VIG: Visual Information Gain as a Reward Signal for Multimodal Chain-of-Thought Compression

• URL: http://arxiv.org/abs/2608.21883v2
• Date: Thu, 27 Aug 2026 14:17:29 GMT
• Title: VIG: Visual Information Gain as a Reward Signal for Multimodal Chain-of-Thought Compression
• Abstract summary: Visual Information Gain (VIG) is an information-theoretic reward that scores each reasoning token by how much the image reduces its predictive uncertainty.<n>VIG consistently improves the accuracy--efficiency trade-off across six main multimodal reasoning benchmarks and three Qwen3-VL-Thinking model sizes.
• Score: 6.871563715334457
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multimodal large reasoning models often rely on long Chain-of-Thought (CoT) traces in which a substantial fraction of tokens, such as repeated visual descriptions, self-reflection, and other visually-disengaged filler, inflate inference cost without contributing to the answer. Existing CoT compression methods optimize output length but never measure whether a reasoning token is actually grounded in the image. We propose \textbf{VIG} (Visual Information Gain), an information-theoretic GRPO reward that scores each reasoning token by how much the image reduces its predictive uncertainty. VIG is computed online from two forward passes of the same policy, one with and one without the image, so no reference chains, external annotations, or auxiliary reward models are needed. Across six main multimodal reasoning benchmarks and three Qwen3-VL-Thinking model sizes (2B/4B/8B), plus an additional R1-Onevision-Bench evaluation on 8B, VIG consistently improves the accuracy--efficiency trade-off, supporting our central claim: \emph{efficient multimodal reasoning emerges from raising visual information density, where every reasoning token earns its place by anchoring to the image, rather than from imposing a length budget.} Our source code is available at https://github.com/chaser682/vig.

Related papers

• Every Token Leaves a Ripple in the Stream of Thought: Eliciting Model-Internal Token Saliency for Chain-of-Thought Compression [13.74199057436712]
  Chain-of-thought (CoT) reasoning improves multi-step problem solving, but long reasoning traces inflate computation cost.<n>We propose textscMIST (Model-Internal Saliency for Token-level CoT compression), which defines token importance along two complementary axes.<n>textscMIST consistently outperforms baseline methods, suggesting that model-internal saliency provides an effective proxy for reasoning-token importance.
  arXiv  Detail & Related papers  (2026-08-31T16:41:14Z)
• Paying More Attention to Visual Tokens in Self-Evolving Large Multimodal Models [54.097381112138]
  Current self-evolving LMMs struggle on vision-language understanding tasks such as image captioning and visual question answering.<n>We propose VISE, a purely unsupervised self-evolving framework that directly regularizes the model's visual conditioning policy.<n>VISE operates within a single model without specialist roles, external reward models, or annotations.
  arXiv  Detail & Related papers  (2026-06-25T17:59:55Z)
• See Less, See Right: Bi-directional Perceptual Shaping For Multimodal Reasoning [58.7125460363147]
  We propose Bi-directional Perceptual Shaping (BiPS), which transforms question-conditioned masked views into bidirectional where-to-look signals.<n>BiPS boosts Qwen2.5-VL-7B by 8.2% on average and shows strong out-of-domain generalization to unseen datasets and image types.
  arXiv  Detail & Related papers  (2025-12-26T18:59:47Z)
• ChainV: Atomic Visual Hints Make Multimodal Reasoning Shorter and Better [59.29940512530982]
  We propose ChainV, a framework that dynamically integrates visual hints into the reasoning process.<n>Our method significantly improves reasoning accuracy and efficiency, especially on math-intensive benchmarks.
  arXiv  Detail & Related papers  (2025-11-21T10:11:17Z)
• VR-Thinker: Boosting Video Reward Models through Thinking-with-Image Reasoning [49.610569478718226]
  multimodal reward models (RMs) have substantially improved post-training for visual generative models.<n>VideoReward Thinker (VR-Thinker) is a thinking-with-image framework that equips the RM with visual reasoning operations and a visual memory window.<n>Our approach delivers state-of-the-art accuracy among open-source models on video preference benchmarks.
  arXiv  Detail & Related papers  (2025-10-12T09:29:50Z)
• Improving Chain-of-Thought Efficiency for Autoregressive Image Generation [55.57836819892392]
  We introduce ShortCoTI, a lightweight optimization framework for image generation.<n>ShortCoTI rewards more concise prompts with an adaptive function that scales according to an estimated difficulty for each task.<n>Our method eliminates verbose explanations and repetitive refinements, producing reasoning prompts that are both concise and semantically rich.
  arXiv  Detail & Related papers  (2025-10-07T05:40:43Z)
• Reinforcing Video Reasoning with Focused Thinking [65.85683941058916]
  We propose TW-GRPO, a novel framework that enhances visual reasoning with focused thinking and dense reward granularity.<n>Specifically, we employ a token weighting mechanism that prioritizes tokens with high informational density.<n>We also reformulate RL training by shifting from single-choice to multi-choice QA tasks.
  arXiv  Detail & Related papers  (2025-05-30T15:42:19Z)

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.