Empowering Lightweight MLLMs with Reasoning via Long CoT SFT
On the map: LLM reasoning → Multimodal reasoning
Abstract (arXiv)
While Reinforcement Learning with Verifiable Rewards has enhanced the reasoning of large-scale language models (LLMs), its efficacy for lightweight multimodal language models (MLLMs) with fewer than seven billion parameters remains underexplored. This paper investigates the role of long Chain-of-Thought (long CoT) data in enhancing the reasoning abilities of such MLLMs. Our findings demonstrate that Supervised Fine-Tuning (SFT) with long CoT data significantly improves MLLM reasoning. Furthermore, we observe that after this initial SFT phase, MLLMs can achieve additional performance gains through a subsequent RL stage. We conclude that a SFT stage with long CoT data is a critical prerequisite for developing the reasoning capabilities of lightweight MLLMs.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 18 of its 19 references that are in the map. First posted on arXiv 3 Sep 2025.
Builds on
18 of 19 references.
Closest ancestors · 8
Same direction, newest first
- 2025Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning4 Jun
- 2025SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models10 Apr
- 2025VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model10 Apr
- 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization13 Mar
- 2025LMM-R1: Empowering 3B LMMs with Strong Reasoning Abilities Through Two-Stage Rule-Based RL10 Mar
- 2024DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models29 Oct
- 2024Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset22 Feb
- 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts3 Oct
Other references · 10
- 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
- 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,678
- 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,276
- 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,099
- 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 948
- 2024MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?cited by 573
- 2025TTRL: Test-Time Reinforcement Learningcited by 253
- 2025R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcement Learningcited by 99
- 2025Vision-R1: Evolving Human-Free Alignment in Large Vision-Language Models via Vision-Guided Reinforcement Learningcited by 68
- 2025Rethinking Reflection in Pre-Trainingcited by 43
Built on by
Citing papers per year: 2026: 2.
2 papers.
Similar, not cited
Close in wording, no citation either way
- 2025Demystifying Long Chain-of-Thought Reasoning in LLMs
- 2025Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models
- 2025The Synergy Dilemma of Long-CoT SFT and RL: Investigating Post-Training Techniques for Reasoning VLMs
- 2026The Art of Efficient Reasoning: Data, Reward, and Optimization
- 2025Reinforced MLLM: A Survey on RL-Based Reasoning in Multimodal Large Language Models
- 2025Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start
- 2025The Challenge of Teaching Reasoning to LLMs Without RL or Distillation
- 2025CARFT: Boosting LLM Reasoning via Contrastive Learning with Annotated Chain-of-Thought-based Reinforced Fine-Tuning