ReSFlow rectifies categorical flows on statistical manifolds for one-step generation
jiqizhixin · x · 2026-11-22
Researchers from Inner Mongolia University, Shanghai University of Finance and Economics and CMU introduce ReSFlow (arXiv preprint). Discrete generators often spend many integration steps following curved paths between distributions; ReSFlow rectifies categorical flows on the statistical manifold, respecting its geometry so learned routes approach geodesics — changing the transport problem itself so a single sampling step suffices. Experiments span image, text and biological generation, reporting improved one-step quality plus geometric analysis of trajectories.
More from Research
- Ultra-fast SAT solvers could make verification loops an ordinary part of inference — _onionesque · 2026-11-22
- NBER study of 500k+ GitHub devs: coding agents boosted commits ~240% but releases only ~30% — Antique_Atmosphere97 · 2026-11-22
- Ben Goertzel on the AI Riemann proof: 'a new era of math'—and math glory now belongs to those with tokens — bengoertzel · 2026-11-22
- LessWrong: looping boosts no-CoT reasoning far more than general skills in NVIDIA's Dory 4B — CFGeek · 2026-11-22
- "Hilbert's program vindicated": Lean brings us near a practical decision machine — teortaxesTex · 2026-11-22
- Simulation scores, not vibes: when can you claim you've explained a neuron? — burny_tech · 2026-11-22