantic responses?. Our new paper investig….

RT @vishakh_pk: What does it mean for #LLM output to be novel?.In work w/ @jcyhc_ai, @JanePan_, @valeriechen_, @hhexiy we argue it needs t….

RT @YulinChen99: Reasoning models overthink, generating multiple answers during reasoning. Is it because they can’t tell which ones are rig….

RT @StringChaos: Excited to release R2E-Gym. - 🔥 8.1K executable environments using synthetic data. - 🧠 Hybrid verifiers for enhanced inf….

RT @yzpang_: First set of Llama 4!!.

RT @yanda_chen_: My first paper @AnthropicAI is out!. We show that Chains-of-Thought often don’t reflect models’ true reasoning—posing chal….

RT @nsaphra: 2018: Saliency maps give plausible interpretations of random weights, triggering skepticism and catalyzing the mechinterp cult….

RT @JanePan_: When benchmarks talk, do LLMs listen?. Our new paper shows that evaluating that code LLMs with interactive feedback significa….

RT @danish037: Remember this study about how LLM generated research ideas were rated to be more novel than expert-written ones? . We find a….

New work where we show that with the right training distribution, transformers can learn to search and internally implement an exponential path-merging algo. But they struggle to learn to search as the graph size increases, and simple solns like scaling doesn't resolve it.

RT @omarsar0: Transformers Struggle to Learn to Search. Finds that transformer-based LLMs struggle to perform search robustly. Suggests t….

RT @vishakh_pk: Had a lot of fun poking holes at how LLMs capture diverse preferences with @chuanyang_jin, @hannahrosekirk and @hhexiy 🧐! N….

RT @LauraRuis: How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we….

RT @lasha_nlp: ✨I’m on the faculty job market for 2024-2025! ✨ . My research focuses on advancing Responsible AI—enhancing factuality, robu….

RT @javirandor: Anyone may be able to compromise LLMs with malicious content posted online. With just a small amount of data, adversaries c….