Pith. sign in

REVIEW 2 major objections 4 minor 48 references

Question Decomposition for Retrieval-Augmented Generation

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Pairing an off-the-shelf cross-encoder reranker with LLM-driven question decomposition bridges the retrieval gap on multi-hop questions, without extra training or specialized indexing.

desk verdict Useful drop-in RAG configuration, honestly reported, but the decomposition-specific benefit is confounded by candidate-pool size. read the letter →

arxiv 2507.00355 v1 pith:KTRUZX56 submitted 2025-07-01 cs.CL

Paul J. L. Ammann , Jonas Golde , Alan Akbik This is my paper · ORCID
classification cs.CL
keywords retrieval-augmentedgenerationquestiondecompositionmulti-hopansweringrerankingcross-encoderdenseretrievalzero-shotRAG
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that ordinary retrieval-augmented generation (RAG) struggles with multi-hop questions because the facts needed to answer them live in different documents, and a single query embedding rarely lands on all of them. The proposed fix is a pipeline that first has an instruction-tuned LLM decompose the original question into several fact-seeking subquestions, retrieves passages for each, merges and deduplicates the candidates, and then reranks the pool against the original question with an off-the-shelf cross-encoder. On MultiHop-RAG this combination raises Hits@10 from 78.1% to 87.2% and MRR@10 from 0.464 to 0.635, and on HotpotQA it produces the best answer exact-match and F1 scores among the four configurations tested, with no fine-tuning or specialized indexing. If the finding holds, any RAG system facing multi-hop or multi-entity questions can adopt this as a drop-in enhancement.

What carries the argument

The central mechanism is a two-stage retrieval pipeline built on top of standard dense retrieval. In the first stage, an instruction-tuned LLM with a fixed prompt generates up to five fact-seeking subquestions per original question; dense retrieval is run for each subquestion and for the original query, and the results are merged and deduplicated. In the second stage, a cross-encoder reranker scores every merged candidate against the original question and returns the top-k passages for generation. Question decomposition is the coverage-expanding stage, and reranking is the precision-restoring stage; the original query is always retained in the retrieval set to preserve baseline performance.

What would settle it

On a multi-hop dataset with gold evidence annotations, compare Hits@10 and MRR@10 of (i) single-query retrieval with reranking (retrieve 2k passages, rerank to k) and (ii) question decomposition plus reranking (retrieve k per subquery, merge, rerank to k). If run (i) matches or exceeds run (ii) on the same queries and passage budget, the claim that decomposition expands coverage is falsified, because the reranker alone would then be doing all the work.

Watch

Extended reading notes

Core claim

The central claim is that the retrieval gap in multi-hop RAG is bridgeable by pairing LLM-driven question decomposition with a cross-encoder reranker, without any extra training or specialized indexing. Decomposition expands the candidate pool so that each subquestion targets a separate document holding a needed fact; reranking then scores every merged candidate against the original query and keeps only the top-k, restoring precision. The paper's QD+RR system reaches 87.2% Hits@10 and 0.635 MRR@10 on MultiHop-RAG, beating the strongest reported baselines from the dataset paper, and it achieves the best answer EM (28.1) and F1 (35.0) on HotpotQA among naive RAG, QD-only, and reranker-only variants. The paper takes this as evidence that coverage from decomposition and precision from reranking are complementary, not redundant.

Load-bearing premise

The whole method depends on the assumption that a fixed prompt makes an instruction-tuned LLM produce fact-seeking subquestions whose independently retrieved passages, after reranking against the original question, retrieve the gold evidence better than the original query alone.

Editorial extensions

If this is right

  • RAG systems operating on multi-hop or multi-entity questions can add this two-module pipeline as a drop-in enhancement, gaining retrieval recall and ranking quality without retraining or reindexing.
  • Because both modules are off-the-shelf, the method is model-agnostic: any instruction-following LLM can generate subquestions, and any cross-encoder can rerank.
  • The ablation shows that decomposition alone yields only small gains, so the reranker is the component that turns broader coverage into usable precision.
  • On MultiHop-RAG, QD+RR outperforms the dataset authors' strongest reranked baselines even with a smaller embedding model, suggesting the method is not tied to a specific retriever.
  • Because the LLM almost always emits exactly five subquestions, the effective behavior is a fixed subquery budget; changing that budget is a direct lever that trades latency against coverage.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The near-zero correlation between subquery count and gold evidence count suggests the decomposition LLM is producing a diverse cover rather than counting hops; this predicts that maximizing subquery diversity, rather than number, would be the better objective for the decomposition prompt.
  • The paper does not test adaptive decomposition, but its own data suggest a fixed five-subquery budget is wasteful on HotpotQA, where most questions need only two supporting facts; a gating rule that decomposes only when the query is predicted to be multi-hop could retain most of the gain at a fraction of the latency.
  • Caching subqueries per raw query, as the paper briefly proposes, makes the amortized latency of QD+RR close to plain RAG, turning the primary overhead into a one-time storage cost.
  • The method's gains are demonstrated on English Wikipedia-style corpora; a natural test is whether the same pipeline holds when the reranker is mismatched with the domain, a failure mode the paper itself flags as a limitation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper proposes a retrieval-augmented generation pipeline that adds an LLM-based question decomposition (QD) module and an off-the-shelf cross-encoder reranker (RR) to standard dense retrieval. For a multi-hop query, the LLM generates up to five subqueries; top-k passages are retrieved for the original query and each subquery, merged, deduplicated, and reranked against the original query before the top-k passages are passed to the generator. The authors evaluate four system variants (Naive RAG, +QD, +RR, +QD+RR) on MultiHop-RAG and HotpotQA, reporting that QD+RR achieves the highest Hits@10 (87.2%) and MRR@10 (0.635) on MultiHop-RAG and the highest answer EM/F1 on HotpotQA. The paper also reports latency overhead, subquery-count distributions, and a correlation analysis showing no relation between subquery count and gold evidence count.

Significance. If the reported retrieval gains survive a properly controlled comparison, the paper offers a practical and modular enhancement for multi-hop RAG: it requires no fine-tuning, uses only an instruction-tuned LLM for decomposition and a standard reranker, and the authors release their code. The four-way internal ablation is a useful design choice, and using exact FAISS inner-product search removes approximate-search confounds. The central scientific claim, that decomposition-specific subqueries assemble complementary evidence that reranking then distills, is plausible but is not yet cleanly established because the main ablation does not hold the candidate-pool size fixed, and the HotpotQA supporting-fact metrics partially contradict the evidence-selection narrative. The paper is an incremental but potentially useful empirical contribution if these issues are resolved.

major comments (2)
  1. [Section 5.1, Table 1; Section 4.2] The headline comparison between +RR and +QD+RR does not control for candidate-pool size. As defined in Section 4.2, +RR retrieves top-2k = 20 passages for the original query and reranks them, while +QD+RR merges top-10 passages for the original query plus top-10 passages for each subquery. Table 3 shows that the LLM emits exactly five subqueries in 93.3% of MultiHop-RAG cases and 98.6% of HotpotQA cases, so the QD+RR pool is typically 60 candidates, three times larger than the RR pool. A reranker over a larger candidate pool can improve Hits@10 and MRR@10 even if the additional candidates are generic paraphrases of the original query, so the specific contribution of decomposition-generated subqueries is not isolated. Please add a controlled condition, for example reranking top-60 passages retrieved with the original query alone, or reducing the per-subquery cutoff so the merged pool matches 20 candidates, and report significance or bootstrap intervals across multiple seeds.
  2. [Section 5.2, Table 2] The claim that QD+RR leads to better evidence-grounded answers is weakened by the supporting-fact and joint metrics on HotpotQA. QD+RR has lower supporting-fact EM and F1 than RR alone (17.9/11.2 vs. 19.6/12.9) and lower joint EM and F1 (8.9/5.8 vs. 9.5/6.4), despite having higher supporting-fact precision (46.8 vs. 44.9). Since the paper's narrative is that decomposition improves coverage and reranking restores precision, the evidence-identification outcome on HotpotQA does not consistently support that narrative. The paper should explicitly discuss this pattern, and should report supporting-fact recall, which is listed in Section 4.3 but omitted from Table 2; without recall it is impossible to tell whether the higher precision comes at the cost of coverage.
minor comments (4)
  1. [Section 5.1] The comparison against text-ada-002 and voyage-02 baselines from Tang and Yang (2024) uses different embedding models and retrieval settings, so the sentence 'our QD+RR thus improves Hits@10 by 16.5% and MRR@10 by 8.4%' is a cross-paper comparison rather than a controlled head-to-head gain; please label it accordingly.
  2. [Algorithm 1] Algorithm 1 calls DECOMPOSE(q0) while Section 3.1 defines DECOMPOSE(q, p); the argument q0 is not defined. Please align the pseudocode with the notation in the text.
  3. [Section 5.4] The claim that 'once decomposed, subqueries can be reused ... so that the latency remains identical to the baseline' depends on an unspecified cache-population assumption; on a cold cache the reported 18.9s/query latency applies, and this tradeoff should be stated more carefully.
  4. [Table 5] Table 5 labels columns 'Pearson (p)' and 'Spearman (p)' but the values shown appear to be the correlation coefficient with a p-value in parentheses; please rename the columns to avoid ambiguity.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the QD+RR pipeline is an empirical system benchmarked on external datasets, with no fitted parameter renamed as a prediction and only minor non-load-bearing self-citations.

full rationale

The paper's central claim is that combining question decomposition with an off-the-shelf reranker improves multi-hop retrieval and answer accuracy. This claim is supported by experiments on MultiHop-RAG and HotpotQA, using fixed public models (bge-large-en-v1.5, bge-reranker-large, Qwen2.5-32B-Instruct) and fixed hyperparameters (k=10, temperature 0.8). No parameter of the pipeline is fitted to the target benchmark labels, and no evaluation metric is defined in terms of the method's own outputs. The retrieval equations R(q) and the reranking score g_phi(q,d) are standard definitions from prior work, not derived from the reported Hits@10 or MRR@10 numbers. The QD module is an LLM call with a fixed prompt; the paper explicitly reports that the LLM almost always emits the maximum of five subqueries, which is a behavioral observation, not a fitted parameter. The self-citations to Golde et al. (2023, 2024) appear only in related work as contextual examples of LLM-based data generation and NER limitations; they are not load-bearing for the main retrieval result, and no uniqueness theorem or prior claim is invoked to forbid alternative approaches. The reviewer-flagged confound that QD+RR reranks up to roughly 60 candidates while RR reranks only 20 is an experimental-design concern about isolating the causal effect of decomposition, but it is not a case where the paper's prediction reduces by construction to its inputs. Consequently, no circular step meets the required evidence standard, and the modest self-citation presence does not affect the validity of the empirical evaluation.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the quality of three off-the-shelf components (dense retriever, cross-encoder reranker, and instruction-tuned LLM decomposer) plus two hand-set hyperparameters (k=10 and a five-subquery prompt cap). No new entities are postulated, and no constants are fitted to the evaluation targets. The main uncharged premises are that the chosen embedder and reranker are adequate for both benchmarks and that gold evidence annotations are a faithful measure of retrieval success.

free parameters (3)
  • k (number of retrieved/reranked passages) = 10
    Fixed in Section 4.5 for both datasets; all Hits@k, MAP@10, MRR@10 results depend on this cutoff.
  • Subquery budget (LLM prompt cap) = 5
    Section 5.3 instructs the LLM to generate at most 5 subqueries, and 93.3% (MultiHop-RAG) and 98.6% (HotpotQA) of queries receive exactly 5, making the effective policy a fixed budget rather than an adaptive count.
  • Sampling temperature and top-p for LLM calls = 0.8 and 0.8
    Section 4.5 sets these for both subquery generation and answer synthesis; they influence decomposition quality and answer variability but are not tuned per dataset.
assumptions (5)
  • domain assumption bge-large-en-v1.5 dense embeddings are an adequate base retriever for both original queries and subqueries.
    Section 4.4 selects this embedder as the retrieval backbone; all gains are conditional on its dense retrieval quality.
  • domain assumption bge-reranker-large cross-encoder reliably scores candidate passages by relevance to the original query.
    Algorithm 1 and Section 4.4 use this reranker to filter the merged candidate pool; if the reranker is mismatched to the domain, the authors note the benefits may vanish.
  • domain assumption An instruction-tuned LLM can decompose questions into useful fact-seeking subquestions without task-specific training.
    Section 3.1 defines DECOMPOSE; the Limitations section explicitly lists prompt and model sensitivity as a limitation.
  • domain assumption Gold evidence annotations in MultiHop-RAG and HotpotQA are complete and correctly define retrieval success.
    Section 4.1 and Section 4.3 compute Hits@k, MAP, MRR, and supporting-fact metrics against these annotations; incomplete gold evidence would create false negatives and distort metric levels.
  • domain assumption Dev-split evaluation is representative of hidden test behavior for these benchmarks.
    The paper evaluates on dev splits because gold test labels are not public (Section 4 and Section 5.2 footnote); this is standard when no training is performed but remains an assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Question Decomposition for Retrieval-Augmented Generation." pith.science (2026). https://pith.science/paper/KTRUZX56

@misc{pith2026250700355,
  author       = {Pith},
  title        = {Pith review of: Question Decomposition for Retrieval-Augmented Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KTRUZX56}},
  note         = {Machine review of arXiv:2507.00355}
}
read the original abstract

Grounding large language models (LLMs) in verifiable external sources is a well-established strategy for generating reliable answers. Retrieval-augmented generation (RAG) is one such approach, particularly effective for tasks like question answering: it retrieves passages that are semantically related to the question and then conditions the model on this evidence. However, multi-hop questions, such as "Which company among NVIDIA, Apple, and Google made the biggest profit in 2023?," challenge RAG because relevant facts are often distributed across multiple documents rather than co-occurring in one source, making it difficult for standard RAG to retrieve sufficient information. To address this, we propose a RAG pipeline that incorporates question decomposition: (i) an LLM decomposes the original query into sub-questions, (ii) passages are retrieved for each sub-question, and (iii) the merged candidate pool is reranked to improve the coverage and precision of the retrieved evidence. We show that question decomposition effectively assembles complementary documents, while reranking reduces noise and promotes the most relevant passages before answer generation. Although reranking itself is standard, we show that pairing an off-the-shelf cross-encoder reranker with LLM-driven question decomposition bridges the retrieval gap on multi-hop questions and provides a practical, drop-in enhancement, without any extra training or specialized indexing. We evaluate our approach on the MultiHop-RAG and HotpotQA, showing gains in retrieval (MRR@10: +36.7%) and answer accuracy (F1: +11.6%) over standard RAG baselines.

Figures

Figures reproduced from arXiv: 2507.00355 by the authors.

Figure 1
Figure 1. (a) Standard retrieval in RAG versus (b) our [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Absolute counts of gold evidences (blue) vs. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 15 canonical work pages

  1. [1]

    Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. 2023. https://doi.org/10.48550/arXiv.2310.11511 Self- RAG : Learning to Retrieve , Generate , and Critique through Self-Reflection . Preprint, arXiv:2310.11511

    arXiv doi:10.48550/arxiv.2310.11511 2023

  2. [2]

    Chi-Min Chan, Chunpu Xu, Ruibin Yuan, Hongyin Luo, Wei Xue, Yike Guo, and Jie Fu. 2024. https://doi.org/10.48550/arXiv.2404.00610 RQ-RAG : Learning to Refine Queries for Retrieval Augmented Generation . Preprint, arXiv:2404.00610

    arXiv doi:10.48550/arxiv.2404.00610 2024

  3. [3]

    Sukmin Cho, Jeongyeon Seo, Soyeong Jeong, and Jong C. Park. 2023. https://arxiv.org/abs/2310.17490 Improving Zero-shot Reader by Reducing Distractions from Irrelevant Documents in Open-Domain Question Answering . Preprint, arXiv:2310.17490

    work page Pith review arXiv 2023

  4. [4]

    Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazar \'e , Maria Lomeli, Lucas Hosseini, and Herv \'e J \'e gou. 2024. https://doi.org/10.48550/arXiv.2401.08281 The Faiss library . Preprint, arXiv:2401.08281

    arXiv doi:10.48550/arxiv.2401.08281 2024

  5. [5]

    Facebookresearch. 2024. Faiss indexes. https://github.com/facebookresearch/faiss/wiki/ Faiss-indexes

    work page 2024

  6. [6]

    Barah Fazili, Koustava Goswami, Natwar Modani, and Inderjeet Nair. 2024. https://doi.org/10.48550/arXiv.2407.10245 GenSco : Can Question Decomposition based Passage Alignment improve Question Answering ? Preprint, arXiv:2407.10245

    work page Pith review arXiv doi:10.48550/arxiv.2407.10245 2024

  7. [7]

    Yair Feldman and Ran El-Yaniv . 2019. https://doi.org/10.48550/arXiv.1906.06606 Multi- Hop Paragraph Retrieval for Open-Domain Question Answering . Preprint, arXiv:1906.06606

    work page Pith review arXiv doi:10.48550/arxiv.1906.06606 2019

  8. [8]

    Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2022. https://doi.org/10.48550/arXiv.2212.10496 Precise Zero-Shot Dense Retrieval without Relevance Labels . Preprint, arXiv:2212.10496

    arXiv doi:10.48550/arxiv.2212.10496 2022

Show all 48 references
  1. [9]

    Michael Glass, Gaetano Rossiello, Md Faisal Mahbub Chowdhury, Ankita Naik, Pengshan Cai, and Alfio Gliozzo. 2022. https://doi.org/10.18653/v1/2022.naacl-main.194 Re2G : Retrieve , Rerank , Generate . In Proceedings of the 2022 Conference of the North American Chapter of the As...

    2022 doi

  2. [10]

    Jonas Golde, Patrick Haller, Felix Hamborg, Julian Risch, and Alan Akbik. 2023. https://doi.org/10.18653/v1/2023.emnlp-demo.1 Fabricator: An open source toolkit for generating labeled training data with teacher LLM s . In Proceedings of the 2023 Conference on Empirical Methods...

    2023 doi

  3. [11]

    Jonas Golde, Felix Hamborg, and Alan Akbik. 2024. https://aclanthology.org/2024.eacl-long.178/ Large-scale label interpretation learning for few-shot named entity recognition . In Proceedings of the 18th Conference of the European Chapter of the Association for Computational L...

    2024

  4. [12]

    Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, and Jong C. Park. 2024. https://doi.org/10.48550/arXiv.2403.14403 Adaptive- RAG : Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity . Preprint, arXiv:2403.14403

    2024 doi arXiv

  5. [13]

    Weld, and Luke Zettlemoyer

    Mandar Joshi, Eunsol Choi, Daniel S. Weld, and Luke Zettlemoyer. 2017. https://doi.org/10.48550/arXiv.1705.03551 TriviaQA : A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension . Preprint, arXiv:1705.03551

    2017 doi arXiv

  6. [14]

    Vladimir Karpukhin, Barlas O g uz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. https://arxiv.org/abs/2004.04906 Dense Passage Retrieval for Open-Domain Question Answering . Preprint, arXiv:2004.04906

    2020 arXiv

  7. [15]

    LangChain. 2025. LangChain . https://www.langchain.com/

    2025

  8. [16]

    u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K \"u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \"a schel, Sebastian Riedel, and Douwe Kiela. 2020. https://arxiv.org/abs/2005.11401 Retrieval- Augmented Generation for Know...

    2020 arXiv

  9. [17]

    Kunze Li and Yu Zhang. 2024. https://doi.org/10.18653/v1/2024.findings-acl.280 Planning first, question second: An LLM -guided method for controllable question generation . In Findings of the Association for Computational Linguistics: ACL 2024, pages 4715--4729, Bangkok, Thail...

    2024 doi

  10. [18]

    Xinbei Ma, Yeyun Gong, Pengcheng He, Hai Zhao, and Nan Duan. 2023. https://arxiv.org/abs/2305.14283 Query Rewriting for Retrieval-Augmented Large Language Models . Preprint, arXiv:2305.14283

    2023 arXiv

  11. [19]

    Vaibhav Mavi, Anubhav Jangra, and Adam Jatowt. 2024. https://doi.org/10.48550/arXiv.2204.09140 Multi-hop Question Answering . Preprint, arXiv:2204.09140

    2024 doi arXiv

  12. [20]

    Sewon Min, Victor Zhong, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2019. https://doi.org/10.48550/arXiv.1906.02916 Multi-hop Reading Comprehension through Question Decomposition and Rescoring . Preprint, arXiv:1906.02916

    2019 doi arXiv

  13. [21]

    Rodrigo Nogueira and Kyunghyun Cho. 2020. https://arxiv.org/abs/1901.04085 Passage Re-ranking with BERT . Technical Report arXiv:1901.04085, arXiv

    2020 arXiv

  14. [22]

    OpenAI . 2022. New and improved embedding model. https://openai.com/index/new-and-improved-embedding-model/

    2022

  15. [23]

    Ethan Perez, Patrick Lewis, Wen-tau Yih, Kyunghyun Cho, and Douwe Kiela. 2020. https://doi.org/10.48550/arXiv.2002.09758 Unsupervised Question Decomposition for Question Answering . Preprint, arXiv:2002.09758

    2020 doi arXiv

  16. [24]

    Smith, and Mike Lewis

    Ofir Press, Muru Zhang, Sewon Min, Ludwig Schmidt, Noah A. Smith, and Mike Lewis. 2023. https://doi.org/10.48550/arXiv.2210.03350 Measuring and Narrowing the Compositionality Gap in Language Models . Preprint, arXiv:2210.03350

    2023 doi arXiv

  17. [25]

    Qwen Team . 2024. Qwen2.5: A Party of Foundation Models ! https://qwenlm.github.io/blog/qwen2.5/

    2024

  18. [26]

    Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. https://doi.org/10.48550/arXiv.1606.05250 SQuAD : 100,000+ Questions for Machine Comprehension of Text . Preprint, arXiv:1606.05250

    2016 doi arXiv

  19. [27]

    Nils Reimers and Iryna Gurevych. 2019. https://arxiv.org/abs/1908.10084 Sentence- BERT : Sentence Embeddings using Siamese BERT-Networks . Technical Report arXiv:1908.10084, arXiv

    2019 arXiv

  20. [28]

    Chaudhary, Kalpa Gunaratna, Srinivasan Parthasarathy, and Manas Gaur

    Yash Saxena, Ankur Padia, Mandar S. Chaudhary, Kalpa Gunaratna, Srinivasan Parthasarathy, and Manas Gaur. 2025. https://doi.org/10.48550/arXiv.2505.16014 Ranking Free RAG : Replacing Re-ranking with Selection in RAG for Sensitive Domains . Preprint, arXiv:2505.16014

    2025 doi arXiv

  21. [29]

    Zhihong Shao, Yeyun Gong, Yelong Shen, Minlie Huang, Nan Duan, and Weizhu Chen. 2023. https://doi.org/10.48550/arXiv.2305.15294 Enhancing Retrieval-Augmented Large Language Models with Iterative Retrieval-Generation Synergy . Preprint, arXiv:2305.15294

    2023 doi arXiv

  22. [30]

    Freda Shi, Xinyun Chen, Kanishka Misra, Nathan Scales, David Dohan, Ed Chi, Nathanael Sch \"a rli, and Denny Zhou. 2023. https://doi.org/10.48550/arXiv.2302.00093 Large Language Models Can Be Easily Distracted by Irrelevant Context . Preprint, arXiv:2302.00093

    2023 doi arXiv

  23. [31]

    Krishna Srinivasan, Karthik Raman, Anupam Samanta, Lingrui Liao, Luca Bertelli, and Mike Bendersky. 2022. https://doi.org/10.48550/arXiv.2210.15718 QUILL : Query Intent with Large Language Models using Retrieval Augmentation and Multi-stage Distillation . Preprint, arXiv:2210.15718

    2022 doi arXiv

  24. [32]

    Yixuan Tang and Yi Yang. 2024. https://doi.org/10.48550/arXiv.2401.15391 MultiHop-RAG : Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries . Preprint, arXiv:2401.15391

    2024 doi arXiv

  25. [33]

    Voyage AI Innovations Inc. 2024. Voyage AI Home . https://www.voyageai.com/

    2024

  26. [34]

    Shuting Wang, Xin Yu, Mang Wang, Weipeng Chen, Yutao Zhu, and Zhicheng Dou. 2024 a . https://arxiv.org/abs/2406.12566 RichRAG : Crafting Rich Responses for Multi-faceted Queries in Retrieval-Augmented Generation . Preprint, arXiv:2406.12566

    2024 arXiv

  27. [35]

    Xiaohua Wang, Zhenghua Wang, Xuan Gao, Feiran Zhang, Yixin Wu, Zhibo Xu, Tianyuan Shi, Zhengyuan Wang, Shizheng Li, Qi Qian, Ruicheng Yin, Changze Lv, Xiaoqing Zheng, and Xuanjing Huang. 2024 b . https://doi.org/10.18653/v1/2024.emnlp-main.981 Searching for Best Practices in R...

    2024 doi

  28. [36]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, R \'e mi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, ...

    2020 doi arXiv

  29. [37]

    Karlsson, and Manabu Okumura

    Jian Wu, Linyi Yang, Yuliang Ji, Wenhao Huang, B \"o rje F. Karlsson, and Manabu Okumura. 2024. https://doi.org/10.48550/arXiv.2402.11166 GenDec : A robust generative Question-decomposition method for Multi-hop reasoning . Preprint, arXiv:2402.11166

    2024 doi arXiv

  30. [38]

    Shitao Xiao, Zheng Liu, Peitian Zhang, Niklas Muennighoff, Defu Lian, and Jian-Yun Nie. 2023. https://doi.org/10.48550/arXiv.2309.07597 C- Pack : Packed Resources For General Chinese Embeddings . Preprint, arXiv:2309.07597

    2023 doi arXiv

  31. [39]

    Shicheng Xu, Liang Pang, Huawei Shen, Xueqi Cheng, and Tat-Seng Chua. 2024. https://doi.org/10.48550/arXiv.2304.14732 Search-in-the- Chain : Interactively Enhancing Large Language Models with Search for Knowledge-intensive Tasks . Preprint, arXiv:2304.14732

    2024 doi arXiv

  32. [40]

    Kota Yamaguchi. 2025. Faiss-cpu: A library for efficient similarity search and clustering of dense vectors

    2025

  33. [41]

    Shi-Qi Yan, Jia-Chen Gu, Yun Zhu, and Zhen-Hua Ling. 2024. https://arxiv.org/abs/2401.15884 Corrective Retrieval Augmented Generation . Technical Report arXiv:2401.15884, arXiv

    2024 arXiv

  34. [42]

    An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, and 43 others. 2024. https://doi.org/10.4...

    2024 doi arXiv

  35. [43]

    Cohen, Ruslan Salakhutdinov, and Christopher D

    Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W. Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. https://arxiv.org/abs/1809.09600 HotpotQA : A Dataset for Diverse , Explainable Multi-hop Question Answering . Preprint, arXiv:1809.09600

    2018 arXiv

  36. [44]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. https://doi.org/10.48550/arXiv.2210.03629 ReAct : Synergizing Reasoning and Acting in Language Models . Preprint, arXiv:2210.03629

    2023 doi arXiv

  37. [45]

    Ben Zhou, Kyle Richardson, Xiaodong Yu, and Dan Roth. 2022. https://doi.org/10.48550/arXiv.2210.16865 Learning to Decompose : Hypothetical Question Decomposition Based on Comparable Texts . Preprint, arXiv:2210.16865

    2022 doi arXiv

  38. [46]

    Rongzhi Zhu, Xiangyu Liu, Zequn Sun, Yiwei Wang, and Wei Hu. 2025. https://arxiv.org/abs/2502.14245 Mitigating lost-in-retrieval problems in retrieval augmented multi-hop question answering . Preprint, arXiv:2502.14245

    2025 arXiv

  39. [47]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  40. [48]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.