List your company

Audio-Visual Perception on WorldSense

47.4ScoreEchoingPixels
3739.742.445.1Dec 11, 2025Jan 14, 2026Feb 17, 2026Mar 23, 2026Apr 26, 2026May 30, 2026Jul 3, 2026

Evaluation results

MethodLinks
EchoingPixelsModel Scale=7B, Token Budget=10%2025.1247.4
Full Model (Qwen2.5-Omni-7B)Model Scale=7B, Token Budget=100%2025.1246.1
Full TokensRetained Ratio (%)=100, Model Backbone=Qwen2.5-Omni-7B2026.0745.81
Full Model (Qwen2.5-Omni-3B)Model Scale=3B, Token Budget=100%2025.1245.4
Full TokensRetained Ratio (%)=100, Model Backbone=Qwen2.5-Omni-3B2026.0745.21
EchoingPixelsModel Scale=3B, Token Budget=20%2025.1245
OmniFocusRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-7B2026.0744.77
OmniZipRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-7B2026.0744.55
OmniFocusRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-3B2026.0744.51
OmniFocusRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-7B2026.0744.23
RandomRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-7B2026.0744.14
OmniFocusRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-3B2026.0744.1
OmniZipRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-3B2026.0743.88
RandomRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-7B2026.0743.82
OmniZipRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-7B2026.0743.69
EchoingPixelsModel Scale=3B, Token Budget=10%2025.1243.5
RandomRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-3B2026.0743.44
DyCokeRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-3B2026.0743.32
DyCokeRetained Ratio (%)=35, Model Backbone=Qwen2.5-Omni-7B2026.0743.28
DyCokeRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-7B2026.0743.25
DyCokeRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-3B2026.0743.25
OmniZipRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-3B2026.0743.06
RandomRetained Ratio (%)=25, Model Backbone=Qwen2.5-Omni-3B2026.0742.78
EchoingPixelsModel Scale=3B, Token Budget=5%2025.1240.9
IntraModalModel Scale=7B, Token Budget=25%2025.1240.6
IntraModalModel Scale=3B, Token Budget=25%2025.1237.4