SDE Matching extended beyond one-time marginals to learn generative models in physical time
canaesseth · x · 2026-10-06
- The author highlights work extending SDE Matching beyond one-time marginals, enabling generative models learned directly in continuous physical time and avoiding the pseudo-time nature of diffusion models, flow matching, and stochastic interpolants.
- The original SDE Matching paper (Bartosh et al.) introduces a simulation-free framework for training latent SDEs with exact posterior process marginals.
- Also recommended: Helmholtz-SDE by Smith, Trippe, and Linderman, studying analytic approximations to the optimal gauge.
More from Research
- QUEEN paper distills AlphaZero-style chess search into language for LLMs — danqi_chen · 2026-10-07
- COLM paper asks whether VLMs can internalize tool calls in latent space instead of calling them — PMinervini · 2026-10-07
- CLeaR 2027 Opens Its Submission Portal — ArthurGretton · 2026-10-07
- MIT paper: a minimalist agent loop that passes history as code variables beats Letta and ACE at half the cost — rohanpaul_ai · 2026-10-07
- doodlestein releases explainer video alongside open-source companion repo — doodlestein · 2026-10-07
- A 3Blue1Brown-Style Explainer Rendered End-to-End in Rust by franken_manim — doodlestein · 2026-10-07