ProRank: Prompt Warmup via Reinforcement Learning for Small Language Models Reranking
Xianming Li, Aamir Shakir, Rui Huang, Julius Lipp, Jing Li, Mixedbread AI, Berlin, Germany, The Hong Kong Polytechnic University, Hong Kong SAR {sean,aamir,rui,julius}@[email protected],[email protected] https://hf.co/mixedbread-ai/mxbai-rerank-large-v2 · 2025
Read ProRank: Prompt Warmup via Reinforcement Learning for Small Language Models Reranking by Xianming Li, Aamir Shakir, Rui Huang, Julius Lipp, Jing Li, Mixedbread AI, Berlin, Germany, The Hong Kong Polytechnic University, Hong Kong SAR {sean,aamir,rui,julius}@[email protected],[email protected] https://hf.co/mixedbread-ai/mxbai-rerank-large-v2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-03487.txt.
Description
Xianming Li and co-authors present ProRank in 2025, a way to make small language models strong at reranking search results. Reranking, which reorders first-pass results by relevance, is core to search and to retrieval-fed AI answers. Most top rerankers use large models that cost too much for many uses.
Small models are cheap but lag in quality, and the authors first measured where they fall short. Their fix warms up the model's prompts through reinforcement learning, tuning how the task is posed rather than growing the model. The warmed small model closes much of the gap to large rivals. Search builders get a cheaper path to quality ranking, useful where volume is high and budgets are thin.
Catalog record
| CREATOR | Xianming Li, Aamir Shakir, Rui Huang, Julius Lipp, Jing Li, Mixedbread AI, Berlin, Germany, The Hong Kong Polytechnic University, Hong Kong SAR {sean,aamir,rui,julius}@[email protected],[email protected] https://hf.co/mixedbread-ai/mxbai-rerank-large-v2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-03487