DecisionRank:面向決策模型重排器的 on-policy 蒸餾
DecisionRank: On-Policy Distillation for Decision-Model Rerankers
DecisionRank 以 item-space on-policy 蒸餾,把生成式教師模型的排序知識轉移至決策模型,在檢索基準上達到最先進表現,推理速度亦顯著提升。1.7B 參數的 DecisionRank 在 TREC DL19–23 超越 27B 生成式教師模型;8B 版本在 TREC DL 上快 24.5 倍,在 BRIGHT 上快 16.7 倍。
Published on Oct 30
Authors:
,
,
,
,
,
,
,
,
,
,
,
Abstract
DecisionRank uses item-space on-policy distillation to transfer ranking knowledge from generative teachers to efficient decision models, achieving state-of-the-art retrieval performance with significantly faster inference.
Decision models target structured tasks by directly predicting choices or scores, avoiding autoregressive decoding latency. However, transferring a generative LLM's ranking knowledge to a decision model remains challenging: the teacher expresses preferences through a ranked list of document ID sequences, while the student determines the ranking using scores from the first-token position. To bridge this gap, we introduce DecisionRank, a decision model trained through item-space on-policy distillation (OPD). Item-space OPD constructs the student's rollout by sorting its document scores into a ranked sequence. A generative teacher takes the resulting ranking list as inputs and provides conditional feedback to directly supervise those scores. Combined with offline ranking supervision, DecisionRank achieves state-of-the-art performance on both general and reasoning-intensive retrieval benchmarks such as TREC DL, BEIR, and BRIGHT. Across four benchmarks, item-space OPD consistently outperforms alternative training objectives. Even at 1.7B parameters, DecisionRank surpasses its 27B generative teacher on TREC DL19--23. At 8B, it outperforms its generative listwise counterpart while running 24.5times faster on TREC DL and 16.7times faster on BRIGHT.
View arXiv page View PDF Add to collection
Get this paper in your agent:
hf papers read 2610.42128
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash
Models citing this paper 0
No model linking this paper
Cite arxiv.org/abs/2610.42128 in a model README.md to link it from this page.
Datasets citing this paper 0
No dataset linking this paper
Cite arxiv.org/abs/2610.42128 in a dataset README.md to link it from this page.
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2610.42128 in a Space README.md to link it from this page.
Collections including this paper 0
No Collection including this paper
Add this paper to a collection to link it from this page.
來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co