Transformers can learn compositional function

Channel:
United States Simons Institute for the Theory of Computing
Subscribers:
68,700
Published on ● Video Link: https://www.youtube.com/watch?v=70-ugtkvqe8


Duration: 0:00
736 views
12

Jason Lee (Princeton University)
https://simons.berkeley.edu/talks/jason-lee-princeton-university-2025-04-03
The Future of Language Models and Transformers



Other Videos By Simons Institute for the Theory of Computing


2025-04-23How to Locate Unentanglement | Quantum Colloquium
2025-04-16Talk by Sophie Morel (ENS de Lyon)
2025-04-16Challenges in State-of-the-Art Bit-Precise Reasoning
2025-04-16Talk by Yannick Forster (INRIA)
2025-04-16Testing Artificial Mathematical Intelligence
2025-04-16Adventures with an Automatic Prover
2025-04-16Formal Reasoning Meets LLMs: Toward AI for Mathematics and Verification
2025-04-16How can Machine Learning Help Mathematicians?
2025-04-16Computer-Assisted Intuition: SAT Solvers in Mathematical Discovery
2025-04-07The Move Toward AGI: Why Large Language Models Surprised Almost Everyone... | Theoretically Speaking
2025-04-07Transformers can learn compositional function
2025-04-07Advancing Diffusion Models for Text Generation
2025-04-07Inference Scaling: A New Frontier for AI Capability
2025-04-07Predicting and optimizing the behavior of large ML models
2025-04-07Mixed-modal Language Modeling: Chameleon, Transfusion, and Mixture of Transformers
2025-04-07Reducing the Dimension of Language: A Spectral Perspective on Transformers
2025-04-07LLM skills and meta-cognition: scaffolding for new forms of learning?
2025-04-07The Key Ingredients of Optimizing Test-Time Compute and What's Still Missing
2025-04-07Controllable and Creative Natural Language Generation
2025-04-04Field-based decoders revisited (partial)
2025-03-24Topological quantum spin glass order and its realization in qLDPC codes (partial)