Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start

• URL: http://arxiv.org/abs/2505.22334v2
• Date: Wed, 23 Jul 2025 07:37:08 GMT
• Title: Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start
• Authors: Lai Wei, Yuting Li, Kaipeng Zheng, Chen Wang, Yue Wang, Linghe Kong, Lichao Sun, Weiran Huang,
• Abstract summary: "aha moment" patterns are often attributed to emergent properties from reinforcement learning (RL)<n>We present a comprehensive study on enhancing multimodal reasoning through a two-stage approach.<n>Our experiments show that this combined approach consistently outperforms both SFT-only and RL-only methods.
• Score: 24.244577648817188
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent advancements in large language models (LLMs) have demonstrated impressive chain-of-thought reasoning capabilities, with reinforcement learning (RL) playing a crucial role in this progress. While "aha moment" patterns--where models exhibit self-correction through reflection--are often attributed to emergent properties from RL, we first demonstrate that these patterns exist in multimodal LLMs (MLLMs) prior to RL training but may not necessarily correlate with improved reasoning performance. Building on these insights, we present a comprehensive study on enhancing multimodal reasoning through a two-stage approach: (1) supervised fine-tuning (SFT) as a cold start with structured chain-of-thought reasoning patterns, followed by (2) reinforcement learning via GRPO to further refine these capabilities. Our extensive experiments show that this combined approach consistently outperforms both SFT-only and RL-only methods across challenging multimodal reasoning benchmarks. The resulting models achieve state-of-the-art performance among open-source MLLMs at both 3B and 7B scales, with our 7B model showing substantial improvements over base models (e.g., 66.3 %$\rightarrow$73.4 % on MathVista, 62.9 %$\rightarrow$70.4 % on We-Math) and our 3B model achieving performance competitive with several 7B models. Overall, this work provides practical guidance for building advanced multimodal reasoning models. Our code is available at https://github.com/waltonfuture/RL-with-Cold-Start.

Related papers

• Skywork-R1V3 Technical Report [14.952041273882639]
  We introduce Skywork-R1V3, an advanced, open-source vision-language model (VLM)<n>Key innovation lies in effectively transferring reasoning skills from text-only Large Language Models to visual tasks.<n>We introduce a unique indicator of reasoning capability, the entropy of critical reasoning tokens, which has proven highly effective for checkpoint selection.
  arXiv  Detail & Related papers  (2025-07-08T16:47:16Z)
• OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling [29.818409458662344]
  Different base language model families, such as Llama and Qwen, exhibit divergent behaviors during post-training with reinforcement learning (RL)<n>Our study reveals that high-quality mathematical corpora, such as MegaMath-Web-Pro, significantly improve both base model and RL performance.<n>We introduce a two-stage mid-training strategy, Stable-then-Decay, in which base models are first trained on 200B tokens with a constant learning rate, followed by 20B tokens across three CoT-focused branches with learning rate decay.
  arXiv  Detail & Related papers  (2025-06-25T14:58:13Z)
• Ring-lite: Scalable Reasoning via C3PO-Stabilized Reinforcement Learning for LLMs [51.21041884010009]
  Ring-lite is a Mixture-of-Experts (MoE)-based large language model optimized via reinforcement learning (RL)<n>Our approach matches the performance of state-of-the-art (SOTA) small-scale reasoning models on challenging benchmarks.
  arXiv  Detail & Related papers  (2025-06-17T17:12:34Z)
• Metis-RISE: RL Incentivizes and SFT Enhances Multimodal Reasoning Model Learning [20.515599491717442]
  We introduce textbfMetis-RISE (textbfRL textbfSFT textbfEnhances) for multimodal reasoning model learning.
  arXiv  Detail & Related papers  (2025-06-16T02:56:13Z)
• Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning [28.92744927199283]
  ReVisual-R1 achieves a new state-of-the-art among open-source 7B MLLMs on challenging benchmarks including MathVerse, MathVision, WeMath, LogicVista, DynaMath, and challenging AIME2024 and AIME2025.
  arXiv  Detail & Related papers  (2025-06-04T17:51:08Z)
• SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models [39.551767637896404]
  This work revisits the dominant supervised fine-tuning (SFT) then reinforcement learning (RL) paradigm for training Large Vision-Language Models (LVLMs)<n>We show that SFT can significantly undermine subsequent RL by inducing pseudo reasoning paths'' imitated from expert models.<n>We introduce VLAA-Thinking, a new multimodal dataset designed to support reasoning in LVLMs.
  arXiv  Detail & Related papers  (2025-04-10T16:54:05Z)
• Do Theory of Mind Benchmarks Need Explicit Human-like Reasoning in Language Models? [14.29992535286614]
  Theory of Mind (ToM) is the ability to attribute mental states to others.<n>Recent advancements in Large Language Models have shown promising performance on ToM benchmarks.<n>Do these benchmarks necessitate explicit human-like reasoning processes, or can models succeed through alternative strategies?
  arXiv  Detail & Related papers  (2025-04-02T12:58:42Z)
• Exploring the Effect of Reinforcement Learning on Video Understanding: Insights from SEED-Bench-R1 [53.894789613838654]
  We introduce SEED-Bench-R1, a benchmark designed to evaluate post-training methods for MLLMs in video understanding.<n>It includes intricate real-world videos and complex everyday planning tasks in the format of multiple-choice questions.<n>Using Qwen2-VL-Instruct-7B as a base model, we compare RL with supervised fine-tuning (SFT)<n>Our detailed analysis reveals that RL enhances visual perception but often produces less coherent reasoning chains.
  arXiv  Detail & Related papers  (2025-03-31T17:55:23Z)
• OpenVLThinker: An Early Exploration to Complex Vision-Language Reasoning via Iterative Self-Improvement [91.88062410741833]
  This study investigates whether similar reasoning capabilities can be successfully integrated into large vision-language models (LVLMs)<n>We consider an approach that iteratively leverages supervised fine-tuning (SFT) on lightweight training data and Reinforcement Learning (RL) to further improve model generalization.<n>OpenVLThinker, a LVLM exhibiting consistently improved reasoning performance on challenging benchmarks such as MathVista, MathVerse, and MathVision, demonstrates the potential of our strategy for robust vision-language reasoning.
  arXiv  Detail & Related papers  (2025-03-21T17:52:43Z)
• VisualPRM: An Effective Process Reward Model for Multimodal Reasoning [76.35753243272521]
  We introduce VisualPRM, which improves the reasoning abilities of existing Multimodal Large Language Models (MLLMs)<n>Our model achieves a 5.9-point improvement across seven multimodal reasoning benchmarks.<n>For the evaluation of multimodal PRMs, we propose VisualProcessBench, a benchmark with human-annotated step-wise correctness labels.
  arXiv  Detail & Related papers  (2025-03-13T12:03:37Z)
• MM-Eureka: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning [55.82649731348012]
  We introduce the MMK12 dataset and MM-EUREKA with 7B and 32B parameters.<n>The former is a high-quality multimodal mathematics reasoning dataset featuring diverse knowledge domains with human-verified answers and solution processes.<n>The latter is a multimodal model employing rule-based reinforcement learning utilizing online filtering and two-stage training strategy to enhance training stability.
  arXiv  Detail & Related papers  (2025-03-10T14:23:12Z)
• LLaVE: Large Language and Vision Embedding Models with Hardness-Weighted Contrastive Learning [76.82159851648711]
  We propose a framework that dynamically improves the embedding model's representation learning for negative pairs.<n>LLaVE establishes stronger baselines that achieve state-of-the-art (SOTA) performance.<n>LLaVE can generalize to text-video retrieval tasks in a zero-shot manner and achieve strong performance.
  arXiv  Detail & Related papers  (2025-03-04T10:21:57Z)
• Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimization [65.64108848398696]
  We introduce a preference optimization (PO) process to enhance the multimodal reasoning capabilities of MLLMs.<n>Specifically, we design an automated preference data construction pipeline to create MMPR, a high-quality, large-scale multimodal reasoning preference dataset.<n>We explore integrating PO with MLLMs, developing a simple yet effective method, termed Mixed Preference Optimization (MPO), which boosts multimodal CoT performance.
  arXiv  Detail & Related papers  (2024-11-15T18:59:27Z)
• Making Long-Context Language Models Better Multi-Hop Reasoners [42.09676404515287]
  We introduce Reasoning with Attributions, a novel approach that prompts LMs to supply attributions for each assertion during their reasoning. We validate our approach through experiments on three multi-hop datasets, employing both proprietary and open-source models. Our model achieves competitive performance on multi-hop reasoning benchmarks, closely paralleling proprietary LMs such as ChatGPT and Claude-instant.
  arXiv  Detail & Related papers  (2024-08-06T15:06:40Z)

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.