The Platonic Representation Hypothesis

The Platonic Representation Hypothesis Minyoung Huh* Brian Cheung* Tongzhou Wang* Phillip Isola* MIT Position Paper in ICML 2024 Paper Code Outline Our hypothesis How to measure convergence? Evidence of convergence What is driving convergence? What are we converging to? The world (Z) can be viewed in many different ways: in images (X), in text (Y), etc. We conjecture that representations learned on each modality on its own will converge to similar representations of Z. Conventionally, different AI systems represent the world in different ways. A vision system might represent shapes and colors, a language model might focus on syntax and semantics. However, in recent years, the architectures and objectives for modeling images and text, and many other signals, are becoming remarkably alike. Are the internal representations in these systems also converging? We argue that they are, and put forth the following hypothesis: Neural networks, trained with different objectives on diffe

--> The Platonic Representation Hypothesis The Platonic Representation Hypothesis Minyoung Huh* Brian Cheung* Tongzhou Wang* Phillip Isola* MIT Position Paper in ICML 2024 Paper Code Outline Our hypothesis How to measure convergence? Evidence of convergence What is driving convergence? What are we converging to? The world (Z) can be viewed in many different ways: in images (X), in text (Y), etc. We conjecture that representations learned on each modality on its own will converge to similar representations of Z. Conventionally, different AI systems represent the world in different ways. A visio

saved by

• Mackenzie Dion
• Ratan Kaliani
• Rikard Saqe
• Jacob G-W
• Gene Yang
• Sophie Wang

related reading

• The Platonic Representation Hypothesisarxiv.org
• [2405.07987] The Platonic Representation Hypothesisarxiv.org
• All AI Models Might Be The Same - by Jack Morrisblog.jxmo.io
• Verbalizable Representations Form a Global Workspace in Language Modelstransformer-circuits.pub
• [2602.15029] Symmetry in language statistics shapes the geometry of model representationsarxiv.org
• Language Models, World Models, and Human Model-Buildinglingo.csail.mit.edu
• Interpretability Dreamstransformer-circuits.pub
• [2210.13382] Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskarxiv.org
• Andrej Karpathy — AGI is still a decade awaydwarkesh.com
• World Models: Computing the Uncomputablenotboring.co
• Formation of Representations in Neural Networksarxiv.org
• Actually, Othello-GPT Has A Linear Emergent World Representation - Neel Nandaneelnanda.io