AI Resonance · 阅读最新日报 · AI 入门推荐

2026-11-16 · America/Los_Angeles · 社区动态 · #2

[Paper & Models] MMFlash: Faster Speculative Decoding for Vision-Language Models

Speculative decoding accelerates generation by letting a lightweight drafter propose tokens for parallel verification by a target model. Diffusion drafters such as DFlash propose an entire token block in a single forward pass, but mainly focus on text-only tasks, leaving multimodal drafting underexplored. To study this setting, we first introduce a metric of token-level visual dependence based on the KL divergence between next-token distributions with and without visual input. We find that acceptance rates decrease from 46% in the lowest visual-dependence quartile to 22% in the highest, with…

[Paper & Models] MMFlash: Faster Speculative Decoding for Vision-Language Models

热度 45.6 / 100;排名与评分保留该期记录。