List your company

Detailed Image Captioning on MS COCO 2014 (val)

10.1CHAIRiLLaVA
7.3968.0988.89.502Mar 20, 2024

Evaluation results

MethodLinks
LLaVABackbone=LLaVA-7B, Decoding Strategy=Standard2024.0310.135.671.1
LLaVA ContrastiveBackbone=LLaVA-7B, Decoding Strategy=Contrastive2024.0310.133.769.9
LLaVA PMIBackbone=LLaVA-7B, Decoding Strategy=PMI2024.039.83470.1
LLaVA M3IDBackbone=LLaVA-7B, Decoding Strategy=M3ID2024.038.828.867.4
LLaVA PMIBackbone=LLaVA-13B, Decoding Strategy=PMI2024.038.628.567.1
LLaVA M3ID + DPOBackbone=LLaVA-7B, Decoding Strategy=M3ID + DPO2024.038.528.168.9
LLaVABackbone=LLaVA-13B, Decoding Strategy=Standard2024.038.531.867
LLaVA ContrastiveBackbone=LLaVA-13B, Decoding Strategy=Contrastive2024.038.128.768.2
LLaVA M3IDBackbone=LLaVA-13B, Decoding Strategy=M3ID2024.037.827.568.2
LLaVA M3ID + DPOBackbone=LLaVA-13B, Decoding Strategy=M3ID + DPO2024.037.526.967.7