REVIEW 7 cited by
Reducing Hallucinations in Vision-Language Models via Latent Space Steering
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Hallucination poses a challenge to the deployment of large vision-language models (LVLMs) in applications. Unlike in large language models (LLMs), hallucination in LVLMs often arises from misalignments between visual inputs and textual outputs. This paper investigates the underlying mechanisms of hallucination, focusing on the unique structure of LVLMs that distinguishes them from large language models (LLMs). We identify that hallucinations often arise from the sensitivity of text decoders to vision inputs, a natural phenomenon when image encoders and text decoders are pre-trained separately. Inspired by this, we introduce Visual and Textual Intervention (VTI), a novel technique designed to reduce hallucinations by steering latent space representations during inference to enhance the stability of vision features. As a task-agnostic test-time intervention, VTI can be easily applied to any problem without additional cost. Extensive experiments demonstrate that it can effectively reduce hallucinations and outperform baseline methods across multiple metrics, highlighting the critical role of vision feature stability in LVLMs.
Forward citations
Cited by 7 Pith papers
-
Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens
Modality bias, an imbalanced attention to text or image during hallucinated outputs, is shown to be mitigated by a training-free attention intervention plus contrastive decoding.
-
GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs
GrAInS uses Integrated Gradients to identify the most influential tokens, then builds layer-wise steering vectors that improve truthfulness, reduce hallucination, and preserve general capabilities in LLMs and VLMs.
-
Not All Tokens and Heads Are Equally Important: Dual-Level Attention Intervention for Hallucination Mitigation
A dual-level attention intervention that boosts salient visual-token attention and suppresses text/system attention during decoding reduces hallucination rates in LLaVA, MiniGPT-4, and mPLUG-Owl2 on POPE and CHAIR.
-
The Latent Space: Foundation, Evolution, Mechanism, Ability, and Outlook
A large survey organizes latent-space work in language-based models by foundation, evolution, four mechanisms, seven abilities, and open challenges.
-
CAI: Caption-Sensitive Attention Intervention for Mitigating Object Hallucination in Large Vision-Language Models
CAI reduces object hallucination in LVLMs by injecting caption-query attention patterns into selected attention heads at inference time.
-
ReCo: Reminder Composition Mitigates Hallucinations in Vision-Language Models
ReCo, a lightweight DPO-trained linear head that re-injects pooled image embeddings at every step, reduces hallucination on five benchmarks across three VLMs and combines with existing mitigation methods.
-
Empowering Multimodal LLMs with External Tools: A Comprehensive Survey
A survey paper maps how external tools are used to augment multimodal large language models across data, tasks, evaluation, and future directions.
Discussion (0). Sign in to comment.