List your company
ResearchFormerly WizWand

Visual Reasoning on NLVR2 (val)

91.1AccuracyLLaVA-NeXT-Interleave-14B
49.29260.1467181.854Sep 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
LLaVA-NeXT-Interleave-14BModel Scale=>14B2024.0991.1
MM1.5-30BModel Scale=>14B2024.0990.6
Mantis-Idefics2-8BModel Scale=7B2024.0989.7
LLaVA-NeXT-Interleave-7BModel Scale=7B2024.0988.8
GPT-4VModel Scale=>14B2024.0988.8
Idefics2-8BModel Scale=7B2024.0986.9
MM1.5-7BModel Scale=7B2024.0986.9
MM1.5-3B-MoEModel Scale=3B2024.0986
MM1.5-3BModel Scale=3B2024.0983.8
MM1.5-1B-MoEModel Scale=1B2024.0983.2
MM1.5-1BModel Scale=1B2024.0979
LLaVA-NeXT-Interleave-0.5BModel Scale=1B2024.0967.8
LLaVAOneVision-0.5BModel Scale=1B2024.0963.4
MM1-30BModel Scale=>14B2024.0963.1
MM1-7BModel Scale=7B2024.0959.9
Emu2-Chat-37BModel Scale=>14B2024.0958.2
LLaVA-v1.5-7BModel Scale=7B2024.0953.9
Phi-3-Vision-4BModel Scale=3B2024.0953.6
MM1-3BModel Scale=3B2024.0951.7
MM1-1BModel Scale=1B2024.0950.9