arxiv:2506.04755

Truth in the Few: High-Value Data Selection for Efficient Multi-Modal Reasoning

Published on Jun 5

· Submitted by

lss727 on Jun 9

#3 Paper of the day

Upvote

Authors:

Lei Wang ,

Abstract

A new data selection paradigm, Reasoning Activation Potential (RAP), enhances multi-modal reasoning in large language models using minimal high-value datasets, improving performance and reducing computational costs.

AI-generated summary

While multi-modal large language models (MLLMs) have made significant progress in complex reasoning tasks via reinforcement learning, it is commonly believed that extensive training data is necessary for improving multi-modal reasoning ability, inevitably leading to data redundancy and substantial computational costs. However, can smaller high-value datasets match or outperform full corpora for multi-modal reasoning in MLLMs? In this work, we challenge this assumption through a key observation: meaningful multi-modal reasoning is triggered by only a sparse subset of training samples, termed cognitive samples, whereas the majority contribute marginally. Building on this insight, we propose a novel data selection paradigm termed Reasoning Activation Potential (RAP), which identifies cognitive samples by estimating each sample's potential to stimulate genuine multi-modal reasoning by two complementary estimators: 1) Causal Discrepancy Estimator (CDE) based on the potential outcome model principle, eliminates samples that overly rely on language priors by comparing outputs between multi-modal and text-only inputs; 2) Attention Confidence Estimator (ACE), which exploits token-level self-attention to discard samples dominated by irrelevant but over-emphasized tokens in intermediate reasoning stages. Moreover, we introduce a Difficulty-aware Replacement Module (DRM) to substitute trivial instances with cognitively challenging ones, thereby ensuring complexity for robust multi-modal reasoning. Experiments on six datasets show that our RAP method consistently achieves superior performance using only 9.3% of the training data, while reducing computational costs by over 43%. Our code is available at https://github.com/Leo-ssl/RAP.

View arXiv page View PDF Add to collection

Community

lss727

Paper submitter 8 days ago

While multi-modal large language models (MLLMs) have made significant
progress in complex reasoning tasks via reinforcement learning, it is commonly
believed that extensive training data is necessary for improving multi-modal
reasoning ability, inevitably leading to data redundancy and substantial
computational costs. However, can smaller high-value datasets match or
outperform full corpora for multi-modal reasoning in MLLMs? In this work, we
challenge this assumption through a key observation: meaningful multi-modal
reasoning is triggered by only a sparse subset of training samples, termed
cognitive samples, whereas the majority contribute marginally. Building on this
insight, we propose a novel data selection paradigm termed Reasoning Activation
Potential (RAP), which identifies cognitive samples by estimating each sample's
potential to stimulate genuine multi-modal reasoning by two complementary
estimators: 1) Causal Discrepancy Estimator (CDE) based on the potential
outcome model principle, eliminates samples that overly rely on language priors
by comparing outputs between multi-modal and text-only inputs; 2) Attention
Confidence Estimator (ACE), which exploits token-level self-attention to
discard samples dominated by irrelevant but over-emphasized tokens in
intermediate reasoning stages. Moreover, we introduce a Difficulty-aware
Replacement Module (DRM) to substitute trivial instances with cognitively
challenging ones, thereby ensuring complexity for robust multi-modal reasoning.
Experiments on six datasets show that our RAP method consistently achieves
superior performance using only 9.3% of the training data, while reducing
computational costs by over 43%.