1 paper
Arnav Bendre, Guneesh Gupta, Kavish Grover +2
Multimodal large language models frequently describe objects that are not present in an image. Contrastive decoding has become a widely used inference-time remedy, and it is assume…