Stars
DASH: Detection and Assessment of Systematic Hallucinations of VLMs
Object Hallucination Benchmark for Vision Language Models
📖 A curated list of resources dedicated to hallucination of multimodal large language models (MLLM).
Unsupervised Concept Discovery Mitigates Spurious Correlations
Refine high-quality datasets and visual AI models
DiG-IN: Diffusion Guidance for Investigating Networks - Uncovering Classifier Differences, Neuron Visualisations, and Visual Counterfactual Explanations
Spurious Features Everywhere - Large-Scale Detection of Harmful Spurious Features in ImageNet
Easily compute clip embeddings and build a clip retrieval system with them
Distilling Model Failures as Directions in Latent Space