List your company

Hallucination Evaluation on MSCOCO (val) (CHAIR/TPS)

23.04CHAIR_iSparseVLM
12.036814.893417.7520.6066Jan 11, 2025

Evaluation results

MethodLinks
SparseVLMBase Model=mPLUG-Owl2, Image Token Sparsity=true2025.0123.0468.4223.19
FastVBase Model=mPLUG-Owl2, Image Token Sparsity=true2025.0122.6567.6824.37
SparseVLMBase Model=MiniGPT-4, Image Token Sparsity=true2025.0119.8555.2737.49
FastVBase Model=MiniGPT-4, Image Token Sparsity=true2025.0119.6754.5938.05
VCDBase Model=mPLUG-Owl22025.0119.4662.449.77
GreedyBase Model=mPLUG-Owl22025.0119.0761.2819.96
SIDBase Model=mPLUG-Owl22025.0118.4760.8212.85
DoLaBase Model=mPLUG-Owl22025.0118.4360.1814.23
WoodpeckerBase Model=mPLUG-Owl2, Post-hoc=true2025.0118.3959.58—
SparseVLMBase Model=LLaVA-1.5, Image Token Sparsity=true2025.0118.0952.432.06
OPERABase Model=mPLUG-Owl22025.0117.8656.293.49
LUREBase Model=mPLUG-Owl2, Post-hoc=true2025.0117.8557.73—
HALCBase Model=mPLUG-Owl22025.0117.6356.121.5
FastVBase Model=LLaVA-1.5, Image Token Sparsity=true2025.0117.6251.9433.18
Beam SearchBase Model=mPLUG-Owl22025.0117.1254.6619.58
VCDBase Model=MiniGPT-42025.0117.0943.817.68
VASparseBase Model=mPLUG-Owl22025.0117.0253.717.86
OPERABase Model=MiniGPT-42025.0115.4242.425.27
VCDBase Model=LLaVA-1.52025.0114.8249.7617.55
GreedyBase Model=LLaVA-1.52025.0114.6349.6631.17
GreedyBase Model=MiniGPT-42025.0114.0643.6536.28
Beam SearchBase Model=MiniGPT-42025.0113.944.4532.1
DoLaBase Model=MiniGPT-42025.0113.8544.224.75
WoodpeckerBase Model=MiniGPT-4, Post-hoc=true2025.0113.7644.07—
DoLaBase Model=LLaVA-1.52025.0113.7550.0323.4
HALCBase Model=MiniGPT-42025.0113.7343.683.68
SIDBase Model=MiniGPT-42025.0113.6843.6522.67
Beam SearchBase Model=LLaVA-1.52025.0113.6248.8929.89
LUREBase Model=MiniGPT-4, Post-hoc=true2025.0113.4943.92—
SIDBase Model=LLaVA-1.52025.0113.2947.0919.57
VASparseBase Model=MiniGPT-42025.0113.2943.0229.74
WoodpeckerBase Model=LLaVA-1.5, Post-hoc=true2025.0113.2749.72—
LUREBase Model=LLaVA-1.5, Post-hoc=true2025.0113.0847.95—
OPERABase Model=LLaVA-1.52025.0112.9847.64.07
HALCBase Model=LLaVA-1.52025.0112.9346.352.04
VASparseBase Model=LLaVA-1.52025.0112.4646.2127.53