paper

From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyond

paperactiveprovisionalfrom-tokens-to-states-llms-as-a-special-case-of-world-models-and-the-continuous-path-beyond-fa0840e8·1 events·first seen 18h ago

Aliases: From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyond

Co-occurring entities

Yann LeCun JEPA

More like this (12)

Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs Follow the Latent Roadmap: Navigating Revocable Decoding for Diffusion LLMs with Anchor Tokens Multimodal Large Language Models long-context LLMs Do LLM Embedding Spaces Recover Expert Structure?LLM inference Where Do Models Find Happiness? Emotion Vectors in Open-Source LLMs SIMMER: Benchmarking Latent Failures in LLM Executable Planning with a World Model Continual LLM Upcycling: A Predictor-Gated Bank-Wise Sparsity Training Recipe for Dense-to-Sparse LLMs Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs Looped World Models RTLLM

Recent events (1)

5arXiv · cs.CL·18h ago·source ↗

Paper argues LLMs are a degenerate special case of world models, maps continuous spectrum from NTP to JEPA

A new arXiv preprint reframes the LLM-vs-world-model debate by arguing that LLMs are a degenerate special case of world models rather than a fundamentally different paradigm, with the state space being token sequences and the only action being token appending. The paper maps a continuous spectrum from next-token prediction through multi-token prediction, future-summary prediction, and next-latent prediction up to JEPA-style architectures. It identifies two open research challenges in moving along this spectrum: the data cliff from self-supervised text to action-labeled environments, and whether transformers generalize to continuous-state prediction or require a new architectural primitive. The work directly engages with Yann LeCun's 2022 argument that general intelligence requires abandoning autoregressive prediction.

From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyond Yann LeCun JEPA