UniRL: Self-Improving Unified Multimodal Models via Supervised and Reinforcement Learning

• URL: http://arxiv.org/abs/2505.23380v1
• Date: Thu, 29 May 2025 12:00:15 GMT
• Title: UniRL: Self-Improving Unified Multimodal Models via Supervised and Reinforcement Learning
• Abstract summary: We introduce UniRL, a self-improving post-training approach for multimodal large language models.<n>Our approach enables the model to generate images from prompts and use them as training data in each iteration.<n>We evaluate UniRL on top of Show-o and Janus, achieving a GenEval score of 0.77 for Show-o and 0.65 for Janus.
• Score: 17.68867710994329
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Unified multimodal large language models such as Show-o and Janus have achieved strong performance across both generation and understanding tasks. However, these models typically rely on large-scale datasets and require substantial computation during the pretraining stage. In addition, several post-training methods have been proposed, but they often depend on external data or are limited to task-specific customization. In this work, we introduce UniRL, a self-improving post-training approach. Our approach enables the model to generate images from prompts and use them as training data in each iteration, without relying on any external image data. Moreover, it enables the two tasks to enhance each other: the generated images are used for understanding, and the understanding results are used to supervise generation. We explore supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO) to optimize the models. UniRL offers three key advantages: (1) it requires no external image data, as all training samples are generated by the model itself during training; (2) it not only improves individual task performance, but also reduces the imbalance between generation and understanding; and (3) it requires only several additional training steps during the post-training stage. We evaluate UniRL on top of Show-o and Janus, achieving a GenEval score of 0.77 for Show-o and 0.65 for Janus. Code and models will be released in https://github.com/showlab/UniRL.

Related papers

• Discover, Learn, and Reinforce: Scaling Vision-Language-Action Pretraining with Diverse RL-Generated Trajectories [33.872433985210876]
  Scaling vision-language-action (VLA) model pre-training requires large volumes of diverse, high-quality manipulation trajectories.<n>We propose Discover, Lea rn and Reinforce, which generates multiple distinct, high-success behavioral patterns for VLA pretraining.<n>When adapted to unseen downstream task suites, VLA models pretrained on our diverse RL data surpass counterparts trained on equal-sized standard RL datasets.
  arXiv  Detail & Related papers  (2025-11-24T07:54:49Z)
• Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models [63.032359320629105]
  We introduce: Unpaired Multimodal, a modality-agnostic training paradigm in which a single model alternately processes inputs from different modalities while sharing parameters across them.<n>We show that using unpaired data from auxiliary modalities consistently improves downstream performance across diverse unimodal targets such as image and audio.
  arXiv  Detail & Related papers  (2025-10-09T17:32:23Z)
• Intention-Conditioned Flow Occupancy Models [69.79049994662591]
  Large-scale pre-training has fundamentally changed how machine learning research is done today.<n>Applying this same framework to reinforcement learning is appealing because it offers compelling avenues for addressing core challenges in RL.<n>Recent advances in generative AI have provided new tools for modeling highly complex distributions.
  arXiv  Detail & Related papers  (2025-06-10T15:27:46Z)
• OpenUni: A Simple Baseline for Unified Multimodal Understanding and Generation [44.34131932968147]
  We present OpenUni, a lightweight, open-source baseline for unifying multimodal understanding and generation.<n>Inspired by prevailing practices in unified model learning, we adopt an efficient training strategy that minimizes the training complexity and overhead.<n>We demonstrate that OpenUni can: 1) generate high-quality and instruction-aligned images, and 2) achieve exceptional performance on standard benchmarks such as GenEval, DPG- Bench, and WISE, with only 1.1B and 3.1B activated parameters.
  arXiv  Detail & Related papers  (2025-05-29T17:09:44Z)
• Do We Need to Design Specific Diffusion Models for Different Tasks? Try ONE-PIC [77.8851460746251]
  We propose a simple, efficient, and general approach to fine-tune diffusion models.<n> ONE-PIC enhances the inherited generative ability in the pretrained diffusion models without introducing additional modules.<n>Our method is simple and efficient which streamlines the adaptation process and achieves excellent performance with lower costs.
  arXiv  Detail & Related papers  (2024-12-07T11:19:32Z)
• Leveraging Foundation Models for Multi-modal Federated Learning with Incomplete Modality [41.79433449873368]
  We propose a novel multi-modal federated learning method, Federated Multi-modal contrastiVe training with Pre-trained completion (FedMVP) FedMVP integrates the large-scale pre-trained models to enhance the federated training. We demonstrate that the model achieves superior performance over two real-world image-text classification datasets.
  arXiv  Detail & Related papers  (2024-06-16T19:18:06Z)
• Aligning Modalities in Vision Large Language Models via Preference Fine-tuning [67.62925151837675]
  In this work, we frame the hallucination problem as an alignment issue, tackle it with preference tuning. Specifically, we propose POVID to generate feedback data with AI models. We use ground-truth instructions as the preferred response and a two-stage approach to generate dispreferred data. In experiments across broad benchmarks, we show that we can not only reduce hallucinations, but improve model performance across standard benchmarks, outperforming prior approaches.
  arXiv  Detail & Related papers  (2024-02-18T00:56:16Z)
• eP-ALM: Efficient Perceptual Augmentation of Language Models [70.47962271121389]
  We propose to direct effort to efficient adaptations of existing models, and propose to augment Language Models with perception. Existing approaches for adapting pretrained models for vision-language tasks still rely on several key components that hinder their efficiency. We show that by freezing more than 99% of total parameters, training only one linear projection layer, and prepending only one trainable token, our approach (dubbed eP-ALM) significantly outperforms other baselines on VQA and Captioning.
  arXiv  Detail & Related papers  (2023-03-20T19:20:34Z)
• EfficientTrain: Exploring Generalized Curriculum Learning for Training Visual Backbones [80.662250618795]
  This paper presents a new curriculum learning approach for the efficient training of visual backbones (e.g., vision Transformers) As an off-the-shelf method, it reduces the wall-time training cost of a wide variety of popular models by >1.5x on ImageNet-1K/22K without sacrificing accuracy.
  arXiv  Detail & Related papers  (2022-11-17T17:38:55Z)
• ASIF: Coupled Data Turns Unimodal Models to Multimodal Without Training [29.240131406803794]
  We show that a common space can be created without any training at all, using single-domain encoders and a much smaller amount of image-text pairs. Our model has unique properties, most notably, deploying a new version with updated training samples can be done in a matter of seconds.
  arXiv  Detail & Related papers  (2022-10-04T16:56:22Z)
• Multimodal Knowledge Alignment with Reinforcement Learning [103.68816413817372]
  ESPER extends language-only zero-shot models to unseen multimodal tasks, like image and audio captioning. Our key novelty is to use reinforcement learning to align multimodal inputs to language model generations without direct supervision. Experiments demonstrate that ESPER outperforms baselines and prior work on a variety of zero-shot tasks.
  arXiv  Detail & Related papers  (2022-05-25T10:12:17Z)
• Meta Internal Learning [88.68276505511922]
  Internal learning for single-image generation is a framework, where a generator is trained to produce novel images based on a single image. We propose a meta-learning approach that enables training over a collection of images, in order to model the internal statistics of the sample image more effectively. Our results show that the models obtained are as suitable as single-image GANs for many common image applications.
  arXiv  Detail & Related papers  (2021-10-06T16:27:38Z)

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.