Share your thoughts, 1 month free Claude Pro on usSee more

Multi-image reasoning on BLINK (val)

52.6Accuracy

LLaVA-NeXT-Interleave-7B

Updated 1mo ago

Evaluation Results

Method	Links
LLaVA-NeXT-Interleave-7B 2024.09		52.6
LLaVAOneVision-0.5B 2024.09		52.1
LLaVA-NeXT-Interleave-14B 2024.09		52.1
GPT-4V 2024.09		51.1
MM1.5-30B 2024.09		50.2
MM1.5-3B-MoE 2024.09		49.8
BLIP-3-4B 2024.09		49.7
Mantis-Idefics2-8B 2024.09		49.1
MM1.5-7B 2024.09		48.2
MM1.5-3B 2024.09		46.8
MM1.5-1B 2024.09		46.3
Idefics2-8B 2024.09		45.2
Phi-3-Vision-4B 2024.09		44.2
MM1.5-1B-MoE 2024.09		43.7
MM1-30B 2024.09		43.5
MM1-3B 2024.09		41.5
MM1-1B 2024.09		40.3
MM1-7B 2024.09		40
LLaVA-NeXT-Interleave-0.5B 2024.09		39.2
LLaVA-v1.5-7B 2024.09		37.1
Emu2-Chat-37B 2024.09		36.2