Paper Shows Decision Models Fall Short as Rerankers

An arXiv paper from the University of Passau and others systematically compares decision models with fine-tuned rerankers, finding that traditional rerankers outperform decision models by 5 to 29 points across 0.6B-9B scales.

2026-10-26 ~ 2026-10-26 · 2 related posts

• Study: Decision (System One) fine-tuning never beats plain reranker fine-tuning on Qwen backbones — _reachsumit · 2026-10-26
• Study: reranker fine-tuning beats decision models by 5-29 nDCG points up to 9B params — kalyan_kpl · 2026-10-26