RL makes MLLMs see better than SFT

• URL: http://arxiv.org/abs/2510.16333v1
• Date: Sat, 18 Oct 2025 03:37:17 GMT
• Title: RL makes MLLMs see better than SFT
• Authors: Junha Song, Sangdoo Yun, Dongyoon Han, Jaegul Choo, Byeongho Heo,
• Abstract summary: We conduct a critical yet under-explored analysis of the vision encoder of Multimodal Language Model (MLLM)<n>Our results demonstrate that MLLM's post-training strategy (i.e., SFT or RL) not only leads to distinct outcomes on MLLM downstream tasks, but also fundamentally reshapes MLLM's underlying visual representations.<n>We then reframe our findings into a simple recipe for building strong vision encoders for MLLMs, Preference-Instructed Vision OpTimization (PIVOT)
• Score: 96.508432109136
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: A dominant assumption in Multimodal Language Model (MLLM) research is that its performance is largely inherited from the LLM backbone, given its immense parameter scale and remarkable capabilities. This has created a void in the understanding of the vision encoder, which determines how MLLMs perceive images. The recent shift in MLLM training paradigms, from Supervised Finetuning (SFT) to Reinforcement Learning (RL), magnifies this oversight-namely, the significant lack of analysis on how such training reshapes the vision encoder as well as the MLLM. To address this, we first investigate the impact of training strategies on MLLMs, where RL shows a clear advantage over SFT in strongly vision-related VQA benchmarks. Motivated by this, we conduct a critical yet under-explored analysis of the vision encoder of MLLMs through diverse and in-depth experiments, ranging from ImageNet classification and segmentation to gradient visualization. Our results demonstrate that MLLM's post-training strategy (i.e., SFT or RL) not only leads to distinct outcomes on MLLM downstream tasks, but also fundamentally reshapes MLLM's underlying visual representations. Specifically, the key finding of our study is that RL produces stronger and precisely localized visual representations compared to SFT, boosting the ability of the vision encoder for MLLM. We then reframe our findings into a simple recipe for building strong vision encoders for MLLMs, Preference-Instructed Vision OpTimization (PIVOT). When integrated into MLLMs, a PIVOT-trained vision encoder outperforms even larger and more heavily-trained counterparts, despite requiring less than 1% of the computational cost of standard vision pretraining. This result opens an effective and efficient path for advancing the vision backbones of MLLMs. Project page available at https://june-page.github.io/pivot/

Related papers

• A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models [85.30893355216486]
  We study how visual token redundancy evolves with different dMLLM architectures and tasks.<n>Our study reveals that visual redundancy emerges only in from-scratch dMLLMs while handling long-answer tasks.<n>Layer-skipping is promising for accelerating AR-to-diffusion dMLLMs, whereas progressive or late-step pruning is more effective for from-scratch dMLLMs.
  arXiv  Detail & Related papers  (2025-11-19T04:13:36Z)
• NaViL: Rethinking Scaling Properties of Native Multimodal Large Language Models under Data Constraints [100.02131897927484]
  This paper focuses on the native training of Multimodal Large Language Models (MLLMs) in an end-to-end manner.<n>We propose a native MLLM called NaViL, combined with a simple and cost-effective recipe.<n> Experimental results on 14 multimodal benchmarks confirm the competitive performance of NaViL against existing MLLMs.
  arXiv  Detail & Related papers  (2025-10-09T17:59:37Z)
• SparseMM: Head Sparsity Emerges from Visual Concept Responses in MLLMs [74.2538340966038]
  We investigate how Multimodal Large Language Models (MLLMs) process visual inputs by analyzing their attention mechanisms.<n>We reveal a surprising sparsity phenomenon: only a small subset of attention heads in LLMs actively contribute to visual understanding.<n>We introduce SparseMM, a KV-Cache optimization strategy that allocates asymmetric computation budgets to heads in LLMs based on their visual scores.
  arXiv  Detail & Related papers  (2025-06-05T17:59:55Z)
• Training-Free Reasoning and Reflection in MLLMs [45.134271969594614]
  This paper introduces FRANK Model, a training-FRee ANd r1-liKe MLLM that imbues off-the-shelf MLLMs with reasoning and reflection abilities.<n>Our key insight is to decouple perception and reasoning across MLLM decoder layers.<n>To this end, we propose a layer-wise, Taylor-derived closed-form fusion mechanism that integrates reasoning capacity into deep decoder layers.
  arXiv  Detail & Related papers  (2025-05-22T02:51:12Z)
• Elevating Visual Perception in Multimodal LLMs with Visual Embedding Distillation [109.5893580175657]
  In recent times, the standard practice for developing MLLMs is to feed features from vision encoder(s) into the LLM and train with natural language supervision.<n>This approach often causes models to lean towards language comprehension and undermine the rich visual perception signals present in the data.<n>We propose VisPer-LM, the first approach that infuses visual perception knowledge from expert vision encoders into the LLM's hidden representations.
  arXiv  Detail & Related papers  (2024-12-12T18:55:18Z)
• Dense Connector for MLLMs [89.50595155217108]
  We introduce the Dense Connector - a plug-and-play vision-language connector that significantly enhances existing MLLMs. Building on this, we also propose the Efficient Dense Connector, which achieves performance comparable to LLaVA-v1.5 with only 25% of the visual tokens. Our model, trained solely on images, showcases remarkable zero-shot capabilities in video understanding as well.
  arXiv  Detail & Related papers  (2024-05-22T16:25:03Z)
• Investigating the Catastrophic Forgetting in Multimodal Large Language Models [43.89009178021342]
  We introduce EMT: evaluating MulTimodality for evaluating the catastrophic forgetting in MLLMs. Almost all evaluated MLLMs fail to retain the same performance levels as their vision encoders on standard image classification tasks. As fine-tuning proceeds, the MLLMs begin to hallucinate, resulting in a significant loss of generalizability.
  arXiv  Detail & Related papers  (2023-09-19T04:51:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.