This repository hosts the APASI Self-Injection (SI) Dataset, presented in the paper Mitigating Hallucinations in Large Vision-Language Models by Self-Injecting Hallucinations.
Code Repository:
https://github.com/davidluciolu/APASI
The APASI (Autonomous Preference Alignment via Self-Injection) method proposes a novel approach to mitigate hallucinations in Large Vision-Language… See the full description on the dataset page:
https://huggingface.co/datasets/lucio36/APASI-SI-dataset.