ResearchBenchmarksConspiracy marker extraction on SemEval Task 10 2026 (test)Follow75Macro F1Agentic18.8433.424862.58Mar 5, 2026Evaluation ResultsMethodMethodLinksMacro F1DeltaAgenticbase model=GPT-5.2, op...base model=GPT-5.2, optimization=GEPA, workflow=agentic pipeline, retrieval=contrastive retrieval, architecture=Council-Judge2026.0375—Agenticbase model=GPT-5.2, op...base model=GPT-5.2, optimization=GEPA, workflow=agentic pipeline, retrieval=contrastive retrieval2026.0321—