Taylor Webb
@TaylorWWebb
Studying cognition in humans and machines.
Excited to announce that I'll be starting a lab at the University of Montreal (psychology) and Mila (Montreal Institute of Learning Algorithms) starting summer 2025. More info to come soon, but I'll be recruiting grad students. Please share / get in touch if you're interested!
Led by postdoc Doyeon Lee and grad student Joseph Pruitt, our lab has a new Perspectives piece in PNAS Nexus: "Metacognitive sensitivity: The key to calibrating trust and optimal decision-making with AI" academic.oup.com/pnasnexus/arti… With co-authors Tianyu Zhou and Eric Du 1/
🚨 Discover the Science of LLM! We uncover how LLMs (Llama3-70B) achieve abstract reasoning through emergent symbolic mechanisms: 1️⃣ Symbol Abstraction Heads: Early layers convert input tokens into abstract variables based on their relationships. 2️⃣ Symbolic Induction Heads:…
New work led by @Aaditya6284: "Strategy coopetition explains the emergence and transience of in-context learning in transformers." We find some surprising things!! E.g. that circuits can simultaneously compete AND cooperate ("coopetition") 😯 🧵👇
Why do pre-o3 LLMs struggle with generalization tasks like @arcprize? It's not what you might think. OpenAI o3 shattered the ARC-AGI benchmark. But the hardest puzzles didn’t stump it because of reasoning, and this has implications for the benchmark as a whole. Analysis below🧵
Truly incredible results. I have been impressed with o1’s capabilities but certainly didn’t expect this leap.
Today OpenAI announced o3, its next-gen reasoning model. We've worked with OpenAI to test it on ARC-AGI, and we believe it represents a significant breakthrough in getting AI to adapt to novel tasks. It scores 75.7% on the semi-private eval in low-compute mode (for $20 per task…
Today OpenAI announced o3, its next-gen reasoning model. We've worked with OpenAI to test it on ARC-AGI, and we believe it represents a significant breakthrough in getting AI to adapt to novel tasks. It scores 75.7% on the semi-private eval in low-compute mode (for $20 per task…
Given a high-quality verifier, language model accuracy can be improved by scaling inference-time compute (e.g., w/ repeated sampling). When can we expect similar gains without an external verifier? New paper: Self-Improvement in Language Models: The Sharpening Mechanism
Introducing our new work on mechanistic intepretability of LLM cognition🤖🧠: why do Transformer-based LLMs have limited working memory capacity, as measured by N-back tasks? (1/7) openreview.net/pdf?id=dXjQgm9…
New paper with @alexanderdfung, @PramodRT9 , @jessica__chomik , @Nancy_Kanwisher, @ev_fedorenko on the representations that underlie our intuitive physical reasoning about the world. Thread 🧵about our new preprint 📄✨linked here: tinyurl.com/intphyslang 1/10
🚨 New paper at @NeurIPSConf w/ @Michael_Lepori! Most work on interpreting vision models focuses on concrete visual features (edges, objects). But how do models represent abstract visual relations between objects? We adapt NLP interpretability techniques for ViTs to find out! 🔍
Even ducklings🐣can represent abstract visual relations. Can your favorite ViT? In our new @NeurIPSConf paper, we use mechanistic interpretability to find out!
🚨 New paper at @NeurIPSConf w/ @Michael_Lepori! Most work on interpreting vision models focuses on concrete visual features (edges, objects). But how do models represent abstract visual relations between objects? We adapt NLP interpretability techniques for ViTs to find out! 🔍
In this new preprint @smfleming and I present a theory of the functions and evolution of conscious vision. This is a big project: osf.io/preprints/psya…. We'd love to get your comments!
Open Post-Training recipes! Some of my personal highlights: 💡 We significantly scaled up our preference data! (using more than 330k preference pairs for our 70b model!) 💡 We used RL with Verifiable Rewards to improve targeted skills like math and precise instruction following…
It would be great to have a precise enough formulation of ‘approximate retrieval’ for this hypothesis to be rigorously tested. There is a concern that virtually any task can be characterized in this way, by appealing to a vague notion of similarity with other tasks.
On the fallacy of "If it ain't strictly retrieval, it must be reasoning" argument.. #SundayHarangue (on Wednesday) There is a tendency among some LLM researchers to claim that LLMs must be somehow capable of doing some sort of reasoning since they are after all not doing the…
This looks like a very useful and important contribution!
How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this: Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢 🧵⬇️
How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this: Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢 🧵⬇️
More evidence that working memory is not persistent activity. Instead, it is dynamic on/off states with short-term synaptic plasticity. Intermittent rate coding and cue-specific ensembles support working memory nature.com/articles/s4158… #neuroscience
Fascinating paper from Paul Smolensky et al illustrating how transformers can implement a form of compositional symbol processing, and arguing that an emergent form of this may account for in-context learning in LLMs: arxiv.org/abs/2410.17498