Deepseek RL (GRPO)-Inspired Research for Vision & Multimodal Reasoning

This repository maintains a curated and updated collection of research papers and repositories focused on Reinforcement Learning (RL)-inspired methods such as Group Relative Policy Optimization (GRPO) applied to Vision and Multimodal Reasoning tasks.

Papers

Date	Paper	Repository
July 2, 2025	Kwai Keye-VL Technical Report	Github
July 1, 2025	GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning	Github
June 28, 2025	Listener-Rewarded Thinking in VLMs for Image Preferences	-
June 27, 2025	MiCo: Multi-image Contrast for Reinforcement Visual Reasoning	-
June 26, 2025	ShotBench: Expert-Level Cinematic Understanding in Vision-Language Models	Github
June 25, 2025	MMSearch-R1: Incentivizing LMMs to Search	Github
June 20, 2025	Aha Moment Revisited: Are VLMs Truly Capable of Self Verification in Inference-time Scaling?	-
June 20, 2025	Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens	Github
June 20, 2025	Enhancing Step-by-Step and Verifiable Medical Reasoning in MLLMs	Github
June 19, 2025	GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning	Github
June 16, 2025	Ego-R1: Chain-of-Tool-Thought for Ultra-Long Egocentric Video Reasoning	Github
June 11, 2025	ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMs	Github
June 10, 2025	VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning	Github
June 9, 2025	DeepVideo-R1: Video Reinforcement Fine-Tuning via Difficulty-aware Regressive GRPO	Github
June 4, 2025	Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning	Github
June 4, 2025	RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics	Github
June 4, 2025	MiMo-VL Technical Report	Github
June 2, 2025	SynthRL: Scaling Visual Reasoning with Verifiable Data Synthesis	Github
June 2, 2025	SRPO: Enhancing Multimodal LLM Reasoning via Reflection-Aware Reinforcement Learning	Github
May 29, 2025	Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence	Github
May 28, 2025	Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO	Github
May 28, 2025	Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start	Github
May 27, 2025	MME-Reasoning: A Comprehensive Benchmark for Logical Reasoning in MLLMs	Github
May 26, 2025	Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning	Github
May 24, 2025	Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models	Github
May 22, 2025	GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning	Github
May 22, 2025	VLM-R^3: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought	-
May 22, 2025	SophiaVL-R1: Reinforcing MLLMs Reasoning with Thinking Reward	Github
May 20, 2025	Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning	Github
May 20, 2025	UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning	Github
May 20, 2025	VisualQuality-R1: Reasoning-Induced Image Quality Assessment via Reinforcement Learning to Rank	Github
May 20, 2025	Visual Agentic Reinforcement Fine-Tuning	Github
May 17, 2025	VisionReasoner: Unified Visual Perception and Reasoning via Reinforcement Learning	Github
May 16, 2025	GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoning	-
May 12, 2025	DanceGRPO: Unleashing GRPO on Visual Generation	Github
May 11, 2025	Seed1.5-VL Technical Report	Github
May 6, 2025	X-Reasoner: Towards Generalizable Reasoning Across Modalities and Domains	Github
May 6, 2025	Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning	Github
May 5, 2025	R1-Reward: Training Multimodal Reward Model Through Stable Reinforcement Learning	Github
Apr 23, 2025	Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning	Github
Apr 21, 2025	Eagle 2.5: Boosting Long-Context Post-Training for Frontier Vision-Language Models	Github
Apr 21, 2025	VisuLogic: A Benchmark for Evaluating Visual Reasoning in Multi-modal Large Language Models	Github
Apr 14, 2025	InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models	Github
Apr 13, 2025	TinyLLaVA-Video-R1: Towards Smaller LMMs for Video Reasoning	Github
Apr 10, 2025	VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning	Github
Apr 10, 2025	SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models	-
Apr 10, 2025	VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model	Github
Apr 10, 2025	VCR-Bench: A Comprehensive Evaluation Framework for Video Chain-of-Thought Reasoning	Github
Apr 9, 2025	VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning	Github
Apr 8, 2025	Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought	Github
Apr 3, 2025	Rethinking RL Scaling for Vision Language Models: A Transparent, From-Scratch Framework and Comprehensive Evaluation Scheme	Github
Apr 1, 2025	Improved Visual-Spatial Reasoning via R1-Zero-Like Training	Github
Apr 1, 2025	Think or Not Think: A Study of Explicit Thinking inRule-Based Visual Reinforcement Fine-Tuning	Github
Mar 31, 2025	Exploring the Effect of Reinforcement Learning on Video Understanding: Insights from SEED-Bench-R1	Github
Mar 28, 2025	OThink-MR1: Stimulating multimodal generalized reasoning capabilities via dynamic reinforcement learning	-
Mar 27, 2025	Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning	Github
Mar 27, 2025	Video-R1: Reinforcing Video Reasoning in MLLMs	Github
Mar 23, 2025	Vision-R1: Evolving Human-Free Alignment in Large Vision-Language Models via Vision-Guided Reinforcement Learning	Github
Mar 21, 2025	OpenVLThinker: An Early Exploration to Complex Vision-Language Reasoning via Iterative Self-Improvement	Github
Mar 18, 2025	Cosmos-Reason1: From Physical Common Sense To Embodied Reasoning	Github
Mar 18, 2025	DeepPerception: Advancing R1-like Cognitive Visual Perception in MLLMs for Knowledge-Intensive Visual Grounding	Github
Mar 17, 2025	R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimization	GitHub
Mar 13, 2025	R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization	GitHub
Mar 10, 2025	LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RL	GitHub
Mar 10, 2025	MM-Eureka: Exploring Visual Aha Moment with Rule-based Large-scale Reinforcement Learning	GitHub
Mar 10, 2025	Boosting the Generalization and Reasoning of Vision Language Models with Curriculum Reinforcement Learning	GitHub
Mar 9, 2025	Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models	GitHub
Mar 7, 2025	R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model	GitHub
Mar 4, 2025	Visual-RFT: Visual Reinforcement Fine-Tuning	GitHub
Feb 27, 2025	MedVLM-R1: Incentivizing Medical Reasoning Capability of Vision-Language Models (VLMs) via Reinforcement Learning	-
Feb 27, 2025	Med-RLVR: Emerging Medical Reasoning from a 3B base model via Reinforcement Learning	-
Feb 20, 2025	AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPO	GitHub

Repositories Only

• R1-V: Reinforcing Super Generalization Ability in Vision Language Models with Less Than $3
• Multimodal Open R1

Surveys

• Thinking with Images for Multimodal Reasoning: Foundations, Methods, and Future Frontiers
• Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Models
• Unified Multimodal Understanding and Generation Models: Advances, Challenges, and Opportunities
• 100 Days After DeepSeek-R1: A Survey on Replication Studies and More Directions for Reasoning Language Models
• Why Reasoning Matters? A Survey of Advancements in Multimodal Reasoning (v1)
• A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond | Github
• Aligning Multimodal LLM with Human Preference: A Survey | Github