This work proposes Exact Adaptive Hybrid Retrieval (EAHR), which fixes the ordered Top-$K$ defined by complete-list weighted RRF as the retrieval target and treats channel depth as request-specific execution state and reproduced the complete-list ordered Top-20 in all 150 query-snapshot combinations.
Abstract
Modern retrieval-augmented generation (RAG) systems often fuse fixed Top-$L$ results from dense and sparse retrievers, treating later contributions as zero. The cutoff therefore determines both the ranking and its execution cost. Yet truncated fusion is not generally equivalent to complete-list fusion: unread cross-list ranks can change Top-$K$ membership or order even when the observed candidates contain every item in the complete-list Top-$K$. Because channel rankings vary across queries and corpus updates, a depth selected from historical queries may not transfer reliably. We propose Exact Adaptive Hybrid Retrieval (EAHR), which fixes the ordered Top-$K$ defined by complete-list weighted RRF as the retrieval target and treats channel depth as request-specific execution state. Per-Vector Scalar Quantization (PVS) and Posting Block-Max (PBM) produce resumable exact dense and sparse rankings. Fusion bounds unread contributions and requests further ranks only while they can change the Top-$K$. Every successful request therefore matches complete-list fusion without a preset Top-$L$; otherwise, execution continues safely to list exhaustion. Across five test collections and five temporal corpus snapshots, complete-list weighted RRF remained competitive, whereas fixed depths selected from historical queries did not transfer reliably. EAHR reproduced the complete-list ordered Top-20 in all 150 query-snapshot combinations. Under a warm-cache, interleaved, order-balanced protocol, the paired geometric-mean latency ratios of exhaustive batch execution to EAHR were 23.35 on TREC-DL 2019 and 30.28 on TREC-DL 2020. Anti-correlated rankings exhausted both lists, and some difficult queries were slower with EAHR. EAHR does not guarantee a speedup for every request; it fixes the exact result while adapting execution depth to the current rankings.
This work presents DESA (Dense Expansion and Sparse Anchoring), which shares generated references across channels but specializes their integration, which improves nDCG@10 and Recall@20 over the unexpanded query and reduces dense and sparse replay stopping depths.
The Adaptive Multi-Stage Vector Retrieval (AMSVR) framework is proposed, prioritising weighted, drift-resistant composition over uniform fusion, and offers tailored configurations: AMSVR-Scientific (dense + tuned hybrid) peaks at NDCG@10 = 0.7570 on SciFact, while AMSVR-Full (seven stages) targets broader, noisier corpora where Recall@100 matters most.
Samsudeen Alabi Bankole, Yakub Kayode Saheed· NLP & Big Data· 0 citations
Top-$K$ sparse attention reduces the cost of Softmax and value aggregation by attending to only a small subset of key--value (KV) entries. However, identifying this subset still requires scoring the current query against the full KV cache and performing global Top-$K$ selection, leaving selector cost linear in context length and limiting the practical efficiency of sparse attention for long-context decoding. In this paper, we introduce ReTopK, a training-free method that accelerates dynamic Top-$K$ attention by reusing historical retrieval decisions. ReTopK builds on the observation that similar queries often attend to overlapping supports and that partially overlapping supports can still preserve most of the Exact Top-$K$ attention mass. For each attention head, it maintains a bounded cache of historical query--support pairs, retrieves the most similar cached queries for each new query, unions their stored supports with a recent window, and reranks only the resulting compact candidate set using exact current-query scores. A similarity-based fallback invokes full-history Exact Top-$K$ when reuse is unreliable, while periodic exact refreshes limit cache drift. ReTopK retains the complete KV cache and reuses only selected indices, rather than historical scores, attention weights, or outputs. Across 16K--128K contexts, ReTopK achieves the lowest PG19 perplexity and the highest NIAH and LongBench scores among the evaluated approximate methods. At 128K with $K=512$, ReTopK incurs only a 0.50\% perplexity increase over Exact Top-$K$ while accelerating attention computation by $3.07\times$.
DARE-RAG is proposed, an adaptive retrieval framework that activates LLM-based query expansion only for retrieval-challenging queries, and formulates expansion activation as a lightweight binary classification problem using probe retrieval signals, including score margin, variance, entropy, query length, and lexical specificity.
Li-Xiang Zhu· Poster Volume 0007 The 2026...· 0 citations
: Reciprocal Rank Fusion (RRF) is the standard recipe for combining a sparse and a dense retriever in retrieval augmented generation, but it assigns both components a fixed and equal weight on every query. This is a strong assumption, since a query of rare specific tokens is best served by the sparse retriever while a paraphrastic query (one composed largely of common words that admits many surface forms of the same intent) is best served by the dense one. We show on four BEIR corpora that equal weight RRF fails to surpass its stronger component on any of them, and we propose Query Adaptive Rank Fusion (QARF), a training free variant whose fusion weight is a closed form function of standard corpus IDF statistics already available at index build time. QARF operates in two regimes selected automatically by the data: per query weighting on corpora with a broad IDF distribution, and automatic corpus level recalibration of RRF on corpora where almost every query is more specific than the mean. Using a BM25 + E5 large backbone, QARF improves over its own RRF baseline on all four corpora by + 1 . 74, + 5 . 12, + 8 . 61, and + 2 . 97 NDCG@10 points on NFCorpus, SciFact, FiQA, and ArguAna, matches pure E5 within one point on SciFact and FiQA, and exceeds it on ArguAna while retaining the hybrid’s wider candidate pool. End to end evaluation with Gemma 3 1B shows that the retrieval gain transfers downstream as a lower abstain rate on FiQA (4 . 6% → 3 . 2%, the lowest of any retriever); on the other corpora the rate stays within noise of RRF, so we report commit rate as a per corpus companion metric rather than a uniform claim.
Md. Rafeeq, Chandramani Chaudhary, N. Boran et al.· Proceedings of the 15th Inte...· 0 citations
This work introduces AnchorQE, a training-free method that separately encodes the original query and its expansion before interpolating them, and shows that AnchorQE improves retrieval effectiveness by up to 12.89% when compared to widely-used expansion-only or text-level concatenation baselines across TREC-DL, LoTTE, and BEIR.
Sixia Sun, Mihai Surdeanu· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
