惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

美团技术团队
Microsoft Azure Blog
Microsoft Azure Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
B
Blog
Y
Y Combinator Blog
博客园_首页
有赞技术团队
有赞技术团队
博客园 - Franky
腾讯CDC
G
Google Developers Blog
Recent Announcements
Recent Announcements
博客园 - 【当耐特】
D
Docker
The GitHub Blog
The GitHub Blog
MyScale Blog
MyScale Blog
H
Help Net Security
Apple Machine Learning Research
Apple Machine Learning Research
A
About on SuperTechFans
D
DataBreaches.Net
T
The Blog of Author Tim Ferriss
V
V2EX
U
Unit 42
aimingoo的专栏
aimingoo的专栏
WordPress大学
WordPress大学

cs.CV updates on arXiv.org

ClawGUI: A Unified Framework for Training, Evaluating, and Deploying GUI Agents On the Robustness of Watermarking for Autoregressive Image Generation Revisiting Compositionality in Dual-Encoder Vision-Language Models: The Role of Inference Anthropogenic Regional Adaptation in Multimodal Vision-Language Model From Redaction to Restoration: Deep Learning for Medical Image Anonymization and Reconstruction The Salami Slicing Threat: Exploiting Cumulative Risks in LLM Systems BoxTuning: Directly Injecting the Object Box for Multimodal Model Fine-Tuning Semantic-Geometric Dual Compression: Training-Free Visual Token Reduction for Ultra-High-Resolution Remote Sensing Understanding Lightweight Low-Light Image Enhancement via Distribution-Normalizing Preprocessing and Depthwise U-Net Back to the Barn with LLAMAs: Evolving Pretrained LLM Backbones in Finetuning Vision Language Models Pseudo-Unification: Entropy Probing Reveals Divergent Information Patterns in Unified Multimodal Models QShield: Securing Neural Networks Against Adversarial Attacks using Quantum Circuits Evaluating the Impact of Medical Image Reconstruction on Downstream AI Fairness and Performance Retinal Cyst Detection from Optical Coherence Tomography Images LoViF 2026 The First Challenge on Weather Removal in Videos STORM: End-to-End Referring Multi-Object Tracking in Videos Data-Efficient Surgical Phase Segmentation in Small-Incision Cataract Surgery: A Controlled Study of Vision Foundation Models Rethinking the Diffusion Model from a Langevin Perspective Zero-shot World Models Are Developmentally Efficient Learners Edu-MMBias: A Three-Tier Multimodal Benchmark for Auditing Social Bias in Vision-Language Models under Educational Contexts VGA-Bench: A Unified Benchmark and Multi-Model Framework for Video Aesthetics and Generation Quality Evaluation Degradation-Consistent Paired Training for Robust AI-Generated Image Detection FREE-Switch: Frequency-based Dynamic LoRA Switch for Style Transfer Demographic and Linguistic Bias Evaluation in Omnimodal Language Models FlowPalm: Optical Flow Driven Non-Rigid Deformation for Geometrically Diverse Palmprint Generation Cross-Cultural Value Awareness in Large Vision-Language Models I Walk the Line: Examining the Role of Gestalt Continuity in Object Binding for Vision Transformers GLEaN: A Text-to-image Bias Detection Approach for Public Comprehension From UAV Imagery to Agronomic Reasoning: A Multimodal LLM Benchmark for Plant Phenotyping Not Your Stereo-Typical Estimator: Combining Vision and Language for Volume Perception
Skywork R1V2: Multimodal Hybrid Reinforcement Learning fo...
Peiyu Wang, Yichen Wei, Yi Peng, Xiaokun Wang, Weijie Qiu, Wei S · 2025-04-23 · via cs.CV updates on arXiv.org

We present Skywork R1V2, a next-generation multimodal reasoning model and a major leap forward from its predecessor, Skywork R1V. At its core, R1V2 introduces a hybrid reinforcement learning paradigm that jointly leverages the Mixed Preference Optimization (MPO) and the Group Relative Policy Optimization (GRPO), which harmonizes reward-model guidance with rule-based strategies, thereby addressing the long-standing challenge of balancing sophisticated reasoning capabilities with broad generalization. To further enhance training efficiency, we propose the Selective Sample Buffer (SSB) mechanism, which effectively addresses the vanishing advantages dilemma inherent in GRPO by prioritizing high-value samples throughout the optimization process. Notably, we observe that excessive reinforcement signals can induce visual hallucinations--a phenomenon we systematically monitor and mitigate through calibrated reward thresholds throughout the training process. Empirical results affirm the exceptional capability of R1V2, with benchmark-leading performances such as 62.6 on OlympiadBench, 78.9 on AIME2024, 63.6 on LiveCodeBench, and 73.6 on MMMU. These results underscore R1V2's superiority over existing open-source models and demonstrate significant progress in closing the performance gap with premier proprietary systems, including Gemini 2.5 and OpenAI-o4-mini. The Skywork R1V2 model weights have been publicly released to promote openness and reproducibility https://huggingface.co/Skywork/Skywork-R1V2-38B.