List your company

Reasoning accuracy on PRONTOQA 4-hop

85AccuracyAVI
50.6859.5968.577.41May 17, 2025

Evaluation results

MethodLinks
AVIModel=Qwen 8B2025.0585
AVIModel=Qwen 4B2025.0572
No CorrectionModel=Qwen 8B2025.0565
Self CorrectionModel=Qwen 4B2025.0560
Self CorrectionModel=Qwen 8B2025.0558
No CorrectionModel=Qwen 4B2025.0552