Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Interactive Machine Comprehension with Information Seeking Agents

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper claims that reframing reading-comprehension datasets as interactive, partially observable environments allows a reinforcement-learning agent to learn to seek relevant information through sequential decisions.

desk verdict The paper's real contribution is the new interactive MRC task and public datasets; the RL agent's search behavior is plausible but under-proven because the experiments lack non-learning baselines and variance reporting. read the letter →

arxiv 1908.10449 v3 pith:35CNR4JS submitted 2019-08-27 cs.CL cs.LGstat.ML

Xingdi Yuan , Jie Fu , Marc-Alexandre Cote , Yi Tay , Christopher Pal , Adam Trischler This is my paper · ORCID
classification cs.CLcs.LGstat.ML
keywords machinereadingcomprehensioninteractiveenvironmentpartialobservabilityreinforcementlearninginformationseekingquestionansweringSQuADNewsQA
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

The paper argues that reading-comprehension benchmarks are too static: the supporting document is fully visible, so models can answer by spotting surface similarities between question and text, and they never learn to gather information the way a person would. To fix this, the authors convert existing datasets into interactive, partially observable environments called iMRC: a document is split into sentences, almost all of them are hidden, and an agent must issue commands — move to the previous or next sentence, search with a Ctrl+F query, or stop — to reveal glimpses of the text before answering. The task is formalized as a partially observable Markov decision process, and a baseline agent couples a reinforcement-learning action generator with a QANet-style answerer. Trained on the resulting iSQuAD and iNewsQA benchmarks, the agent learns to seek the sentences that contain answers, and its errors are dominated by information gathering rather than by answer extraction. If this reframing transfers, it offers a route toward MRC systems that can search large, partially observable knowledge sources such as the web.

What carries the argument

The load-bearing mechanism is the iMRC environment itself: a POMDP in which each document is split into sentences, all but one are hidden, and the agent reveals text by issuing one of a small set of commands (previous, next, Ctrl+F QUERY, stop). The agent's encoder produces an attention-aggregated representation of the question and current observation; an action generator scores action and query tokens, and a QANet-style answerer produces head and tail pointers over the observed text when the agent stops. Two design elements carry the learning: a bounded memory queue that stores recently seen sentences so the agent can reason over context larger than one sentence, and a shaped 'sufficient information reward' that pays the agent for stopping with the ground-truth answer substring in view. These convert the sparse, delayed reward of question answering into a signal that guides exploration.

What would settle it

Run an agent trained on iMRC against a real open-domain QA setup in which it must issue search commands over a large corpus, and compare it with a non-interactive baseline that receives a retrieved passage: if the interactive agent gains nothing or its performance does not track its iMRC F1info, the transfer claim fails. A sharper within-paper test would ablate the sufficient-information reward and check whether the sequential policy still learns to seek.

Watch

Extended reading notes

Core claim

The paper's central claim is that sequential information seeking can be learned and measured in natural-language MRC by occluding documents and letting the agent act. Concretely, iMRC withholds all but the first sentence of a paragraph, gives the agent a small command set, and asks it to stop when it has enough context to answer; easy mode adds previous/next navigation, hard mode forces the agent to use Ctrl+F queries alone. The authors show that a baseline agent trained with DQN and A2C on this POMDP improves its answer F1 as it learns to find the answer sentence, that scores are consistently higher when the terminating observation actually contains the answer (the F1info measure), and that larger action spaces and harder navigation modes degrade performance. The discovery is therefore that the hard part of partially observable comprehension is deciding where to look, not reading what is found, and that a simple word-level search action can be acquired through reinforcement learning.

Load-bearing premise

The results rest on the assumption that the synthetic interactive environment — deterministic moves, one-word queries, sentence-sized glimpses, a 20-step budget — is a faithful stand-in for real-world information seeking; if that proxy is too artificial, progress on iMRC would not demonstrate that models can scale to web-level question answering.

Editorial extensions

If this is right

  • Information gathering, not answer extraction, is the bottleneck in partially observable reading: across nearly all settings the reported F1 is lower than F1info, the score an agent achieves when it stops on a sentence containing the answer.
  • Action-space size controls difficulty: agents do best when the query token is drawn from the question (a small action space) and worst when it is drawn from the full vocabulary, and they do better in easy mode with navigation commands than in hard mode with Ctrl+F only.
  • Exploration method matters for generalization: DQN-like explicit exploration generalizes to held-out questions markedly better than A2C's implicit sampling, suggesting that exploration strategy is a primary determinant of interactive MRC success.
  • Larger memory helps: keeping three or five observed sentences in a queue almost always improves test F1 over a single-slot memory, since the environment becomes closer to fully observed.
  • The reframing is dataset-agnostic: because it only requires splitting documents into sentences and hiding them, nearly any extractive MRC dataset can be converted into an interactive information-seeking benchmark without new annotation.

Reading between the lines

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

  • Editorial inference: the paper's own numbers imply that the shaped 'sufficient information reward' may be doing much of the work; a testable ablation that removes that reward and compares learning curves would isolate the contribution of the sequential policy.
  • Editorial inference: if the interactive framing transfers, the natural next step — which the paper names as future work — is replacing single-token vocabulary queries with latent vector queries and fuzzy matching, which would bring the setup closer to learned information retrieval and could sharply reduce exploration cost.
  • Editorial inference: because iMRC is built on documents that are short enough for a 20-step budget to cover, the current benchmarks may understate the difficulty of web-scale search; converting long-document or multi-hop datasets such as HotpotQA would be a stronger stress test of the information-seeking claim.
  • Editorial inference: a comparison of an iMRC-trained agent against the same answerer given the full document would quantify how much performance is lost to active search, and whether the sequential policy is worth its added complexity.
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

3 major / 5 minor

Summary. The paper proposes iMRC, a reformulation of static extractive MRC datasets (SQuAD v1.1 and NewsQA) as interactive, partially observable environments. A document is split into sentences, only a small window or memory of sentences is observable at a time, and the agent must issue commands (previous, next, Ctrl+F QUERY, stop) to reveal further text before producing an extractive answer. The authors define easy and hard modes, three QUERY-generation settings, and a POMDP formulation. As a baseline, they adapt QA-DQN with a QANet-style answerer and train it with DQN and A2C, with optional memory queues and a 'sufficient information' reward. The experiments report training curves and test F1 and F1info scores for iSQuAD and iNewsQA across these settings, and the paper discusses trends such as easy mode outperforming hard mode, DQN generally outperforming A2C, and larger memories helping.

Significance. If the empirical claims are supported, this is a useful contribution to the study of information-seeking behavior in NLP: it provides a simple, general recipe for converting existing MRC datasets into interactive benchmarks, releases the data and code publicly, and formulates a clean RL baseline. The evaluation uses held-out test splits of standard datasets, and the sufficient-information reward is used only as reward shaping, not as the evaluation metric, so there is no circularity in the main performance measure. The work also honestly discusses the preliminary nature of the experiments. However, the central behavioral claim that the agent 'seeks relevant information through sequential decision making' is underdetermined by the current experiments because no non-learning controls, no full-document QA upper bound, and no variance estimates are reported.

major comments (3)
  1. [Section 5.2, Table 3] The central claim that the agent learns to seek relevant information is not supported without non-learning control policies. In easy mode, the next action deterministically scans sentences within the 20-step budget; in hard mode, a policy that issues Ctrl+F for question tokens in order is available. None of these trivial baselines appears, so the reported DQN F1 scores (e.g., 0.524 hard-mode iSQuAD, 0.352 hard-mode iNewsQA) cannot be decomposed into the answerer's contribution and the learned search policy. I ask for random-action, first-sentence-only, stop-at-first-sentence, and exhaustive-scan baselines to be added.
  2. [Section 5.2, Table 3 and Section 4.3.3] The interpretation that information gathering is the main bottleneck relies on F1info, but there is no full-document QANet upper bound and no control for the answerer's training distribution. The answerer is trained only on terminal observations that contain the ground-truth answer, yet at test time it must produce a span from arbitrary terminal observations, including those without sufficient information. This train/test mismatch could itself explain much of the gap between overall F1 and F1info, and it should be addressed by reporting a full-document baseline, a stop-at-first-sentence baseline, or an answerer trained on all terminal observations.
  3. [Section 5.2, Table 3] All experimental results are from single runs, with no multiple seeds, error bars, or significance tests. Differences such as 0.524 versus 0.551 in hard-mode iSQuAD, or 0.352 versus 0.367 in hard-mode iNewsQA, may be within noise. The paper should report variance across seeds or at least state the number of runs used, so that the claimed trends (memory size, query source, DQN over A2C) can be assessed.
minor comments (5)
  1. [Section 3.2] Easy mode does not specify what Ctrl+F does when there is no next occurrence of the query, or whether the current sentence is revisited if it already contains the query; this should be stated explicitly for reproducibility.
  2. [Section 5.2 and Table 3] F1info is defined only in prose; the caption should state formally whether it is a conditional F1 computed only over episodes whose terminal memory contains the answer, or an upper-bound oracle measure.
  3. [Figures 2 and 3] The training curves show no error bars or indication of the number of runs; please clarify whether these are single runs and whether the trends are stable across seeds.
  4. [Section 2, Related Works] The statement that supervised learning 'requires human labeled or heuristically generated trajectories' and that generalization can be limited is plausible but is presented as fact without experimental support; it should be softened or paired with a reference that demonstrates the limitation for the cited methods.
  5. [Section 6, Discussion and Future Work] The paragraph on scaling to web-level QA via hyperlink-click actions is appropriately framed as a hypothesis, but the abstract's phrasing 'We believe that this setting can contribute in scaling models to web-level QA scenarios' is consistent with that framing; consider making the abstract equally explicit that web-level scaling is not demonstrated in this paper.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central empirical evaluation is self-contained against external benchmarks, and the existing self-citations are not load-bearing.

full rationale

The paper's central contribution is a reframing of existing MRC datasets (SQuAD v1.1 and NewsQA) into interactive, partially observable environments, followed by an empirical evaluation of a baseline agent on held-out test splits of those external benchmarks. The load-bearing evidence for the claim that the agent 'seeks relevant information through sequential decision making' consists of F1 and F1info scores computed against ground-truth answers on test sets, which are external to the method. The sufficient-information reward used during training is reward shaping, not a fitted parameter later reported as a prediction, and F1info is a diagnostic metric conditioned on the ground-truth answer appearing in the observation; neither constitutes a derivation that reduces to its own inputs by construction. The paper does cite prior work by its own authors, notably QA-DQN (Yuan et al., 2019), as the basis for the baseline agent and as inspiration, but this is a method appropriation rather than a load-bearing argument: no uniqueness theorem or circular definition is imported, and the results are not asserted to follow from that citation alone. The absence of non-learning baselines in Table 3 is a legitimate experimental-design concern about whether the reported F1 scores demonstrate learned search behavior, but that is a weakness in evidence strength, not an instance of circularity under the definitions used here. Overall, the derivation chain is self-contained against external benchmarks, with at most a minor non-load-bearing self-citation.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central results rest on standard RL training, a heuristic reward based on answer substrings, and the assumption that the synthetic occlusion game captures real information-seeking difficulty. Hyperparameters such as step budget and memory size are hand-selected and affect the reported numbers.

free parameters (4)
  • Step budget = 20
    Chosen by hand in Section 3; limits how much of the document can be revealed and directly sets task difficulty.
  • Memory queue size = 1, 3, 5
    Chosen by hand in Section 4.2.1; larger memory is shown to improve results, so reported performance depends on this inductive bias.
  • Discount factor gamma = 0.9
    Standard RL hyperparameter, hand-set in Appendix B.
  • Entropy regularization coefficient = 0.01
    Hand-set in Appendix B for A2C exploration.
assumptions (4)
  • domain assumption Off-the-shelf RL algorithms (Rainbow DQN, A2C) will learn effective information-gathering policies in this POMDP given the designed reward signal.
    Section 4.3 assumes these algorithms can solve the task without task-specific exploration heuristics; the paper only provides one set of runs.
  • ad hoc to paper If the ground-truth answer is a substring of the current observation, the agent has sufficient information to answer correctly.
    Section 4.2.2 defines the sufficient information reward on this premise; it is a heuristic, not a proven property of the environment.
  • domain assumption SQuAD v1.1 and NewsQA extractive spans are valid for studying interactive comprehension.
    Section 3 repurposes these datasets without validating whether extractive span answering transfers to the interactive setting.
  • domain assumption Deterministic environment transitions with noiseless observations adequately model document occlusion.
    Section 3.1 sets transition and observation probabilities to 0 or 1, simplifying the POMDP and possibly missing real-world stochasticity.
invented entities (1)
  • Sufficient information reward
    purpose: A shaping reward that gives positive feedback when the terminal observation contains the ground-truth answer as a substring, used to guide the RL agent toward answer-bearing sentences.
    It is defined and used only within this paper; its validity as a proxy for sufficient information is not independently demonstrated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Interactive Machine Comprehension with Information Seeking Agents." pith.science (2026). https://pith.science/paper/35CNR4JS

@misc{pith2026190810449,
  author       = {Pith},
  title        = {Pith review of: Interactive Machine Comprehension with Information Seeking Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/35CNR4JS}},
  note         = {Machine review of arXiv:1908.10449}
}
read the original abstract

Existing machine reading comprehension (MRC) models do not scale effectively to real-world applications like web-level information retrieval and question answering (QA). We argue that this stems from the nature of MRC datasets: most of these are static environments wherein the supporting documents and all necessary information are fully observed. In this paper, we propose a simple method that reframes existing MRC datasets as interactive, partially observable environments. Specifically, we "occlude" the majority of a document's text and add context-sensitive commands that reveal "glimpses" of the hidden text to a model. We repurpose SQuAD and NewsQA as an initial case study, and then show how the interactive corpora can be used to train a model that seeks relevant information through sequential decision making. We believe that this setting can contribute in scaling models to web-level QA scenarios.

Figures

Figures reproduced from arXiv: 1908.10449 by the authors.

Figure 1
Figure 1. A demonstration of the proposed iMRC pipeline, in which the agent is illustrated as a shaded area. At a game step t, it encodes the question and text observation into hidden representations Mt. An action generator takes Mt as input to generate commands to interact with the environment. If the agent generates stop at this game step, Mt is used to answer question by a question answerer. Otherwise, the iMRC environ￾men… view at source ↗
Figure 2
Figure 2. Training F1 scores in easy mode with differ￾ent QUERY types and memory sizes. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5. that contains sufficient information to answer the question, the task becomes a standard extractive QA setting, where an agent learns to point to a span from its observation. When this condition is met, it is also natural to adopt standard supervised learning methods to tr… view at source ↗
Figure 7
Figure 7. Validation performance on iSQuAD, hard mode. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5 [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figures from the paper (4 more)
Figure 8
Figure 8. Figure 8: Training performance on iNewsQA, easy mode. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5 [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Validation performance on iNewsQA, easy mode. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5 [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Training performance on iNewsQA, hard mode. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5 [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: Validation performance on iNewsQA, hard mode. Solid line: DQN, dashed line: A2C; number of memory slots: 1, 3, 5. randomly initialize character embeddings by 200- dimensional vectors. In all transformer blocks, block size is 96. Dimensionality of MLPshared in Eqn. 1 i…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Semantic Web and Creative AI -- A Technical Report from ISWS 2023

    cs.AI 2025-01 conditional novelty 5.0 of 10

    Ten exploratory studies from ISWS 2023 test how knowledge graphs and large language models can combine for creative and knowledge-engineering tasks, with mostly small proof-of-concept evidence.

Reference graph

Works this paper leans on

33 extracted references · 11 canonical work pages · cited by 1 Pith paper

  1. [1]

    Philip Bachman, Alessandro Sordoni, and Adam Trischler. 2016. Towards information-seeking agents. arXiv preprint arXiv:1612.02605

    arXiv 2016

  2. [2]

    Eunsol Choi, Daniel Hewlett, Jakob Uszkoreit, Illia Polosukhin, Alexandre Lacoste, and Jonathan Berant. 2017. Coarse-to-fine question answering for long documents. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 209--220

    work page 2017

  3. [3]

    Meire Fortunato, Mohammad Gheshlaghi Azar, Bilal Piot, Jacob Menick, Ian Osband, Alex Graves, Vlad Mnih, R \' e mi Munos, Demis Hassabis, Olivier Pietquin, Charles Blundell, and Shane Legg. 2017. http://arxiv.org/abs/1706.10295 Noisy networks for exploration . CoRR, abs/1706.10295

    arXiv 2017

  4. [4]

    Mor Geva and Jonathan Berant. 2018. Learning to search in long documents using document structure. arXiv preprint arXiv:1806.03529

    work page Pith review arXiv 2018

  5. [5]

    Moonsu Han, Minki Kang, Hyunwoo Jung, and Sung Ju Hwang. 2019. Episodic memory reader: Learning what to remember for question answering from streaming data. arXiv preprint arXiv:1903.06164

    work page Pith review arXiv 2019

  6. [6]

    Christian Hansen, Casper Hansen, Stephen Alstrup, Jakob Grue Simonsen, and Christina Lioma. 2019. Neural speed reading with structural-jump-lstm. arXiv preprint arXiv:1904.00761

    arXiv 2019

  7. [7]

    Matteo Hessel, Joseph Modayil, Hado van Hasselt, Tom Schaul, Georg Ostrovski, Will Dabney, Daniel Horgan, Bilal Piot, Mohammad Gheshlaghi Azar, and David Silver. 2017. http://arxiv.org/abs/1710.02298 Rainbow: Combining improvements in deep reinforcement learning . CoRR, abs/1710.02298

    arXiv 2017

  8. [8]

    Leslie Pack Kaelbling, Michael L Littman, and Anthony R Cassandra. 1998. Planning and acting in partially observable stochastic domains. Artificial intelligence, 101(1-2):99--134

    work page 1998

Show all 33 references
  1. [9]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  2. [10]

    Tomas Mikolov, Edouard Grave, Piotr Bojanowski, Christian Puhrsch, and Armand Joulin. 2018. Advances in pre-training distributed word representations. In Proceedings of the International Conference on Language Resources and Evaluation (LREC 2018)

    2018

  3. [11]

    Lillicrap, Tim Harley, David Silver, and Koray Kavukcuoglu

    Volodymyr Mnih, Adri \` a Puigdom \` e nech Badia, Mehdi Mirza, Alex Graves, Timothy P. Lillicrap, Tim Harley, David Silver, and Koray Kavukcuoglu. 2016. http://arxiv.org/abs/1602.01783 Asynchronous methods for deep reinforcement learning . CoRR, abs/1602.01783

    2016 arXiv

  4. [12]

    Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A Rusu, Joel Veness, Marc G Bellemare, Alex Graves, Martin Riedmiller, Andreas K Fidjeland, Georg Ostrovski, et al. 2015. Human-level control through deep reinforcement learning. Nature, 518(7540):529--533

    2015

  5. [13]

    Karthik Narasimhan, Adam Yala, and Regina Barzilay. 2016. http://arxiv.org/abs/1603.07954 Improving information extraction by acquiring external evidence with reinforcement learning . CoRR, abs/1603.07954

    2016 arXiv

  6. [14]

    Tri Nguyen, Mir Rosenberg, Xia Song, Jianfeng Gao, Saurabh Tiwary, Rangan Majumder, and Li Deng. 2016. http://arxiv.org/abs/1611.09268 MS MARCO: A human generated machine reading comprehension dataset . CoRR, abs/1611.09268

    2016 arXiv

  7. [15]

    Rodrigo Nogueira and Kyunghyun Cho. 2016. http://arxiv.org/abs/1602.02261 Webnav: A new large-scale task for natural language based sequential decision making . CoRR, abs/1602.02261

    2016 arXiv

  8. [16]

    Rodrigo Nogueira and Kyunghyun Cho. 2017. http://arxiv.org/abs/1704.04572 Task-oriented query reformulation with reinforcement learning . CoRR, abs/1704.04572

    2017 arXiv

  9. [17]

    Peng Qi, Xiaowen Lin, Leo Mehr, Zijian Wang, and Christopher D. Manning. 2019. https://nlp.stanford.edu/pubs/qi2019answering.pdf Answering complex open-domain questions through iterative query generation . In 2019 Conference on Empirical Methods in Natural Language Processing ...

    2019

  10. [18]

    Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. http://arxiv.org/abs/1606.05250 Squad: 100, 000+ questions for machine comprehension of text . CoRR, abs/1606.05250

    2016 arXiv

  11. [19]

    Siva Reddy, Danqi Chen, and Christopher D. Manning. 2018. http://arxiv.org/abs/1808.07042 Coqa: A conversational question answering challenge . CoRR, abs/1808.07042

    2018 arXiv

  12. [20]

    Tom Schaul, John Quan, Ioannis Antonoglou, and David Silver. 2016. Prioritized experience replay. In International Conference on Learning Representations, Puerto Rico

    2016

  13. [21]

    Minjoon Seo, Sewon Min, Ali Farhadi, and Hannaneh Hajishirzi. 2017. Neural speed reading via skim-rnn. arXiv preprint arXiv:1711.02085

    2017 arXiv

  14. [22]

    Lei Sha, Feng Qian, and Zhifang Sui. 2017. Will repeated reading benefit natural language understanding? In National CCF Conference on Natural Language Processing and Chinese Computing, pages 366--379. Springer

    2017

  15. [23]

    Yelong Shen, Po-Sen Huang, Jianfeng Gao, and Weizhu Chen. 2017. Reasonet: Learning to stop reading in machine comprehension. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1047--1055. ACM

    2017

  16. [24]

    Saku Sugawara, Kentaro Inui, Satoshi Sekine, and Akiko Aizawa. 2018. http://arxiv.org/abs/1808.09384 What makes reading comprehension questions easier? CoRR, abs/1808.09384

    2018 arXiv

  17. [25]

    Adam Trischler, Tong Wang, Xingdi Yuan, Justin Harris, Alessandro Sordoni, Philip Bachman, and Kaheer Suleman. 2016. http://arxiv.org/abs/1611.09830 Newsqa: A machine comprehension dataset . CoRR, abs/1611.09830

    2016 arXiv

  18. [26]

    Christopher J. C. H. Watkins and Peter Dayan. 1992. https://doi.org/10.1007/BF00992698 Q-learning . Machine Learning, 8(3):279--292

    1992 doi

  19. [27]

    Ronald J Williams. 1992. Simple statistical gradient-following algorithms for connectionist reinforcement learning. Machine learning, 8(3-4):229--256

    1992

  20. [28]

    Cohen, Ruslan Salakhutdinov, and Christopher D

    Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W. Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. http://arxiv.org/abs/1809.09600 Hotpotqa: A dataset for diverse, explainable multi-hop question answering . CoRR, abs/1809.09600

    2018 arXiv

  21. [29]

    Adams Wei Yu, David Dohan, Minh - Thang Luong, Rui Zhao, Kai Chen, Mohammad Norouzi, and Quoc V. Le. 2018. http://arxiv.org/abs/1804.09541 Qanet: Combining local convolution with global self-attention for reading comprehension . CoRR, abs/1804.09541

    2018 arXiv

  22. [30]

    Adams Wei Yu, Hongrae Lee, and Quoc V Le. 2017. Learning to skim text. arXiv preprint arXiv:1704.06877

    2017 arXiv

  23. [31]

    Xingdi Yuan, Marc-Alexandre C\^ot\' e , Jie Fu, Zhouhan Lin, Christopher Pal, Yoshua Bengio, and Adam Trischler. 2019. Interactive language learning by question answering

    2019

  24. [32]

    URL: " 'urlintro :=

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

  25. [33]

    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 14, 2026 · model on record in the stance chip above.