mlsanityData to 7 Oct 2026

Understanding vs. Generation: Navigating Optimization Dilemma in Multimodal Models

arXiv 2602.15772 ↗First posted 17 Feb 2026cs.CV cs.AIcited by 5

Not on the map. Why?

Abstract (arXiv)

Current research in multimodal models faces a key challenge where enhancing generative capabilities often comes at the expense of understanding, and vice versa. We analyzed this trade-off and identify the primary cause might be the potential conflict between generation and understanding, which creates a competitive dynamic within the model. To address this, we propose the Reason-Reflect-Refine (R3) framework. This innovative algorithm re-frames the single-step generation task into a multi-step process of "generate-understand-regenerate". By explicitly leveraging the model's understanding capability during generation, we successfully mitigate the optimization dilemma, achieved stronger generation results and improved understanding ability which are related to the generation process. This offers valuable insights for designing next-generation unified multimodal models. Code is available at https://github.com/sen-ye/R3.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Builds on 32 of its 39 references that are in the map. First posted on arXiv 17 Feb 2026.

Builds on

32 of 39 references.

References · 32

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  2. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  3. 2025Qwen-Image Technical Reportcited by 973
  4. 2023GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignmentcited by 946
  5. 2024Chameleon: Mixed-Modal Early-Fusion Foundation Modelscited by 904
  6. 2025Emerging Properties in Unified Multimodal Pretrainingcited by 855
  7. 2025Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scalingcited by 787
  8. 2024Emu3: Next-Token Prediction is All You Needcited by 761
  9. 2024Show-o: One Single Transformer to Unify Multimodal Understanding and Generationcited by 734
  10. 2024SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformerscited by 707
  11. 2023NExT-GPT: Any-to-Any Multimodal LLMcited by 610
  12. 2025Flow-GRPO: Training Flow Matching Models via Online RLcited by 605
20 more
  1. 2020Scaling Laws for Autoregressive Generative Modelingcited by 510
  2. 2024Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Modelcited by 468
  3. 2024Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generationcited by 447
  4. 2025Kimi-VL Technical Reportcited by 381
  5. 2023DreamLLM: Synergistic Multimodal Comprehension and Creationcited by 296
  6. 2024VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generationcited by 290
  7. 2025Show-o2: Improved Native Unified Multimodal Modelscited by 242
  8. 2022DALL-Eval: Probing the Reasoning Skills and Social Biases of Text-to-Image Generation Modelscited by 230
  9. 2025MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDEcited by 168
  10. 2025R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Modelcited by 163
  11. 2025T2I-R1: Reinforcing Image Generation with Collaborative Semantic-level and Token-level CoTcited by 162
  12. 2024Mixture-of-Transformers: A Sparse and Scalable Architecture for Multi-Modal Foundation Modelscited by 151
  13. 2025UniTok: A Unified Tokenizer for Visual Generation and Understandingcited by 148
  14. 2025Echo-4o: Harnessing the Power of GPT-4o Synthetic Images for Improved Image Generationcited by 95
  15. 2025TIIF-Bench: How Does Your T2I Model Follow Your Instructions?cited by 64
  16. 2025UniCTokens: Boosting Personalized Understanding and Generation via Unified Concept Tokenscited by 55
  17. 2025GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learningcited by 51
  18. 2025UniLiP: Adapting CLIP for Unified Multimodal Understanding, Generation and Editingcited by 43
  19. 2025ReasonGen-R1: CoT for Autoregressive Image generation models through SFT and RLcited by 30
  20. 2026GENIUS: Generative Fluid Intelligence Evaluation Suitecited by 11

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026UniGRPO: Unified Policy Optimization for Reasoning-Driven Visual Generationcited by 11
  2. 2026PEARL: Personalized Streaming Video Understanding Modelcited by 8
  3. 2026Dreaming in Flow: Generative Grounding Feedback for Self-Evolving Unified Multimodal Modelscited by 1
  4. 2026Uni-Synergy: Bridging Understanding and Generation for Personalized Reasoning via Co-operative Reinforcement Learningcited by 0
  5. 2026Noisy-Channel Minimum Bayes Risk Decodingcited by 0