List your company
ResearchFormerly WizWand

inductive Prolog rule synthesis on SLR-Bench Hard tier 250 tasks 1

58.4AccuracyDF + CC + QO
13.05624.82836.648.372May 6, 2026
Updated 6mo ago

Evaluation results

MethodLinks
DF + CC + QOCategory=Hybrid (symbolic + LLM fallback)2026.0558.4
DF + CCCategory=Hybrid (symbolic + LLM fallback)2026.0558
BoK + CCCategory=Hybrid (symbolic + LLM fallback)2026.0552.8
BoK + CC + QOCategory=Hybrid (symbolic + LLM fallback)2026.0552.8
DF + QOCategory=Hybrid (symbolic + LLM fallback)2026.0549.6
CC SolverCategory=Symbolic solvers2026.0546.8
gpt-5†Category=LLM-only methods2026.0546
o3†Category=LLM-only methods2026.0545
BoK + QOCategory=Hybrid (symbolic + LLM fallback)2026.0540.8
QO SolverCategory=Symbolic solvers2026.0537.2
DF-32Category=LLM-only methods2026.0534.4
BoKCategory=LLM-only methods2026.0517.2
Qwen3.6-35B-A3B (OpenHands)Category=LLM-only methods2026.0514.8