DAMO: Decoding by Accumulating Activations Momentum for Mitigating Hallucinations in Vision-Language Models.

Kaishen Wang, Hengrui Gu, Meijun Gao, Kaixiong Zhou

Browse the full ICLR paper archive.