{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreigq4kyf5kbcypto2zemxhlbhonthq5srebffectk6h2rdtzrt43rm",
"uri": "at://did:plc:3fychdutjjusoqeq24ljch6q/app.bsky.feed.post/3mn7eexkbvii2"
},
"coverImage": {
"$type": "blob",
"ref": {
"$link": "bafkreiflo6xt7is6b2iafwghkjahlgggocme5jwjsbeuqqwcywuvjhmszm"
},
"mimeType": "image/png",
"size": 24783
},
"path": "/abs/2605.31176v1",
"publishedAt": "2026-06-01T00:00:00.000Z",
"site": "https://arxiv.org",
"tags": [
"Miltiadis Stouras",
"Vincent Cohen-Addad",
"Silvio Lattanzi",
"Ola Svensson"
],
"textContent": "**Authors:** Miltiadis Stouras, Vincent Cohen-Addad, Silvio Lattanzi, Ola Svensson\n\nRetrieval-augmented generation (RAG) systems typically rely on a single retriever and a single set of hyperparameters, despite facing highly heterogeneous queries that range from simple factoid questions to complex multi-hop reasoning. We propose a method that automatically selects a small, diverse subset of retrievers (a portfolio) from a large pool of candidates, to cover different regions of the target query distribution. We formalize this setting via an expected best-of-$k$ objective over the query distribution and show that it admits an efficient portfolio construction algorithm with near-optimal guarantees. Across multiple QA benchmarks, our learned portfolios and router pipeline consistently outperform single-retriever and naive multi-retriever baselines on both retrieval metrics and answer quality. In addition, compared to inference-time hyperparameter tuning approaches, fixed portfolios enable parallel retrieval and LLM calls, achieving comparable (and sometimes better) accuracy with substantially lower latency and token cost.",
"title": "Retriever Portfolios: A Principled Approach to Adaptive RAG"
}