AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Co-Evolving Interpolants and Flows via Path-Flow Alignment

Zeyu Michael Li, William X Chen, Xiang Cheng

NeurIPS 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Flow matching and flow sampling

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes jointly training an endpoint-preserving path network and a flow network with a shared path-flow alignment loss for flow matching.

Problem

Flow matching typically fixes the interpolation path and learns only the velocity field, which may be suboptimal.

Approach

Co-evolve the interpolation path and the flow: the flow matches the path velocity while the path aligns its velocity to the current flow, using one alignment loss with an endpoint-preserving path network.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add path-flow alignment with stochastic regularization to continued fine-tuningSiT-S · ImageNet-256 (FM)50k-FID: 9.022 -> 8.359Improved
Add path-flow alignment with stochastic regularization to continued fine-tuningSiT-B · ImageNet-256 (FM)50k-FID: 5.057 -> 4.415Improved
Add path-flow alignment with stochastic regularization to continued fine-tuningSiT-L · ImageNet-256 (FM)50k-FID: 3.390 -> 2.742Improved
Add path-flow alignment with stochastic regularization to continued fine-tuningSiT-XL · ImageNet-256 (FM)50k-FID: 2.091 -> 1.825Improved
Add path-flow alignment to model-guidance fine-tuningSiT-B-MG · ImageNet-256 (MG)50k-FID: 7.940 -> 5.585Improved
Add path-flow alignment to model-guidance fine-tuningSiT-XL-MG · ImageNet-256 (MG)50k-FID: 1.494 -> 1.322Improved
Remove stochastic regularization (rho_t = 1) from path-flow alignment (ablation)SiT-S · ImageNet-256 (FM)50k-FIDHurt
Add stochastic source perturbation (rho_t < 1) to path-flow alignment (ablation)SiT-S · ImageNet-256 (FM)late-time flow divergence / FIDImproved
Increase kinetic energy weight beta (with rho_t = 1) (ablation)SiT-S · ImageNet-256 (FM)FID late degradationMixed
Add acceleration regularization to path objective (ablation)SiT-S · ImageNet-256 (FM)50k-FIDNo clear effect

Experimental setup

Builds on: flow matching

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy