Cover Image

RL makes MLLMs see better than SFT

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Song, Junha, Yun, Sangdoo, Han, Dongyoon, Choo, Jaegul, Heo, Byeongho
Format:	Preprint
Published:	2025
Subjects:	Computer Vision and Pattern Recognition Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Learning to See What You Need: Gaze Attention for Multimodal Large Language Models
by: Song, Junha, et al.
Published: (2026)

Rotary Position Embedding for Vision Transformer
by: Heo, Byeongho, et al.
Published: (2024)

Masking meets Supervision: A Strong Learning Alliance
by: Heo, Byeongho, et al.
Published: (2023)

Match me if you can: Semi-Supervised Semantic Correspondence Learning with Unpaired Images
by: Kim, Jiwon, et al.
Published: (2023)

DNNs May Determine Major Properties of Their Outputs Early, with Timing Possibly Driven by Bias
by: Park, Song, et al.
Published: (2025)

DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
by: Kim, Donghyun, et al.
Published: (2024)

MagiCapture: High-Resolution Multi-Concept Portrait Customization
by: Hyung, Junha, et al.
Published: (2023)

Token Bottleneck: One Token to Remember Dynamics
by: Kim, Taekyung, et al.
Published: (2025)

Model Stock: All we need is just a few fine-tuned models
by: Jang, Dong-Hwan, et al.
Published: (2024)

DaWin: Training-free Dynamic Weight Interpolation for Robust Adaptation
by: Oh, Changdae, et al.
Published: (2024)

Similarity of Neural Architectures using Adversarial Attack Transferability
by: Hwang, Jaehui, et al.
Published: (2022)

Adapting Pretrained ViTs with Convolution Injector for Visuo-Motor Control
by: Hwang, Dongyoon, et al.
Published: (2024)

Morphing Tokens Draw Strong Masked Image Models
by: Kim, Taekyung, et al.
Published: (2023)

Investigating Pre-Training Objectives for Generalization in Vision-Based Reinforcement Learning
by: Kim, Donghu, et al.
Published: (2024)

Temporal In-Context Fine-Tuning with Temporal Reasoning for Versatile Control of Video Diffusion Models
by: Kim, Kinam, et al.
Published: (2025)

Lipsum-FT: Robust Fine-Tuning of Zero-Shot Models Using Random Text Guidance
by: Nam, Giung, et al.
Published: (2024)

Learning with Unmasked Tokens Drives Stronger Vision Learners
by: Kim, Taekyung, et al.
Published: (2023)

SeiT++: Masked Token Modeling Improves Storage-efficient Training
by: Lee, Minhyun, et al.
Published: (2023)

Scratching Visual Transformer's Back with Uniform Attention
by: Hyeon-Woo, Nam, et al.
Published: (2022)

LongProLIP: A Probabilistic Vision-Language Model with Long Context Text
by: Chun, Sanghyuk, et al.
Published: (2025)

Sparse autoencoders reveal selective remapping of visual concepts during adaptation
by: Lim, Hyesu, et al.
Published: (2024)

Is user feedback always informative? Retrieval Latent Defending for Semi-Supervised Domain Adaptation without Source Data
by: Song, Junha, et al.
Published: (2024)

Probabilistic Language-Image Pre-Training
by: Chun, Sanghyuk, et al.
Published: (2024)

Towards Calibrated Robust Fine-Tuning of Vision-Language Models
by: Oh, Changdae, et al.
Published: (2023)

MaskRIS: Semantic Distortion-aware Data Augmentation for Referring Image Segmentation
by: Lee, Minhyun, et al.
Published: (2024)

Exploring Conditions for Diffusion models in Robotic Control
by: Shin, Heeseong, et al.
Published: (2025)

Vector Prism: Animating Vector Graphics by Stratifying Semantic Structure
by: Yun, Jooyeol, et al.
Published: (2025)

Scaling Up Personalized Image Aesthetic Assessment via Task Vector Customization
by: Yun, Jooyeol, et al.
Published: (2024)

SelfSwapper: Self-Supervised Face Swapping via Shape Agnostic Masked AutoEncoder
by: Lee, Jaeseong, et al.
Published: (2024)

Why Does RL Generalize Better Than SFT? A Data-Centric Perspective on VLM Post-Training
by: Lu, Aojun, et al.
Published: (2026)

VidGuard-R1: AI-Generated Video Detection and Explanation via Reasoning MLLMs and RL
by: Park, Kyoungjun, et al.
Published: (2025)

Metis-RISE: RL Incentivizes and SFT Enhances Multimodal Reasoning Model Learning
by: Qiu, Haibo, et al.
Published: (2025)

When Test-Time Adaptation Meets Self-Supervised Models
by: Han, Jisu, et al.
Published: (2025)

Infinite-Homography as Robust Conditioning for Camera-Controlled Video Generation
by: Kim, Min-Jung, et al.
Published: (2025)

HYPE: Hyperbolic Entailment Filtering for Underspecified Images and Texts
by: Kim, Wonjae, et al.
Published: (2024)

Spatiotemporal Skip Guidance for Enhanced Video Diffusion Sampling
by: Hyung, Junha, et al.
Published: (2024)

PromptRL: Prompt Matters in RL for Flow-Based Image Generation
by: Wang, Fu-Yun, et al.
Published: (2026)

SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
by: Chu, Tianzhe, et al.
Published: (2025)

Devil is in the Detail: Towards Injecting Fine Details of Image Prompt in Image Generation via Conflict-free Guidance and Stratified Attention
by: Jo, Kyungmin, et al.
Published: (2025)

EgoX: Egocentric Video Generation from a Single Exocentric Video
by: Kang, Taewoong, et al.
Published: (2025)