SRUM: Fine-Grained Self-Rewarding for Unified Multimodal Models

• URL: http://arxiv.org/abs/2510.12784v1
• Date: Tue, 14 Oct 2025 17:56:11 GMT
• Title: SRUM: Fine-Grained Self-Rewarding for Unified Multimodal Models
• Abstract summary: Unified Multimodal Models (UMMs) integrate vision-language generation and understanding capabilities within a single framework.<n>A model might correctly understand an image based on user instructions, yet be unable to generate a faithful image from text prompts.<n>We introduce SRUM, a self-rewarding post-training framework that can be directly applied to existing UMMs of various designs.
• Score: 44.79249920949795
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recently, remarkable progress has been made in Unified Multimodal Models (UMMs), which integrate vision-language generation and understanding capabilities within a single framework. However, a significant gap exists where a model's strong visual understanding often fails to transfer to its visual generation. A model might correctly understand an image based on user instructions, yet be unable to generate a faithful image from text prompts. This phenomenon directly raises a compelling question: Can a model achieve self-improvement by using its understanding module to reward its generation module? To bridge this gap and achieve self-improvement, we introduce SRUM, a self-rewarding post-training framework that can be directly applied to existing UMMs of various designs. SRUM creates a feedback loop where the model's own understanding module acts as an internal ``evaluator'', providing corrective signals to improve its generation module, without requiring additional human-labeled data. To ensure this feedback is comprehensive, we designed a global-local dual reward system. To tackle the inherent structural complexity of images, this system offers multi-scale guidance: a \textbf{global reward} ensures the correctness of the overall visual semantics and layout, while a \textbf{local reward} refines fine-grained, object-level fidelity. SRUM leads to powerful capabilities and shows strong generalization, boosting performance on T2I-CompBench from 82.18 to \textbf{88.37} and on T2I-ReasonBench from 43.82 to \textbf{46.75}. Overall, our work establishes a powerful new paradigm for enabling a UMMs' understanding module to guide and enhance its own generation via self-rewarding.

Related papers

• Ask, Solve, Generate: Self-Evolving Unified Multimodal Understanding and Generation via Self-Consistency Rewards [52.42920996842378]
  Most unified large multimodal models (LMMs) that support both visual understanding and image generation rely on curated post-training supervision.<n>We propose a self-evolving training framework with three internal roles: a Proposer that generates visual questions, a solver that answers and evaluates them, and a Generator that synthesizes images.<n>For image generation, we design a multi-scale internal evaluation scheme that combines question-answer fidelity scoring with cycle-consistent captioning.
  arXiv  Detail & Related papers  (2026-06-25T17:59:57Z)
• Free Lunch for Unified Multimodal Models: Enhancing Generation via Reflective Rectification with Inherent Understanding [20.397510070808238]
  Unified Multimodal Models (UMMs) aim to integrate visual understanding and generation within a single structure.<n>UMMs exhibit a notable capability mismatch, where their understanding capability significantly outperforms their generation.<n>We propose UniRect-CoT, a training-free unified rectification chain-of-thought framework.
  arXiv  Detail & Related papers  (2026-04-15T06:41:56Z)
• InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing [87.5130783098133]
  InternVL-U is a lightweight 4B- parameter UMM that democratizes multimodal understanding and reasoning capabilities.<n>It integrates a state-of-the-art Multimodal Large Language Model (MLLM) with a specialized MMDiT-based visual generation head.<n>It consistently outperforms unified baseline models with over 3x larger scales such as BAGEL (14B) on various generation and editing tasks.
  arXiv  Detail & Related papers  (2026-03-10T16:38:33Z)
• Learning to Generate via Understanding: Understanding-Driven Intrinsic Rewarding for Unified Multimodal Models [98.8608163448532]
  unified multimodal models (UMMs) have made remarkable progress in integrating visual understanding and generation.<n>We propose a token-level intrinsic text-image alignment reward mechanism, GvU, enabling the UMM to act simultaneously as teacher and student.<n>We show that our method substantially boosts UMMs' generation, which in turn strengthens their fine-grained visual understanding.
  arXiv  Detail & Related papers  (2026-03-06T08:56:14Z)
• UniG2U-Bench: Do Unified Models Advance Multimodal Understanding? [50.92401586025528]
  Unified multimodal models have recently demonstrated strong generative capabilities, yet whether and when generation improves understanding remains unclear.<n>We introduce UniG2U-Bench, a comprehensive benchmark categorizing generation-to-understanding (G2U) evaluation into 7 regimes and 30 subtasks.
  arXiv  Detail & Related papers  (2026-03-03T18:36:16Z)
• Endogenous Reprompting: Self-Evolving Cognitive Alignment for Unified Multimodal Models [23.128973540926552]
  Endogenous Reprompting transforms the model's understanding into an explicit generative reasoning step.<n>We show that SEER consistently outperforms state-of-the-art baselines in evaluation accuracy, reprompting efficiency, and generation quality.
  arXiv  Detail & Related papers  (2026-01-28T06:54:36Z)
• Forge-and-Quench: Enhancing Image Generation for Higher Fidelity in Unified Multimodal Models [23.529904770014735]
  This paper introduces a novel perspective: leveraging understanding to enhance the fidelity and detail richness of generated images.<n>We propose Forge-and-Quench, a new unified framework that puts this principle into practice.<n>Experiments show that Forge-and-Quench significantly improves image fidelity and detail across multiple models.
  arXiv  Detail & Related papers  (2026-01-08T08:18:44Z)
• Draw-In-Mind: Rebalancing Designer-Painter Roles in Unified Multimodal Models Benefits Image Editing [53.197392152109636]
  We introduce Draw-In-Mind (DIM), a dataset consisting of two complementary subsets: (i) DIM-T2I, containing 14M long-context image-text pairs to enhance complex instruction comprehension; and (ii) DIM-Edit, consisting of 233K chain-of-thought imaginations generated by GPT-4o, serving as explicit design blueprints for image edits.<n>DIM-4.6B-T2I/Edit achieves competitive performance on the ImgEdit and GEdit-Bench benchmarks, outperforming much larger models such as UniWorld-V1 and Step1X-Edit.
  arXiv  Detail & Related papers  (2025-09-02T06:06:52Z)
• SUDER: Self-Improving Unified Large Multimodal Models for Understanding and Generation with Dual Self-Rewards [55.99492656542475]
  We propose textbfSUDER (textbfSelf-improving textbfUnified LMMs with textbfDual stextbfElf-textbfRewards), a framework reinforcing the understanding and generation capabilities of LMMs.
  arXiv  Detail & Related papers  (2025-06-09T17:38:45Z)
• VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation [45.52926475981602]
  VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation.<n>VILA-U employs a single autoregressive next-token prediction framework for both tasks.
  arXiv  Detail & Related papers  (2024-09-06T17:49:56Z)
• USER: Unified Semantic Enhancement with Momentum Contrast for Image-Text Retrieval [115.28586222748478]
  Image-Text Retrieval (ITR) aims at searching for the target instances that are semantically relevant to the given query from the other modality. Existing approaches typically suffer from two major limitations.
  arXiv  Detail & Related papers  (2023-01-17T12:42:58Z)
• S2RMs: Spatially Structured Recurrent Modules [105.0377129434636]
  We take a step towards exploiting dynamic structure that are capable of simultaneously exploiting both modular andtemporal structures. We find our models to be robust to the number of available views and better capable of generalization to novel tasks without additional training.
  arXiv  Detail & Related papers  (2020-07-13T17:44:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.