跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2602.02227 · v2

Show, Don't Tell: Morphing Latent Reasoning into Image Generation

首次提交 2026-02-02 23:29
版本更新 2026-11-19 12:39
雷达观测 2026-11-20 10:50
arXiv ↗PDF ↗

摘要

Text-to-image (T2I) generation has achieved remarkable progress, yet existing methods often lack the ability to dynamically reason and refine during generation--a hallmark of human creativity. Current reasoning-augmented paradigms most rely on explicit thought processes, where intermediate reasoning is decoded into discrete text at fixed steps with frequent image decoding and re-encoding, leading to inefficiencies, information loss, and cognitive mismatches. To bridge this gap, we introduce LatentMorph, a novel framework that seamlessly integrates implicit latent reasoning into the T2I generation process. At its core, LatentMorph introduces four lightweight components: (i) a condenser for summarizing intermediate generation states into compact visual memory, (ii) a translator for converting latent thoughts into actionable guidance, (iii) a shaper for dynamically steering next image token predictions, and (iv) an RL-trained invoker for adaptively determining when to invoke reasoning. By performing reasoning entirely in continuous latent spaces, LatentMorph avoids the bottlenecks of explicit reasoning and enables more adaptive self-refinement. Extensive experiments demonstrate that LatentMorph (I) enhances the base model Janus-Pro by 16%16\% on GenEval and 25%25\% on T2I-CompBench; (II) outperforms explicit paradigms (e.g., TwiG) by 15%15\% and 11%11\% on abstract reasoning tasks like WISE and IPV-Txt, (III) while reducing inference time by 44%44\% and token consumption by 51%51\%; and (IV) exhibits 71%71\% cognitive alignment with human intuition on reasoning invocation.

作者备注:ICML 2026; Code: https://github.com/EnVision-Research/LatentMorph; 1119-update: a smaller-sized file

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-19 12:39 · 最新观测版本

Show, Don't Tell: Morphing Latent Reasoning into Image Generation

Text-to-image (T2I) generation has achieved remarkable progress, yet existing methods often lack the ability to dynamically reason and refine during generation--a hallmark of human creativity. Current reasoning-augmented paradigms most rely on explicit thought processes, where intermediate reasoning is decoded into discrete text at fixed steps with frequent image decoding and re-encoding, leading to inefficiencies, information loss, and cognitive mismatches. To bridge this gap, we introduce LatentMorph, a novel framework that seamlessly integrates implicit latent reasoning into the T2I generation process. At its core, LatentMorph introduces four lightweight components: (i) a condenser for summarizing intermediate generation states into compact visual memory, (ii) a translator for converting latent thoughts into actionable guidance, (iii) a shaper for dynamically steering next image token predictions, and (iv) an RL-trained invoker for adaptively determining when to invoke reasoning. By performing reasoning entirely in continuous latent spaces, LatentMorph avoids the bottlenecks of explicit reasoning and enables more adaptive self-refinement. Extensive experiments demonstrate that LatentMorph (I) enhances the base model Janus-Pro by 16%16\% on GenEval and 25%25\% on T2I-CompBench; (II) outperforms explicit paradigms (e.g., TwiG) by 15%15\% and 11%11\% on abstract reasoning tasks like WISE and IPV-Txt, (III) while reducing inference time by 44%44\% and token consumption by 51%51\%; and (IV) exhibits 71%71\% cognitive alignment with human intuition on reasoning invocation.

查看此版本解读 →查看此版本原文 ↗