Meta's DecisionRank: 1.7B Distilled Reranker Beats 27B Generative Teacher

Meta and Purdue propose DecisionRank, which distills generative LLM ranking knowledge into an efficient decision-model reranker via on-policy distillation. The 1.7B student outperforms its 27B generative teacher while running 24.5x faster.

2026-11-02 ~ 2026-11-03 · 3 related posts

• Meta's DecisionRank: 1.7B Reranker Beats 27B Teacher While Running 24.5x Faster — _reachsumit · 2026-11-02
• Meta's DecisionRank: 1.7B distilled reranker beats its 27B generative teacher — meta · 2026-11-03

1 near-duplicate retellings: kalyan_kpl