Latent Reasoning in Continuous Space for Unified Multimodal Models
cs.CV, cs.CL, cs.LG
Submitted: 2026-10-30
Updated: 2026-10-30
Project page: https://rootyjeon.github.io/larc-umm
Terminology
Sources
- Hierarchical Text-Conditional Image Generation with CLIP Latents
- Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation
- Training-Free Structured Diffusion Guidance for Compositional Text-to-Image Synthesis
- World Models
- Dream to Control: Learning Behaviors by Latent Imagination
- Show-o2: Improved Native Unified Multimodal Models
- Uni-cot: Towards Unified Chain-of-Thought Reasoning Across Text and Vision
- Emu3: Next-Token Prediction is All You Need
- Emerging Properties in Unified Multimodal Pretraining
- Training Large Language Models to Reason in a Continuous Latent Space
- FLUX-Reason-6M & PRISM-Bench: A Million-Scale Text-to-Image Reasoning Dataset and Comprehensive Benchmark
- Extending Test-Time Scaling: A 3D Perspective with Context, Batch, and Turn
- ScaleEdit-12M: Scaling Open-Source Image Editing Data Generation via Multi-Agent Framework
- ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment
- OneIG-Bench: Omni-dimensional Nuanced Evaluation for Image Generation
- CompBench: Benchmarking Complex Instruction-guided Image Editing
- When Visualizing is the First Step to Reasoning: MIRA, a Benchmark for Visual Chain-of-Thought
- Vision-aligned Latent Reasoning for Multi-modal Large Language Model
- GPT-4o System Card
- FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs