List your company

Theorem Question Answering on TheoremQA standard (test)

56AccuracyOurs (theory-guided context selection strategy)
25.63233.51641.449.284Feb 9, 2026

Evaluation results

MethodLinks
Ours (theory-guided context selection strategy)Model=Qwen3-8B2026.0256
ExpRAGModel=Qwen3-8B2026.0255.6
ReMemModel=Qwen3-8B2026.0255.6
DCModel=Qwen3-8B2026.0255.4
BM25Model=Qwen3-8B2026.0254.8
ZeroModel=Qwen3-8B2026.0254.6
Ours (theory-guided context selection strategy)Model=Llama-3.1-8B2026.0228
ReMemModel=Llama-3.1-8B2026.0227.7
DCModel=Llama-3.1-8B2026.0227.6
ExpRAGModel=Llama-3.1-8B2026.0227.6
BM25Model=Llama-3.1-8B2026.0227
ZeroModel=Llama-3.1-8B2026.0226.8