🛰️ Daily AI Frontier
‹ back to 2026-07-25

Expanding Flow Maps

Research Multimodal & Generative

Merged summary

TL;DR - Expanding Flow Maps extend flow-based generative models to outputs whose dimensionality or length can grow during generation. This enables efficient few-step generation of variable-size objects such as graphs and sequences.

  • Introduces expanding interpolants that add state dimensions using conditional noise.
  • Factors each map into a learned expand operator and a transport map for joint expansion and denoising.
  • Supports continuous spaces and discrete simplices, including variable-size graphs and variable-length sequences.
  • Recovers fixed-dimensional flow models when expansion is the identity operation.

Sources (1)

Expanding Flow Maps

arXiv cs.LG Sophia Tang, Pranam Chatterjee 2026-07-23 arXiv:2607.21585

TL;DR - Expanding Flow Maps extend flow-based generative models to outputs whose dimensionality or length can grow during generation. This enables efficient few-step generation of variable-size objects such as graphs and sequences.

  • Introduces expanding interpolants that add state dimensions using conditional noise.
  • Factors each map into a learned expand operator and a transport map for joint expansion and denoising.
  • Supports continuous spaces and discrete simplices, including variable-size graphs and variable-length sequences.
  • Recovers fixed-dimensional flow models when expansion is the identity operation.
item →