paper-with-me

Papers

How Many Counterfactuals Does It Take? Probing VLM Hallucinations Through Circuits and Causal Effects

2026-06-07 · Abhivansh Gupta, Simardeep Singh, Advika Sinha, Shreyansh Modi, Akshat Tomar arxiv

Visual Language Models (VLMs) are known to produce hallucinated predictions that are not grounded in visual evidence, yet existing approaches lack a principled understanding of how robust such predictions are under counterfactual perturbations. In this work, we study the sample complexity of counterfactual robustness for hallucinated outputs in VLMs. We define a causal influence metric based on log-probability differences between factual, counterfactual, and activation-patched runs, and use it to characterize the stability of hallucinated predictions. By leveraging circuit discovery techniques (CD-T), we identify model components responsible for these predictions and track their activation differences across counterfactual samples. We then derive empirical bounds on the minimum number of counterfactual samples m required to reliably detect instability in hallucinated outputs, using concentration inequalities and variance estimates of the causal influence distribution.

📄 PDF Abstract BibTeX arXiv:2606.08777

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Flexible text generation for counterfactual fairness probing

2022-06-28 · NAACL (WOAH) 2022 7 · Zee Fryer, Vera Axelrod, Ben Packer, Alex Beutel 외

A common approach for testing fairness issues in text-based classifiers is through the use of counterfactuals: does the classifier output change if a sensitive attribute in the input is changed? Existing counterfactual g…

AttributecounterfactualFairnessText Generation

SocialCounterfactuals: Probing and Mitigating Intersectional Social Biases in Vision-Language Models with Counterfactual Examples

2023-11-30 · CVPR 2024 1 · Phillip Howard, Avinash Madasu, Tiep Le, Gustavo Lujan Moreno 외

While vision-language models (VLMs) have achieved remarkable performance improvements recently, there is growing evidence that these models also posses harmful biases with respect to social attributes such as gender and …

counterfactual

CAGE: Probing Causal Relationships in Deep Generative Models

2021-09-29 · Joey Bose, Ricardo Pio Monti, Aditya Grover

Deep generative models excel at generating complex, high-dimensional data, often exhibiting impressive generalization beyond the training distribution. The learning principle for these models is however purely based on s…

Robust classificationSynthetic Data Generation

Probing the Robustness of Time-series Forecasting Models with CounterfacTS

2024-03-06 · Håkon Hanisch Kjærnli, Lluis Mas-Ribas, Aida Ashrafi, Gleb Sizov 외

A common issue for machine learning models applied to time-series forecasting is the temporal evolution of the data distributions (i.e., concept drift). Because most of the training data does not reflect such changes, th…

Time SeriesTime Series Forecasting

Towards Fair Graph Neural Networks via Graph Counterfactual

2023-07-10 · Zhimeng Guo, Jialiang Li, Teng Xiao, Yao Ma 외

Graph neural networks have shown great ability in representation (GNNs) learning on graphs, facilitating various tasks. Despite their great performance in modeling graphs, recent works show that GNNs tend to inherit and …

counterfactualFairnessGraph LearningNode Classification