List your company

Reasoning accuracy on PRONTOQA 3-hop

87AccuracyAVI
52.6861.5970.579.41May 17, 2025

Evaluation results

MethodLinks
AVIModel=Qwen 8B2025.0587
AVIModel=Qwen 4B2025.0568
No CorrectionModel=Qwen 4B2025.0560
No CorrectionModel=Qwen 8B2025.0554
Self CorrectionModel=Qwen 4B2025.0554
Self CorrectionModel=Qwen 8B2025.0554