Pith. sign in

REVIEW 3 major objections 5 minor 37 references

Momentum Posterior Regularization for Multi-hop Dense Retrieval

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read MoPo makes posterior knowledge distillation work for multi-hop dense retrieval by treating each hop's query-focused summary as the posterior signal and maintaining the posterior teacher as a momentum average of the student.

desk verdict Useful empirical work on multi-hop dense retrieval with a new dataset and a simple training trick, but the posterior-distillation story is not proven: the teacher is an EMA of the student, so the gains may come from regularization. read the letter →

arxiv 2502.20399 v1 pith:6AJ73ZY7 submitted 2024-12-17 cs.CL cs.LG

Zehua Xia , Yuyang Wu , Yiyun Xia , Cam-Tu Nguyen This is my paper · ORCID
classification cs.CLcs.LG
keywords multi-hopdenseretrievalposteriorregularizationmomentummovingaverageknowledgedistillationquery-focusedsummaryPostSumQAHotpotStrategy
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

MoPo is a training method for multi-hop dense retrieval that moves posterior knowledge into the retriever without changing how retrieval runs at inference time. The paper's claim is that the right posterior information for hop $t$ is not the final answer but a short query-focused summary of the golden documents seen so far, and that a teacher model maintained as a momentum average of the student makes distillation stable. On HotpotQA and the held-out StrategyQA, MoPo reports exact-match retrieval at 20 candidates rising from 72.65 to 76.74 and from 32.92 to 35.61 respectively, while the downstream retrieval-reranking-generation pipeline reports joint EM of 46.7 on dev and 45.7 on test. The result is a stronger multi-hop retriever at no extra inference cost.

What carries the argument

Momentum Posterior Regularization (MoPo): the posterior model $M_\phi$ is never trained on posterior summaries directly; instead its parameters are a momentum moving average of the prior student $\theta$, updated at each step by $\phi \leftarrow m\phi + (1-m)\theta$ (Eq. 9). It produces the posterior distribution $p_\phi(d_t|s_t)$ over documents given the hop summary, which is distilled into the student through the KL term in Eq. 8, alongside the InfoNCE retrieval loss. The other load-bearing piece is PostSumQA, a 22,696-item dataset of per-hop query-focused summaries built from HotpotQA by backward summary generation: the last-hop summary is produced by QA2D from question and answer, and earlier summaries are generated by an LLM using the next-hop summary as look-ahead.

What would settle it

Run the MoPo training objective with an oracle posterior: replace the model-produced summaries with gold summaries, and compare a variant trained with and without the KL term. If exact-match retrieval is statistically identical, the KL term is not transferring posterior knowledge and the reported improvements come from the summary-based query reformulation or the momentum smoothing rather than from the claimed distillation.

Watch

Extended reading notes

Core claim

The paper argues that previous attempts to distill posterior information into retrieval fail in multi-hop settings for two reasons: the answer is too far from the query to anchor intermediate hops, and the gap between prior and posterior retrievers is too large for two-stage distillation. MoPo replaces the answer as posterior signal with a query-focused summary of the gold knowledge at the previous and current hops, and replaces the separately trained teacher with one that is updated as a momentum moving average of the student, $\phi \leftarrow m\phi + (1-m)\theta$. The training objective combines the InfoNCE retrieval loss with a KL divergence between the posterior distribution $p_\phi(d_t|s_t)$ and the prior distribution $p_\theta(d_t|q_t)$ at every hop. The paper reports that this one-stage, smoother distillation yields better exact-match retrieval than both the multi-hop dense retrieval baselines and the two-stage posterior-regularization baselines, and that the trained retriever transfers to StrategyQA without further fine-tuning.

Load-bearing premise

The momentum-averaged teacher is never supervised on the posterior summaries, so the entire method assumes it still produces a meaningful posterior distribution to distil; if that distribution is not trustworthy, the KL regularizer is only a self-smoothing term and the reported gains do not demonstrate posterior knowledge transfer.

Editorial extensions

If this is right

  • Posterior information for intermediate hops should be the hop's own query-focused summary, not the final answer, because the answer does not point back to the missing bridge evidence.
  • A momentum-averaged teacher makes posterior regularization stable enough to train in one stage, while a fixed two-stage teacher makes the KL term dominate the loss and hurts retrieval.
  • MoPo's gains appear specifically in exact-match (EM) metrics, the metric that reflects whether the full golden chain is retrieved, on both HotpotQA and StrategyQA.
  • The trained retriever adds no inference-time cost relative to the summary-based prior retrieval, so downstream reranking and QA pipelines inherit the improvement for free.
  • The MoPo-based retrieval-reranking-generation pipeline reports higher joint EM/F1 than the compared multi-hop reranking baselines and LLM-reasoning baselines on HotpotQA.

Reading between the lines

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

  • If momentum smoothing is the active ingredient, the same update rule should improve other distillation setups with a large teacher-student gap, such as answer-based posterior distillation in single-hop dense retrieval.
  • The backward summary-generation recipe could be lifted to three-hop or other multi-hop benchmarks: it generates each summary from the next summary as look-ahead, so the recursion should extend as long as a reliable next-hop summary exists.
  • A sharper test of the posterior-distillation story would be to use an oracle summary at inference time and compare MoPo's student with the same student trained without the KL term; if EM is unchanged, the regularization is acting as a stabilizer rather than as a carrier of posterior knowledge.
  • Since the paper found the answer-based posterior signal hurts, replacing the final-answer posterior with per-hop summaries may also benefit knowledge-grounded dialogue, where posterior summaries at each turn could replace the single response posterior.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes MoPo, a training method for multi-hop dense retrieval in which a query-focused summary of the golden previous and current hops is used as posterior information for each retrieval step. A 'posterior' retriever M_phi is updated as a momentum average of the prior retriever M_theta, and the prior is trained with an InfoNCE loss plus a KL term that aligns p_theta(d_t|q_t) with p_phi(d_t|s_t). The authors construct a dataset, PostSumQA, using backward summary generation with QA2D and GPT-3.5, and evaluate on HotpotQA and held-out StrategyQA, reporting improvements over MDR and posterior-regularization baselines in retrieval recall/EM and in downstream reranking and QA.

Significance. If the proposed mechanism is what drives the gains, MoPo is a simple and inexpensive improvement to multi-hop dense retrieval: it adds no inference cost over MDRsum and yields consistent EM gains on HotpotQA and StrategyQA, including a held-out setting. The paper also contributes PostSumQA and a backward summary-generation procedure, both of potential reuse. The strengths include the held-out StrategyQA evaluation, comparisons against several MDR variants and two posterior-regularization baselines, ablations over lambda and m, and an additional base-model experiment in the appendix. The main weakness is that the 'posterior' teacher is not trained on posterior summaries, so the central distillation claim is not yet isolated from EMA-style self-regularization.

major comments (3)
  1. [3.4 / Algorithm 1 / Eq. (8)-(9)] The posterior teacher M_phi is updated only by phi <- m*phi + (1-m)*theta (Eq. 9) and is never trained on the posterior summaries Sseq. Consequently, p_phi(d_t|s_t) is not established as a posterior distribution: the model parameters have only been optimized on the prior query q_t, not on the posterior summary s_t. The KL term in Eq. (8) may therefore be performing EMA-style self-regularization rather than posterior knowledge distillation. The ablations in Table 3 (m=0, m=0.5, m=0.9, m=0.99, m=1) change both the EMA dynamics and the teacher input, so they do not separate these explanations. I request an additional control in which the teacher input is q_t instead of s_t while keeping the same EMA update and KL weight; this would directly test whether the posterior summary input is the source of the gains. Without it, the central claim that MoPo 'distills posterior knowledge' is not supported.
  2. [5.1 / Tables 1-3 and 5] The paper repeatedly states that MoPo 'significantly outperforms' baselines, but every table reports a single run without variance or significance tests. Some margins are small, e.g. R@100 on HotpotQA is 96.70 for MoPo versus 96.38 for MDRsum, and the R@20 differences between PRdyn and MoPo are about one point; the EM margins are larger, but they still need multiple seeds or a paired significance test to support the claim of significance. Please report means and standard deviations over at least three training runs, or perform paired tests on the dev/test sets, for at least the key comparisons in Table 1 and Table 3.
  3. [5.1 / B.1 / C.2] The base retriever is described inconsistently. The main text says MoPo is initialized with E5-v2-base (Section 5.1), while Appendix B.1 says 'We use E5-base' and states that its pre-training data does not include HotpotQA, and Appendix C.2 explicitly distinguishes e5-base from e5-base-v2 and reports that e5-base-v2 includes HotpotQA in pre-training. Since Table C4 shows that the choice changes R@20 from 68.52 to 75.92 in the zero-shot MDR framework, the exact checkpoint and its pre-training corpus must be stated precisely for the results in Table 1 to be reproducible.
minor comments (5)
  1. [Eq. (5)] In Eq. (5), the summary recursion is written as st-1 = Gs(st-2, dt-1, q), but at t=2 the term s0 is not defined; please clarify the base case and the exact inputs to the summary generator.
  2. [Algorithm 1] The notation M_phi(q, Sseq, Dseq) is inconsistent with Eq. (7), where M_phi conditions on s_t and d; align the pseudocode with the equations.
  3. [Table 2] The unit 'PG(‰)' is unclear; if this is percent change, use '%' and define the performance-gap formula explicitly.
  4. [6.2 / Table 6] Table 6 reports F1 on only 100 HotpotQA instances and gives no confidence intervals or significance tests; the statement that MoPo 'outperforms all baselines' on that subset should be softened accordingly.
  5. [6.2 / Table 5] The text says MoPo 'surpasses the revised MDR by a margin of 5.2%', but the corresponding values in Table 5 are 72.7 vs. 67.5 in Joint F1, a 5.2-point difference; please state the comparison as absolute points.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MoPo's empirical claims are grounded in held-out dev/test evaluations against published baselines, and the momentum teacher update (Eq. 9) is a design choice rather than a fitted prediction of the reported metrics.

full rationale

The paper's central claim is empirical: MoPo improves retrieval EM over MDR-family and PR baselines on HotpotQA and StrategyQA (Table 1) and downstream QA (Tables 4-6). These numbers are measured on held-out sets and compared with externally published baselines, so the reported gains are not fitted to the result. The training objective (Eq. 8) is not equivalent to the evaluation metric by construction: the KL term regularizes the student toward an EMA-smoothed teacher distribution conditioned on posterior summaries, and no model parameter is chosen so that the reported EM values follow directly. The main methodological concern—that M_phi is updated only by EMA (Eq. 9) and never supervised on posterior summaries—is a question of mechanism identification/confounding (the gains might reflect self-distillation or regularization rather than posterior knowledge), not circularity: the teacher is not fitted to the test labels, and the posterior input s_t is a genuinely different conditioning signal from the student's q_t. The only self-citation (Xia et al., 2023, Appendix A.1, used to support QA2D on HotpotQA) is incidental and not load-bearing for the main claims. No step reduces to its own inputs by definition, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central method depends on two tuned hyperparameters (lambda, m) and on several assumptions about summary quality, summary-distribution match, and the validity of an EMA teacher as a posterior model. No new physical or formal entities are introduced; the new artifact is the PostSumQA dataset, which is a training resource rather than a postulated entity.

free parameters (2)
  • lambda (KL regularization weight) = 0.3 for PR baseline; MoPo lambda not explicitly stated in main text, sensitivity range 0.3 to 1.0 explored
    Grid-searched over {0.1, 0.2, ..., 1.0} in Section 5.2 and Table 2; performance varies with lambda, so it is a tuned free parameter.
  • momentum coefficient m = 0.99
    Grid-searched over {0, 0.5, 0.9, 0.99, 1.0} in Section 5.5; Table 3 shows m=0.99 gives the best EM@20 and Recall@100.
assumptions (5)
  • domain assumption Dual-encoder dense retrieval with inner product scoring (f_theta) is suitable for multi-hop relevance, and the E5 initialization is strong enough to support the learned multi-hop queries.
    Used throughout Sections 3.2 and 5.1; Table 1 shows the base encoder choice matters more than the training method for recall, so this assumption carries weight.
  • domain assumption Backward-generated query-focused summaries from QA2D and GPT-3.5 are accurate and do not leak future-hop information into earlier training targets.
    Section 4 and Appendix A.3; only a small human review (Table A3) is presented, and roughly 9 percent of samples are discarded during generation.
  • ad hoc to paper The momentum-averaged model M_phi provides a valid posterior distribution p_phi(d_t|s_t) even though phi is never trained with posterior summaries.
    Section 3.4 and Algorithm 1; this is the enabling assumption for the distillation loss in Eq. 8 and is not directly verified.
  • domain assumption At inference, summaries generated by the fine-tuned Flan-T5 model are distributionally close to the gold summaries used in training.
    Sections 4 and 5.1; the retriever is trained with gold summaries but deployed with generated summaries, and the gap is not evaluated.
  • standard math Standard properties of InfoNCE contrastive loss and KL divergence are assumed without proof.
    Eqs. 4 and 8 use these losses as in the cited literature; no derivation is needed for the empirical claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Momentum Posterior Regularization for Multi-hop Dense Retrieval." pith.science (2026). https://pith.science/paper/6AJ73ZY7

@misc{pith2026250220399,
  author       = {Pith},
  title        = {Pith review of: Momentum Posterior Regularization for Multi-hop Dense Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6AJ73ZY7}},
  note         = {Machine review of arXiv:2502.20399}
}
read the original abstract

Multi-hop question answering (QA) often requires sequential retrieval (multi-hop retrieval), where each hop retrieves missing knowledge based on information from previous hops. To facilitate more effective retrieval, we aim to distill knowledge from a posterior retrieval, which has access to posterior information like an answer, into a prior retrieval used during inference when such information is unavailable. Unfortunately, current methods for knowledge distillation in one-time retrieval are ineffective for multi-hop QA due to two issues: 1) Posterior information is often defined as the response (i.e. the answer), which may not clearly connect to the query without intermediate retrieval; and 2) The large knowledge gap between prior and posterior retrievals makes existing distillation methods unstable, even resulting in performance loss. As such, we propose MoPo (Momentum Posterior Regularization) with two key innovations: 1) Posterior information of one hop is defined as a query-focus summary from the golden knowledge of the previous and current hops; 2) We develop an effective training strategy where the posterior retrieval is updated along with the prior retrieval via momentum moving average method, allowing smoother and effective distillation. Experiments on HotpotQA and StrategyQA demonstrate that MoPo outperforms existing baselines in both retrieval and downstream QA tasks.

Figures

Figures reproduced from arXiv: 2502.20399 by the authors.

Figure 1
Figure 1. An example from HotpotQA benchmark. Given the 2-hop question, an iterative retriever is expected [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Total absolute and relative loss curve of PR [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 8 canonical work pages

  1. [1]

    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 INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Xiuyi Chen, Fandong Meng, Peng Li, Feilong Chen, Shuang Xu, Bo Xu, and Jie Zhou. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.275 Bridging the gap between prior and posterior knowledge selection for knowledge-grounded dialogue generation . In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 3426--3...

    work page doi:10.18653/v1/2020.emnlp-main.275 2020

  4. [4]

    Zheng Chu, Jingchang Chen, Qianglong Chen, Haotian Wang, Kun Zhu, Xiyuan Du, Weijiang Yu, Ming Liu, and Bing Qin. 2024. https://aclanthology.org/2024.acl-long.67 B eam A gg R : Beam aggregation reasoning over multi-source knowledge for multi-hop question answering . In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics...

    work page 2024

  5. [5]

    Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V

    Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Alex Castro-Ros, Marie Pellat, Kevin Robinson, Dasha Valter, Sharan Narang, Gaurav Mishra, Adams Yu, Vincent Zhao, Yanping H...

    arXiv 2022

  6. [6]

    Dorottya Demszky, Kelvin Guu, and Percy Liang. 2018. https://arxiv.org/abs/1809.02922 Transforming question answering datasets into natural language inference datasets . Preprint, arXiv:1809.02922

    arXiv 2018

  7. [7]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. https://doi.org/10.18653/v1/N19-1423 BERT : Pre-training of deep bidirectional transformers for language understanding . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long a...

    doi:10.18653/v1/n19-1423 2019

  8. [8]

    Shaoxiong Feng, Hongshen Chen, Kan Li, and Dawei Yin. 2020. Posterior-gan: Towards informative and coherent response generation with posterior generative adversarial network. Proceedings of the AAAI Conference on Artificial Intelligence, 34:7708--7715

    work page 2020

Show all 37 references
  1. [9]

    Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. 2021. https://doi.org/10.1162/tacl_a_00370 Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies . Transactions of the Association for Computational Lingu...

    2021 doi

  2. [10]

    Michael Günther, Jackmin Ong, Isabelle Mohr, Alaeddine Abdessalem, Tanguy Abel, Mohammad Kalim Akram, Susana Guzman, Georgios Mastrapas, Saba Sturua, Bo Wang, Maximilian Werk, Nan Wang, and Han Xiao. 2023. https://arxiv.org/abs/2310.19923 Jina embeddings 2: 8192-token general-...

    2023 arXiv

  3. [11]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. https://openreview.net/forum?id=nZeVKeeFYf9 Lo RA : Low-rank adaptation of large language models . In International Conference on Learning Representations

    2022

  4. [12]

    Gautier Izacard, Mathilde Caron, Lucas Hosseini, Sebastian Riedel, Piotr Bojanowski, Armand Joulin, and Edouard Grave. 2022. https://arxiv.org/abs/2112.09118 Unsupervised dense information retrieval with contrastive learning . Preprint, arXiv:2112.09118

    2022 arXiv

  5. [13]

    Zhengbao Jiang, Frank Xu, Luyu Gao, Zhiqing Sun, Qian Liu, Jane Dwivedi-Yu, Yiming Yang, Jamie Callan, and Graham Neubig. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.495 Active retrieval augmented generation . In Proceedings of the 2023 Conference on Empirical Methods in...

    2023 doi

  6. [14]

    Jeff Johnson, Matthijs Douze, and Hervé Jégou. 2021. https://doi.org/10.1109/TBDATA.2019.2921572 Billion-scale similarity search with gpus . IEEE Transactions on Big Data, 7(3):535--547

    2021

  7. [15]

    Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.550 Dense passage retrieval for open-domain question answering . In Proceedings of the 2020 Conference on Empiric...

    2020 doi

  8. [16]

    Byeongchang Kim, Jaewoo Ahn, and Gunhee Kim. 2020. https://openreview.net/forum?id=Hke0K1HKwr Sequential Latent Knowledge Selection for Knowledge-Grounded Dialogue . In ICLR

    2020

  9. [17]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. 2015. http://arxiv.org/abs/1412.6980 Adam: A method for stochastic optimization . In 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings

    2015 arXiv

  10. [18]

    Shaobo Li, Xiaoguang Li, Lifeng Shang, Xin Jiang, Qun Liu, Chengjie Sun, Zhenzhou Ji, and Bingquan Liu. 2021. Hopretriever: Retrieve hops over wikipedia to answer complex questions. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 13279--13287

    2021

  11. [19]

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. https://arxiv.org/abs/1907.11692 Roberta: A robustly optimized BERT pretraining approach . CoRR, abs/1907.11692

    2019 arXiv

  12. [20]

    Ilya Loshchilov and Frank Hutter. 2019. https://openreview.net/forum?id=Bkg6RiCqY7 Decoupled weight decay regularization . In International Conference on Learning Representations

    2019

  13. [21]

    Kaixin Ma, Hao Cheng, Yu Zhang, Xiaodong Liu, Eric Nyberg, and Jianfeng Gao. 2023. https://doi.org/10.18653/v1/2023.acl-long.89 Chain-of-skills: A configurable model for open-domain question answering . In Proceedings of the 61st Annual Meeting of the Association for Computati...

    2023 doi

  14. [22]

    Niklas Muennighoff, Nouamane Tazi, Loic Magne, and Nils Reimers. 2023. https://doi.org/10.18653/v1/2023.eacl-main.148 MTEB : Massive text embedding benchmark . In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages...

    2023 doi

  15. [23]

    Ofir Press, Muru Zhang, Sewon Min, Ludwig Schmidt, Noah Smith, and Mike Lewis. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.378 Measuring and narrowing the compositionality gap in language models . In Findings of the Association for Computational Linguistics: EMNLP 20...

    2023 doi

  16. [24]

    Peng Qi, Haejun Lee, Tg Sido, and Christopher Manning. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.292 Answering open-domain questions of varying reasoning steps from text . In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages ...

    2021 doi

  17. [25]

    Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. https://doi.org/10.18653/v1/D16-1264 SQ u AD : 100,000+ questions for machine comprehension of text . In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2383-...

    2016 doi

  18. [26]

    Nandan Thakur, Nils Reimers, Andreas R \"u ckl \'e , Abhishek Srivastava, and Iryna Gurevych. 2021. https://openreview.net/forum?id=wCu6T5xFjeJ BEIR : A heterogeneous benchmark for zero-shot evaluation of information retrieval models . In Thirty-fifth Conference on Neural Info...

    2021

  19. [27]

    Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2023. https://doi.org/10.18653/v1/2023.acl-long.557 Interleaving retrieval with chain-of-thought reasoning for knowledge-intensive multi-step questions . In Proceedings of the 61st Annual Meeting of th...

    2023 doi

  20. [28]

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. 2019. https://arxiv.org/abs/1807.03748 Representation learning with contrastive predictive coding . Preprint, arXiv:1807.03748

    2019 arXiv

  21. [29]

    Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2024. https://arxiv.org/abs/2212.03533 Text embeddings by weakly-supervised contrastive pre-training . Preprint, arXiv:2212.03533

    2024 arXiv

  22. [30]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems 35: Annual Conference on Neur...

    2022

  23. [31]

    Zehua Xia, Qi Gou, Bowen Yu, Haiyang Yu, Fei Huang, Yongbin Li, and Nguyen Cam-Tu. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.57 Improving question generation with multi-level content planning . In Findings of the Association for Computational Linguistics: EMNLP 202...

    2023 doi

  24. [32]

    Wenhan Xiong, Xiang Lorraine Li, Srinivasan Iyer, Jingfei Du, Patrick Lewis, William Yang Wang, Yashar Mehdad, Wen-tau Yih, Sebastian Riedel, Douwe Kiela, and Barlas O g uz. 2021. Answering complex open-domain questions with multi-hop dense retrieval. International Conference ...

    2021

  25. [33]

    Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. https://doi.org/10.18653/v1/D18-1259 H otpot QA : A dataset for diverse, explainable multi-hop question answering . In Proceedings of the 2018 Conference...

    2018 doi

  26. [34]

    Jiahao Zhang, Haiyang Zhang, Dongmei Zhang, Liu Yong, and Shen Huang. 2024. https://doi.org/10.18653/v1/2024.naacl-long.96 End-to-end beam retrieval for multi-hop question answering . In Proceedings of the 2024 Conference of the North American Chapter of the Association for Co...

    2024 doi

  27. [35]

    Xinyu Zhang, Ke Zhan, Enrui Hu, Chengzhen Fu, Lan Luo, Hao Jiang, Yantao Jia, Fan Yu, Zhicheng Dou, Zhao Cao, and Lei Chen. 2021. https://doi.org/10.1145/3404835.3462942 Answer complex questions: Path ranker is all you need . In Proceedings of the 44th International ACM SIGIR ...

    2021

  28. [36]

    Chen Zhao, Chenyan Xiong, Jordan Boyd-Graber, and Hal Daum \'e III. 2021. https://doi.org/10.18653/v1/2021.naacl-main.368 Multi-step reasoning over unstructured text with beam dense retrieval . In Proceedings of the 2021 Conference of the North American Chapter of the Associat...

    2021 doi

  29. [37]

    Yunchang Zhu, Liang Pang, Yanyan Lan, Huawei Shen, and Xueqi Cheng. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.293 Adaptive information seeking for open-domain question answering . In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing...

    2021 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.