Reasoning Portability: Guiding Continual Learning for MLLMs in the RLVR Era

2026-05-17 · Qiuhe Hong, Yuyang Liu, Shuo Yang, Tiantian Peng, Fei Zhu, Yonghong Tian arxiv

Vision-Language Models in Continual Learning (VLM-CL) aim to continuously adapt to new multimodal tasks while retaining prior knowledge. The emerging paradigm that couples Multimodal Large Language Models (MLLMs) with Reinforcement Learning with Verifiable Rewards (RLVR) calls for a new pattern to guide continual adaptation. Advances in reasoning capability now make it feasible to impose constraints at the reasoning level. We formalize portability, a sample-level measure of how reusable the previous policy's behavior is on a new task, and empirically show that reasoning-level signals remain reliable on out-of-distribution samples while answer-level signals do not. We instantiate this as Reasoning Portability (RP) and propose Reasoning-based Dynamic Balance Continual Learning (RDB-CL), which modulates the per-sample Kullback-Leibler regularization in RLVR according to RP: a tight anchor preserves reusable reasoning on high-RP samples, while a relaxed anchor on low-RP samples permits exploration of new reasoning pathways. Experiments show that RDB-CL consistently outperforms baselines, improving Last accuracy by +12.0% over the vanilla RLVR baseline.

📄 PDF Abstract BibTeX arXiv:2605.18903

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning Continual Learning

Continual Reasoning Gym: Diagnosing and Harnessing Shared Reasoning in Continual RLVR

2026-08-19 · Lirui Luo, Guoxi Zhang, Hongming Xu, Rongqing Li 외 arxiv

Reinforcement learning with verifiable rewards (RLVR) commonly post-trains reasoning models on multiple tasks, while rerunning multitask RLVR (MTRL) as new tasks are added makes capability expansion costly. We therefore …

Reinforcement Learning Visual Reasoning

SpaceR: Reinforcing MLLMs in Video Spatial Reasoning

2025-04-02 · Kun Ouyang, Yuanxin Liu, HaoNing Wu, Yi Liu 외

Video spatial reasoning, which involves inferring the underlying spatial structure from observed video frames, poses a significant challenge for existing Multimodal Large Language Models (MLLMs). This limitation stems pr…

MME Spatial Reasoning Video MME Video Understanding

DyCo-RL: Dynamic Cross-Modal Coordination for Visual Reasoning

2026-06-06 · Hangui Lin, Yan Shu, Zhengyang Liang, Chi Liu 외 arxiv

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a leading paradigm for enhancing visual reasoning in Multimodal Large Language Models (MLLMs). However, existing RLVR methods optimize primarily for th…

Reinforcement Learning Mathematical Reasoning Visual Reasoning

M2-Reasoning: Empowering MLLMs with Unified General and Spatial Reasoning

2025-07-11 · Inclusion AI, :, Fudong Wang, Jiajia Liu 외

Recent advancements in Multimodal Large Language Models (MLLMs), particularly through Reinforcement Learning with Verifiable Rewards (RLVR), have significantly enhanced their reasoning abilities. However, a critical gap …

Spatial Reasoning

Rethinking Facial Expression Recognition in the Era of Multimodal Large Language Models: Benchmark, Datasets, and Beyond

2025-11-01 · Fan Zhang, Haoxuan Li, Shengju Qian, Xin Wang 외 arxiv

Multimodal Large Language Models (MLLMs) have revolutionized numerous research fields, including computer vision and affective computing. As a pivotal challenge in this interdisciplinary domain, facial expression recogni…

Facial Expression Recognition Reinforcement Learning