List your company
ResearchFormerly WizWand

Multimodal Understanding on MME Cognition

1,293.8ScoreBLIP-2
182.248470.824759.41,047.976May 24, 2023Oct 27, 2023Mar 31, 2024Sep 4, 2024Feb 7, 2025Jul 13, 2025Dec 17, 2025
Updated 5mo ago

Evaluation results

MethodLinks
BLIP-2zero-shot=true2023.051,293.8
LLaMA-Adapter V2zero-shot=true2023.05972.6
LaVINzero-shot=true2023.05963.6
MiniGPT-4zero-shot=true2023.05866.5
DiffusionVLSize=7B, Type=Diff., Samples=738K2025.12675
Qwen2.5VLSize=7B, Type=AR, Samples=>9M2025.12646
Qwen2.5VLSize=3B, Type=AR, Samples=>9M2025.12620
DiffusionVLSize=3B, Type=Diff., Samples=738K2025.12594
LLaVAzero-shot=true2023.05502.8
LLaDA-VSize=8B, Type=Diff., Samples=16.5M2025.12491
DimpleSize=7B, Type=Diff., Samples=1.3M2025.12432
LyricsLLM Backbone=Vicuna-13B2023.12431.6
LLaVA-OVSize=7B, Type=AR, Samples=7.8M2025.12418
Qwen2-VL-2BModel Scale=2B, Model Access Type=Open-source2025.03413.9
MiniCPM-V2-3BModel Scale=3B, Model Access Type=Open-source2025.03396.8
ShareGPT4VLLM Backbone=Vicuna-7B2023.12376.4
AKI-4BModel Scale=4B, Model Access Type=Open-source2025.03362.9
Qwen-VL-7B-ChatModel Size=7B, Mode=Chat2023.11360.71
Qwen-VL-ChatLLM Backbone=Qwen-7B2023.12360.7
LLaMA-Adapter V22023.11356.43
LaViDa-LSize=8B, Type=Diff., Samples=1.6M2025.12341
PRISM2025.02330
SPHINX-2kResolution=2k2023.11326.8
SPHINX2023.11322.2
TIVE2025.02322.1
MM1.5-3BModel Scale=3B, Model Access Type=Proprietary2025.03319.6
DataTailor2025.02319.2
Full-Finetune2025.02311.9
SPHINX-1kResolution=1k2023.11310
Honeybee-C-7BModel Scale=7B, Model Access Type=Open-source2025.03307.1
Length2025.02306
Phi-3-Vision-4BModel Scale=4B, Model Access Type=Open-source2025.03302.9
BLIP-3-4BModel Scale=4B, Model Access Type=Open-source2025.03302.9
LLaVA-1.5-7BModel Scale=7B, Model Access Type=Open-source2025.03302.1
LLaVA-1.5LLM Backbone=Vicuna-13B2023.12295.4
LLaVA1.5-13BModel Size=13B2023.11295.36
EL2N2025.02294.7
BLIP-2LLM Backbone=FLAN-T52023.12290
GraNd2025.02287.1
MM1-3BModel Scale=3B, Model Access Type=Proprietary2025.03279.3
VILA-1.5-3BModel Scale=3B, Model Access Type=Open-source2025.03268.2
Perplexity2025.02260.7
LLaVALLM Backbone=Vicuna-7B2023.12247.9
Random2025.02233.5
DeepSeek-VL-1.3BModel Scale=1.3B, Model Access Type=Open-source2025.03225