List your company

Multi-page Question Answering on LongBench multi-page

48.9HQA F1AGAR
23.9430.4236.943.38Jun 11, 2026

Evaluation results

MethodLinks
AGARBackbone=InternVL3.5-8B2026.0648.943.825.638—
Plain VQABackbone=InternVL3.5-8B2026.0648.643.12437.2—
UniformBackbone=InternVL3.5-8B2026.0648.643.12437.2—
VEA-BrBackbone=InternVL3.5-8B2026.0648.331.37.121—
RandomBackbone=InternVL3.5-8B2026.0647.542.923.837.3—
AGARBackbone=GLM-4.1V-9B-Thinking2026.0646.950.726.825.9—
VEA-BrBackbone=GLM-4.1V-9B-Thinking2026.0646.750.224.723.6—
VEA-ConBackbone=GLM-4.1V-9B-Thinking2026.0645.949.923.924.9—
CGRBackbone=GLM-4.1V-9B-Thinking2026.0645.824.721.919.9—
AGLABackbone=GLM-4.1V-9B-Thinking2026.0645.738.926.518.6—
VARBackbone=GLM-4.1V-9B-Thinking2026.0645.650.324.522.7—
Plain VQABackbone=GLM-4.1V-9B-Thinking2026.0644.550.124.824.8—
UniformBackbone=GLM-4.1V-9B-Thinking2026.0644.550.224.824.2—
AGARAvg. tok.=5,3842026.0644.340.822.134.2—
RandomBackbone=GLM-4.1V-9B-Thinking2026.0643.546.525.523.9—
VEA-ConAvg. tok.=4,9162026.0642.136.118.229.9—
VEA-BrAvg. tok.=4,9162026.0641.835.221.330.3—
AGLAAvg. tok.=4,9162026.0641.239.121.731.5—
RandomAvg. tok.=5,3512026.0640.837.317.633.8—
Plain VQAAvg. tok.=4,9162026.0639.837.517.229.5—
UniformAvg. tok.=5,3782026.0639.837.516.933.3—
VARAvg. tok.=4,9162026.0637.739.218.530.9—
VEA-ConBackbone=InternVL3.5-8B2026.063732.910.122.4—
CGRAvg. tok.=4,9162026.0634.136.621.229.3—
AGLABackbone=InternVL3.5-8B2026.0631.729.517.817.9—
VARBackbone=InternVL3.5-8B2026.063030.213.412.4—
CGRBackbone=InternVL3.5-8B2026.0624.923.89.519.5—