Cover Image

Why Reinforcement Fine-Tuning Enables MLLMs Preserve Prior Knowledge Better: A Data Perspective

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhang, Zhihao, Dong, Qiaole, Zhang, Qi, Zhao, Jun, Zhou, Enyu, Xi, Zhiheng, Jin, Senjie, Fan, Xiaoran, Zhou, Yuhao, Wu, Mingqi, Fu, Yanwei, Ji, Tao, Gui, Tao, Huang, Xuanjing, Chen, Kai
Format:	Preprint
Published:	2025
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Reasoning or Memorization? Unreliable Results of Reinforcement Learning Due to Data Contamination
by: Wu, Mingqi, et al.
Published: (2025)

Self-Polish: Enhance Reasoning in Large Language Models via Problem Refinement
by: Xi, Zhiheng, et al.
Published: (2023)

EliteKV: Scalable KV Cache Compression via RoPE Frequency Selection and Joint Low-Rank Projection
by: Zhou, Yuhao, et al.
Published: (2025)

Distill Visual Chart Reasoning Ability from LLMs to MLLMs
by: He, Wei, et al.
Published: (2024)

Subspace Defense: Discarding Adversarial Perturbations by Learning a Subspace for Clean Signals
by: Zheng, Rui, et al.
Published: (2024)

Mitigating Object Hallucinations in MLLMs via Multi-Frequency Perturbations
by: Li, Shuo, et al.
Published: (2025)

LoRAMoE: Alleviate World Knowledge Forgetting in Large Language Models via MoE-Style Plugin
by: Dou, Shihan, et al.
Published: (2023)

JFTA-Bench: Evaluate LLM's Ability of Tracking and Analyzing Malfunctions Using Fault Trees
by: Wang, Yuhui, et al.
Published: (2026)

Parrot: A Training Pipeline Enhances Both Program CoT and Natural Language CoT for Reasoning
by: Jin, Senjie, et al.
Published: (2025)

AgentPRM: Process Reward Models for LLM Agents via Step-Wise Promise and Progress
by: Xi, Zhiheng, et al.
Published: (2025)

Steering LLMs via Scalable Interactive Oversight
by: Zhou, Enyu, et al.
Published: (2026)

StepCoder: Improve Code Generation with Reinforcement Learning from Compiler Feedback
by: Dou, Shihan, et al.
Published: (2024)

MemFlow: Optical Flow Estimation and Prediction with Memory
by: Dong, Qiaole, et al.
Published: (2024)

Online Dense Point Tracking with Streaming Memory
by: Dong, Qiaole, et al.
Published: (2025)

Entropy Polarity in Reinforcement Fine-Tuning: Direction, Asymmetry, and Control
by: Zhang, Jiazheng, et al.
Published: (2026)

The Role of Entropy in Visual Grounding: Analysis and Optimization
by: Li, Shuo, et al.
Published: (2025)

VRPO: Rethinking Value Modeling for Robust RL Training under Noisy Supervision
by: Zhu, Dingwei, et al.
Published: (2025)

Unveiling Linguistic Regions in Large Language Models
by: Zhang, Zhihao, et al.
Published: (2024)

RoCoIns: Enhancing Robustness of Large Language Models through Code-Style Instructions
by: Zhang, Yuansen, et al.
Published: (2024)

MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning
by: Lin, Jiahang, et al.
Published: (2026)

Analyzing the Effects of Supervised Fine-Tuning on Model Knowledge from Token and Parameter Levels
by: Ye, Junjie, et al.
Published: (2025)

Better Process Supervision with Bi-directional Rewarding Signals
by: Chen, Wenxiang, et al.
Published: (2025)

60 Data Points are Sufficient to Fine-Tune LLMs for Question-Answering
by: Ye, Junjie, et al.
Published: (2024)

LLaMA Beyond English: An Empirical Study on Language Capability Transfer
by: Zhao, Jun, et al.
Published: (2024)

Feedback-Driven Tool-Use Improvements in Large Language Models via Automated Build Environments
by: Ye, Junjie, et al.
Published: (2025)

Secrets of RLHF in Large Language Models Part II: Reward Modeling
by: Wang, Binghai, et al.
Published: (2024)

Beyond Scaling: Measuring and Predicting the Upper Bound of Knowledge Retention in Language Model Pre-Training
by: Jiang, Changhao, et al.
Published: (2025)

RMB: Comprehensively Benchmarking Reward Models in LLM Alignment
by: Zhou, Enyu, et al.
Published: (2024)

What Makes a Good Speech Tokenizer for LLM-Centric Speech Generation? A Systematic Study
by: Fan, Xiaoran, et al.
Published: (2025)

Towards Trustworthy Dermatology MLLMs: A Benchmark and Multimodal Evaluator for Diagnostic Narratives
by: Shen, Yuhao, et al.
Published: (2025)

Training Large Language Models for Reasoning through Reverse Curriculum Reinforcement Learning
by: Xi, Zhiheng, et al.
Published: (2024)

Have the VLMs Lost Confidence? A Study of Sycophancy in VLMs
by: Li, Shuo, et al.
Published: (2024)

MHA2MLA-VLM: Enabling DeepSeek's Economical Multi-Head Latent Attention across Vision-Language Models
by: Fan, Xiaoran, et al.
Published: (2026)

Measuring Data Diversity for Instruction Tuning: A Systematic Analysis and A Reliable Metric
by: Yang, Yuming, et al.
Published: (2025)

ChartE$^{3}$: A Comprehensive Benchmark for End-to-End Chart Editing
by: Li, Shuo, et al.
Published: (2026)

Unlocking the Essence of Beauty: Advanced Aesthetic Reasoning with Relative-Absolute Policy Optimization
by: Liu, Boyang, et al.
Published: (2025)

ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios
by: Ye, Junjie, et al.
Published: (2024)

MouSi: Poly-Visual-Expert Vision-Language Models
by: Fan, Xiaoran, et al.
Published: (2024)

PFDial: A Structured Dialogue Instruction Fine-tuning Method Based on UML Flowcharts
by: Zhang, Ming, et al.
Published: (2025)

Counteracting Matthew Effect in Self-Improvement of LVLMs through Head-Tail Re-balancing
by: Guo, Xin, et al.
Published: (2025)