mlsanityData to 7 Oct 2026

TPCD: Tone-Pressure Contrastive Decoding and the Label-Free Gating Bottleneck in Vision-Language Models

arXiv 2607.26536 ↗First posted 29 Jul 2026cs.CVcited by 0

On the map: Multimodal LLMs / vision-language models → Hallucination mitigation in LVLMs

Abstract (arXiv)

High-pressure prompts can push vision-language models (VLMs) into unsupported commitments, such as reading illegible text, reporting indeterminate times, or affirming absent objects. This paper asks whether the pressure-induced distribution itself can serve as a contrastive-decoding negative branch. Tone-pressure contrastive decoding (TPCD) subtracts logits produced under a high-pressure instruction from logits produced under a safe neutral instruction. On the 800-example tone-matters benchmark, LLaVA-1.5-7B under pressure reaches 66.75% attack success rate (ASR); safe neutralization reduces ASR to 9.88%; full TPCD reaches 0.50% but collapses positives to 15.56%. A benchmark-specific task-prior/disagreement gate preserves measured positive accuracy (54.44%) while lowering ASR to 1.63% on LLaVA. Treating this LLaVA analysis as the design split, full n=800 negative and n=780 matched-positive held-out runs on GLM-4.6V and Llama-3.2-Vision show that simple gates can improve over safe neutralization, with sensitivity analyses bounding the weak time-positive subtask. A category-prior-free answer-disagreement router reduces held-out aggregate ASR to 6.93%, improving over both safe neutralization (10.98%) and branch disagreement (9.67%) while matching branch disagreement's 79.94% positive accuracy, although it remains post-hoc and surface-form based. We conclude that pressure is a useful probe of commitment bias and a viable mitigation signal, but the current gates are not yet independently validated grounding-aware detectors.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Hallucination mitigation in LVLMs, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 8 of its 8 references that are in the map. First posted on arXiv 29 Jul 2026.

Builds on

8 of 8 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026LLM-as-Judge Framework for Evaluating Tone-Induced Hallucination in Vision-Language Models20 Apr
  2. 2024Mitigating Hallucinations in Large Vision-Language Models with Instruction Contrastive Decoding27 Mar
  3. 2023Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding28 Nov
  4. 2023Evaluating Object Hallucination in Large Vision-Language Models17 May

Other references · 4

  1. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,624
  2. 2018Object Hallucination in Image Captioningcited by 661
  3. 2023DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Modelscited by 367
  4. 2024Sycophancy in Vision-Language Models: A Systematic Analysis and an Inference-Time Mitigation Frameworkcited by 17

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Cross-Image Contrastive Decoding: Precise, Lossless Suppression of Language Priors in Large Vision-Language Models
  2. 2026Rethinking CD: A Reproducibility Study and Extension on the Ineffectiveness of Contrastive Decoding at Mitigating Object Hallucinations in MLLMs
  3. 2026YARD: Y-Architecture Register Decoding for Efficient Hallucination Mitigation in Large Vision-Language Models
  4. 2026PromptCD: Test-Time Behavior Enhancement via Polarity-Prompt Contrastive Decoding
  5. 2025MaskCD: Mitigating LVLM Hallucinations by Image Head Masked Contrastive Decoding
  6. 2025ASCD: Attention-Steerable Contrastive Decoding for Reducing Hallucination in MLLM
  7. 2025ONLY: One-Layer Intervention Sufficiently Mitigates Hallucinations in Large Vision-Language Models
  8. 2026CHASD: Language Increment-Calibrated Contrastive Decoding against Hallucination in LVLMs