List your company

Question Answering on NQ-Conflict (test)

71.4Correctness (%)ConflictRAG
51.32856.53961.7566.961May 17, 2026

Evaluation results

MethodLinks
ConflictRAGSeeds=32026.0571.452.671.8
CoT DetectionTemperature=02026.0565.348.745.1
NLI-FilterTemperature=02026.0561.546.537.5
CRAGTemperature=02026.0557.245.126
RAG+RerankTemperature=02026.0555.843.824.7
Standard RAGTemperature=02026.0553.441.223.4
Self-RAGTemperature=02026.0552.139.523.7