Our new X account is live! Follow @wizwand_team for updates

Multimodal Evaluation on LLaVA-bench in-the-wild

97.3Score

MoE-LLaVA-2.7Bx4-Top2†

Updated 2d ago

Evaluation Results

Method	Links
MoE-LLaVA-2.7Bx4-Top2† 2024.01		97.3
MoE-LLaVA-2.7Bx4-Top2 2024.01		94.1
MoE-LLaVA-1.6Bx4-Top2† 2024.01		90.3
MoE-LLaVA-1.8Bx4-Top2 2024.01		88.7
MoE-LLaVA-1.6Bx4-Top2 2024.01		86.8
LLaVA-NeXT-8B 2024.08		80.1
LongVILA-7B (S3) 2024.08		77.6
SliME 2024.11		73.9
VILA 2024.08		73
GroundingGPT 2024.01		70.9
LLaVA-1.5 2024.08		70.7
LLaVA-1.5 2024.01		70.7
LLaVA-1.5 2024.11		70.5
VILA 2024.08		69.7
LLaVolta 2024.06		69.2
LLaVolta 2024.06		68.2
LLaVolta 2024.06		67.7
LLaVolta 2024.06		67.5
LLaVolta 2024.06		67.3
Mini-Gemini (MGM-2B) 2024.06		65.9
LLaVolta (MGM-2B backbone) 2024.06		65.7
FocusLLaVA 2024.11		65.6
LLaVolta 2024.06		65.5
LLaVA-1.5 2024.11		65.4
LLaVA-NeXT 2024.11		64.7
LLaVolta 2024.06		64.3
LLaVA-1.5 2024.08		63.4
LLaVA-1.5 2024.01		63.4
LLaVA-1.5 2024.01		63.4
Instruct-BLIP 2024.11		60.9
InstructBLIP 2024.08		60.9
InstructBLIP 2024.01		60.9
Trim 2024.11		58.7
InstructBLIP 2024.01		58.2
BLIP-2 2024.08		38.1
BLIP-2 2024.01		38.1