List your company

Single-document Question Answering on Singledoc QA

32.06DuReader ScoreMemSearch-o1
14.608819.139423.6728.2006Apr 19, 2026

Evaluation results

MethodLinks
MemSearch-o1Backbone=Qwen 2.5-72B-Instruct2026.0432.0617.336.1849.0559.9145.53
Direct RAGBackbone=Qwen 2.5-72B-Instruct2026.0427.4917.7335.9445.4352.2938.51
MemSearch-o1Backbone=Deepseek V3.12026.0427.2323.0437.9452.2656.8148.17
AmberBackbone=Qwen 2.5-72B-Instruct2026.0420.5716.6527.1937.9852.2638.4
MemoryBankBackbone=Qwen 2.5-72B-Instruct2026.0419.2416.2626.2832.4243.4632.26
Search-o1 (RAgent)Backbone=Qwen 2.5-72B-Instruct2026.0418.9114.3722.8837.7749.9738.62
A-MemBackbone=Qwen 2.5-72B-Instruct2026.0418.7415.1822.0934.5950.7936.96
Direct RAGBackbone=Deepseek V3.12026.0418.7221.4832.4543.4148.1536.11
AmberBackbone=Deepseek V3.12026.0418.3922.4832.8639.2749.3139.79
MemoryBankBackbone=Deepseek V3.12026.0417.5721.8328.633.3442.3334.49
A-MemBackbone=Deepseek V3.12026.0417.120.9227.5528.2747.2932.22
Search-o1 (Refined)Backbone=Deepseek V3.12026.0416.1218.5922.9733.8244.7931.85
Search-o1 (Refined)Backbone=Qwen 2.5-72B-Instruct2026.0415.316.1217.5133.7944.4531.63
Search-o1 (RAgent)Backbone=Deepseek V3.12026.0415.2820.5933.0438.6749.7639.05