List your company
ResearchFormerly WizWand

Real-world Multi-modal Question Answering on RWQA

74.77AccuracyFlash Linear Attention
61.177264.706168.23571.7639Oct 27, 2025Dec 2, 2025Jan 8, 2026Feb 14, 2026Mar 22, 2026Apr 28, 2026Jun 4, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Flash Linear AttentionModel scale=4B2026.0674.77
Multiplication-Only Matrix Inversion ApproximationModel scale=4B2026.0674.64
Flash Linear AttentionModel scale=9B2026.0674.38
Multiplication-Only Matrix Inversion ApproximationModel scale=9B2026.0674.12
MergeMixbaseline=SFT Vision2025.1070.46
VisionThink-7B2025.1069.28
Qwen2.5-VL-Ins-7B2025.1068.63
SFT Vision2025.1068.63
Multiplication-Only Matrix Inversion ApproximationModel scale=2B2026.0665.62
Flash Linear AttentionModel scale=2B2026.0665.23
Flash Linear AttentionModel scale=0.8B2026.0662.35
Multiplication-Only Matrix Inversion ApproximationModel scale=0.8B2026.0661.7