TL;DR
Prism-Reranker enhances retrieval reranking by jointly producing relevance judgments, contribution summaries, and evidence passages, improving downstream agent performance and interpretability.
Contribution
It introduces a reranker model that outputs not only relevance scores but also contribution statements and evidence passages, advancing beyond traditional scalar relevance scoring.
Findings
Prism-Reranker achieves strong results across multiple sizes on BEIR and LLM-judged evaluations.
Augmenting existing rerankers with contribution and evidence improves NDCG@10 by +1.54.
Model weights, training recipes, and evaluation tools are publicly released.
Abstract
Modern retrieval pipelines increasingly serve downstream consumers like retrieval-augmented generation (RAG) and autonomous agents that need more than a scalar relevance score. A reranker that only tells the caller "how relevant" forces the agent to dump entire documents into the language-model context, wasting tokens on tangential passages and boilerplate. We introduce Prism-Reranker, a family of reranker models built on Qwen3.5 at four sizes (0.8B, 2B, 4B, 9B) that goes beyond scalar scoring. In addition to the standard yes/no relevance judgement, whenever the verdict is yes the model emits (i) a contribution statement summarizing how the document helps the query, and (ii) an evidence passage: a self-contained rewrite that preserves every query-relevant signal while discarding noise. Prism-Reranker is trained with a hybrid objective combining point-wise distillation from a strong…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
- 🤗infgrad/Prism-Qwen3.5-Reranker-0.8Bmodel· 2.0k dl· ♡ 102.0k dl♡ 10
- 🤗infgrad/Prism-Qwen3.5-Reranker-9Bmodel· 273 dl· ♡ 4273 dl♡ 4
- 🤗infgrad/Prism-Qwen3.5-Reranker-2Bmodel· 191 dl· ♡ 2191 dl♡ 2
- 🤗infgrad/Prism-Qwen3.5-Reranker-4Bmodel· 396 dl· ♡ 2396 dl♡ 2
- 🤗infgrad/Prism-Qwen3-Reranker-4B-expmodel· 89 dl· ♡ 289 dl♡ 2
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
