pith. sign in

arxiv: 2606.27598 · v1 · pith:M52FNJMYnew · submitted 2026-06-25 · 💻 cs.CL · cs.AI

Narrative-UFET: Narrative Generation for Ultra-Fine Entity Typing

Mreedul Gupta , Advait Deshmukh , Ashwin Umadi , Matt Pauk , Maria Leonor Pacheco This is my paper

Pith reviewed 2026-06-29 01:12 UTC · model grok-4.3

classification 💻 cs.CL cs.AI
keywords ultra-fine entity typingnarrative generationdiscourse contextlong-tail typessynthetic dataentity typingcontext extension
0
0 comments X

The pith

Narrative context improves ultra-fine entity typing accuracy on long-tail types compared to sentence-only or natural contexts.

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

Current ultra-fine entity typing systems rely on single-sentence context and therefore miss disambiguating evidence that appears across multiple sentences. The paper tests this limitation by automatically generating short coherent narratives for each entity mention and pairing them with the original sentence. Two controlled variants are created: one where the entity's type stays constant across the narrative and one where it changes. Experiments show consistent gains on long-tail types from the narrative versions, with the change variant supplying the stronger signal, and synthetic narratives outperforming naturally occurring multi-sentence contexts. The results indicate that deliberate discourse construction can surface signals left implicit in real text.

Core claim

Narrative-UFET pairs each entity mention with an automatically generated short coherent narrative in two paired variants (Maintain, where the type remains constant, and Change, where it shifts), and demonstrates that these narrative contexts produce consistent accuracy gains on long-tail ultra-fine types over sentence-level baselines, with the Change variant stronger, while also outperforming naturally occurring contexts.

What carries the argument

Controlled narrative generation that isolates specific discourse properties (Maintain versus Change) to supply multi-sentence context beyond the original sentence.

If this is right

  • Disambiguating evidence for rare ultra-fine types is often distributed across sentences rather than contained in one.
  • The type-shift property in narrative context supplies a stronger training signal than type maintenance.
  • Deliberately constructed synthetic text can reveal discourse signals that naturally occurring text leaves implicit.
  • Substantial performance headroom remains after adding narrative context.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same controlled-narrative technique could be tested on other span-level tasks such as fine-grained relation extraction or event argument linking.
  • Hybrid architectures that jointly encode the original sentence and the generated narrative may capture both local and discourse-level cues.
  • If narrative generation quality improves, the performance gap between synthetic and natural contexts could widen further.

Load-bearing premise

The automatically generated narratives remain coherent and free of artifacts that would confound the comparison between discourse variants and sentence-level or natural contexts.

What would settle it

No accuracy improvement on long-tail types when models are given the generated narratives instead of only the sentence, or no advantage for the Change variant over the Maintain variant.

Figures

Figures reproduced from arXiv: 2606.27598 by Advait Deshmukh, Ashwin Umadi, Maria Leonor Pacheco, Matt Pauk, Mreedul Gupta.

Figure 1
Figure 1. Figure 1: Type F1 Scores Across UFET Bins: MLM and [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Mean TinyStories narrative quality scores across models for grammar, creativity, consistency, and plot for [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Mean coherence scores across models for model testing. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mean coreference scores across models for model testing. [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Mean TinyStories narrative quality scores [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Mean coherence scores for character testing [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Mean coreference scores for character testing. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Mean TinyStories narrative quality scores [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Mean coherence scores for narrative lengths. [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Mean coreference density scores for different [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: F1 Scores Across Bins. Comparing MLM and CLM performance across Narrative-UFET-Change of Narrative-UFET and the Standard UFET (sentence￾level) [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: F1 Scores Across Bins. Comparing MLM and CLM performance between Narrative-UFET￾Maintain of Narrative-UFET and the Standard UFET (sentence-level). 14 [PITH_FULL_IMAGE:figures/full_fig_p014_12.png] view at source ↗
Figure 15
Figure 15. Figure 15: F1 Scores Across Bins. Comparing Llama Model between both datasets of Narrative-UFET and the Standard UFET (sentence-level) [PITH_FULL_IMAGE:figures/full_fig_p015_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: F1 Scores Across Bins. Comparing Qwen Model between both datasets of Narrative-UFET and the Standard UFET (sentence-level). 15 [PITH_FULL_IMAGE:figures/full_fig_p015_16.png] view at source ↗
read the original abstract

Ultra-fine entity typing (UFET) assigns highly specific types to entity mentions, but current approaches struggle with types in the long tail. We hypothesize that a key limitation is the reliance on sentence-level context, since disambiguating evidence is often spread across multiple sentences. Testing this has been difficult because all existing UFET resources are sentence-level. We present Narrative-UFET, a controlled extension of UFET in which each entity mention is paired with an automatically generated short, coherent narrative. Synthesizing narratives lets us isolate the effect of specific discourse properties. We experiment with two paired variants: one in which the entity's type is held constant across the narrative (Maintain) and one in which it shifts (Change). We show that narrative context yields consistent improvements on long-tail types over sentence-level baselines, with the Change variant providing the stronger signal. A comparison against naturally occurring contexts shows that synthetic narratives yield stronger gains, indicating that controlled discourse construction can surface signals that real text leaves implicit. Substantial room for improvement remains, suggesting open directions in both discourse modeling and narrative construction.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 3 minor

Summary. The manuscript introduces Narrative-UFET, a controlled extension of existing ultra-fine entity typing (UFET) resources. Each entity mention is paired with an automatically generated short coherent narrative, with two paired variants: Maintain (entity type held constant across the narrative) and Change (type shifts). Experiments claim that narrative context yields consistent gains on long-tail types over sentence-level baselines (stronger for Change), and that synthetic narratives outperform naturally occurring contexts.

Significance. If the experimental results hold after controlling for generation artifacts, the work would establish that discourse-level context can surface disambiguating evidence for long-tail UFET types and that deliberate narrative construction can outperform natural text. It would also motivate further research on discourse modeling and controlled synthetic data for entity typing tasks.

major comments (2)
  1. [Narrative generation and experimental setup sections] The central attribution of long-tail gains to the Maintain/Change discourse manipulation and to synthetic construction requires explicit verification that the narrative generator introduces no differential type leakage, coherence artifacts, or stylistic regularities that correlate with the Change condition or with long-tail performance. Without such controls or post-hoc analyses, the sentence-level baseline comparison and the synthetic-vs-natural contrast cannot be cleanly interpreted.
  2. [Results and discussion sections] The comparison to naturally occurring contexts must include matched statistics on narrative length, coherence, and entity-type distribution to ensure the reported stronger gains for synthetic narratives are not driven by uncontrolled differences in the natural-context baseline.
minor comments (3)
  1. [Abstract] The abstract should name the base UFET dataset(s) being extended and report the number of narratives generated per variant.
  2. [Throughout] Notation for the two variants (Maintain vs. Change) should be introduced once with a clear definition and used consistently in all tables and figures.
  3. [Evaluation section] Any human evaluation of narrative coherence or quality should be described with inter-annotator agreement and sample size.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight important aspects of experimental controls needed to support our claims about narrative context in ultra-fine entity typing. We address each major comment below and will revise the manuscript accordingly to include the requested verifications and matched statistics.

read point-by-point responses

  1. Referee: [Narrative generation and experimental setup sections] The central attribution of long-tail gains to the Maintain/Change discourse manipulation and to synthetic construction requires explicit verification that the narrative generator introduces no differential type leakage, coherence artifacts, or stylistic regularities that correlate with the Change condition or with long-tail performance. Without such controls or post-hoc analyses, the sentence-level baseline comparison and the synthetic-vs-natural contrast cannot be cleanly interpreted.

    Authors: We agree that explicit verification is required to rule out generation artifacts as alternative explanations for the observed gains. The manuscript's generation process uses a controlled prompt template that conditions only on the target entity mention, sentence context, and desired type behavior (Maintain vs. Change), with no direct type leakage in the prompt itself. However, we did not include post-hoc checks in the original submission. In the revision we will add: (i) type-mention frequency analysis across conditions to quantify leakage, (ii) automated coherence metrics (e.g., entity-grid coherence and perplexity under a held-out language model) stratified by Maintain/Change and by tail vs. head types, and (iii) stylistic regularity checks via n-gram overlap and sentence-length distributions. These analyses will be reported in a new subsection of the experimental setup. If any differential artifacts are detected, we will discuss their magnitude and potential impact on the long-tail gains. revision: yes

  2. Referee: [Results and discussion sections] The comparison to naturally occurring contexts must include matched statistics on narrative length, coherence, and entity-type distribution to ensure the reported stronger gains for synthetic narratives are not driven by uncontrolled differences in the natural-context baseline.

    Authors: We acknowledge that the current synthetic-vs-natural comparison lacks explicit matching statistics, which limits interpretability of why synthetic narratives yield stronger gains. The natural contexts were drawn from the same underlying documents as the original UFET sentences, while synthetic narratives were generated to a target length of approximately 3–5 sentences. In the revised manuscript we will add a table in the results section reporting: average narrative length (in tokens and sentences), coherence scores, and the distribution of ultra-fine types for both the synthetic Maintain/Change sets and the natural-context baseline. We will also report whether any imbalances exist and, if so, whether they correlate with performance differences. This will allow readers to evaluate the fairness of the contrast. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper is an empirical contribution that introduces a new dataset (Narrative-UFET) via automatic narrative generation and reports experimental gains on long-tail UFET types for Maintain/Change variants versus sentence-level and natural-context baselines. No equations, parameter-fitting steps, or first-principles derivations are present in the provided text. Claims rest on direct comparisons of model performance rather than any self-definitional mapping, fitted-input-as-prediction, or load-bearing self-citation chain. The central result (narrative context improves long-tail typing) is externally falsifiable against held-out data and does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no information on free parameters, background axioms, or new postulated entities.

pith-pipeline@v0.9.1-grok · 5729 in / 1129 out tokens · 41277 ms · 2026-06-29T01:12:10.408760+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

42 extracted references · 20 canonical work pages

  1. [1]

    Suchanek, and Chlo \'e Clavel

    Cyril Chhun, Pierre Colombo, Fabian M. Suchanek, and Chlo \'e Clavel. 2022. https://aclanthology.org/2022.coling-1.509/ Of human criteria and automatic metrics: A benchmark of the evaluation of story generation . In Proceedings of the 29th International Conference on Computational Linguistics, pages 5794--5836, Gyeongju, Republic of Korea. International C...

    2022

  2. [2]

    Eunsol Choi, Omer Levy, Yejin Choi, and Luke Zettlemoyer. 2018. https://doi.org/10.18653/v1/P18-1009 Ultra-fine entity typing . In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 87--96, Melbourne, Australia. Association for Computational Linguistics

    work page doi:10.18653/v1/p18-1009 2018

  3. [3]

    Hongliang Dai, Yangqiu Song, and Haixun Wang. 2021. https://doi.org/10.18653/v1/2021.acl-long.141 Ultra-fine entity typing with weak supervision from a masked language model . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long...

    work page doi:10.18653/v1/2021.acl-long.141 2021

  4. [4]

    Advait Deshmukh, Ashwin Umadi, Dananjay Srinivas, and Maria Leonor Pacheco. 2025. https://doi.org/10.18653/v1/2025.starsem-1.15 All entities are not created equal: Examining the long tail for ultra-fine entity typing . In Proceedings of the 14th Joint Conference on Lexical and Computational Semantics (*SEM 2025), pages 189--201, Suzhou, China. Association...

    work page doi:10.18653/v1/2025.starsem-1.15 2025

  5. [5]

    Ning Ding, Guangwei Xu, Yulin Chen, Xiaobin Wang, Xu Han, Pengjun Xie, Haitao Zheng, and Zhiyuan Liu. 2021. https://doi.org/10.18653/v1/2021.acl-long.248 Few- NERD : A few-shot named entity recognition dataset . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural...

    work page doi:10.18653/v1/2021.acl-long.248 2021

  6. [6]

    Ronen Eldan and Yuanzhi Li. 2023. https://arxiv.org/abs/2305.07759 Tinystories: How small can language models be and still speak coherent english? Preprint, arXiv:2305.07759

    Pith/arXiv arXiv 2023

  7. [9]

    Seraphina Goldfarb-Tarrant, Tuhin Chakrabarty, Ralph Weischedel, and Nanyun Peng. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.351 Content planning for neural story generation with aristotelian rescoring . In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 4319--4338, Online. Association for Compu...

    work page doi:10.18653/v1/2020.emnlp-main.351 2020

  8. [10]

    K.L. Gwet. 2014. https://books.google.com/books?id=fac9BQAAQBAJ Handbook of Inter-Rater Reliability, 4th Edition: The Definitive Guide to Measuring The Extent of Agreement Among Raters . Advanced Analytics, LLC

    2014

  9. [11]

    Fabrice Y Harel-Canada, Hanyu Zhou, Sreya Muppalla, Zeynep Senahan Yildiz, Miryung Kim, Amit Sahai, and Nanyun Peng. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.953 Measuring psychological depth in language models . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 17162--17196, Miami, Florida, USA. Ass...

    work page doi:10.18653/v1/2024.emnlp-main.953 2024

  10. [12]

    Mitchell Koch, John Gilmer, Stephen Soderland, and Daniel S. Weld. 2014. https://doi.org/10.3115/v1/D14-1203 Type-aware distantly supervised relation extraction with linked arguments . In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing ( EMNLP ) , pages 1891--1901, Doha, Qatar. Association for Computational Linguistics

    work page doi:10.3115/v1/d14-1203 2014

  11. [13]

    Jinyuan Li, Han Li, Di Sun, Jiahao Wang, Wenkun Zhang, Zan Wang, and Gang Pan. 2024 a . https://doi.org/10.18653/v1/2024.findings-acl.76 LLM s as bridges: Reformulating grounded multimodal named entity recognition . In Findings of the Association for Computational Linguistics: ACL 2024, pages 1302--1318, Bangkok, Thailand. Association for Computational Li...

    work page doi:10.18653/v1/2024.findings-acl.76 2024

  12. [14]

    Muzhi Li, Minda Hu, Irwin King, and Ho-fung Leung. 2024 b . https://doi.org/10.18653/v1/2024.naacl-long.369 The integration of semantic and structural knowledge in knowledge graph entity typing . In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long...

    work page doi:10.18653/v1/2024.naacl-long.369 2024

  13. [15]

    Na Li, Zied Bouraoui, and Steven Schockaert. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.786 Ultra-fine entity typing with prior knowledge about labels: A simple clustering based strategy . In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 11744--11756, Singapore. Association for Computational Linguistics

    work page doi:10.18653/v1/2023.findings-emnlp.786 2023

  14. [16]

    Xiao Ling, Sameer Singh, and Daniel S. Weld. 2015. https://doi.org/10.1162/tacl_a_00141 Design challenges for entity linking . Transactions of the Association for Computational Linguistics, 3:315--328

    work page doi:10.1162/tacl_a_00141 2015

  15. [18]

    Ben Naismith, Phoebe Mulcaire, and Jill Burstein. 2023. https://doi.org/10.18653/v1/2023.bea-1.32 Automated evaluation of written discourse coherence using GPT -4 . In Proceedings of the 18th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2023), pages 394--403, Toronto, Canada. Association for Computational Linguistics

    work page doi:10.18653/v1/2023.bea-1.32 2023

  16. [19]

    Yasumasa Onoe and Greg Durrett. 2020. https://doi.org/10.18653/v1/2020.findings-emnlp.54 Interpretable entity representations through large-scale typing . In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 612--624, Online. Association for Computational Linguistics

    work page doi:10.18653/v1/2020.findings-emnlp.54 2020

  17. [21]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. http://jmlr.org/papers/v21/20-074.html Exploring the limits of transfer learning with a unified text-to-text transformer . Journal of Machine Learning Research, 21(140):1--67

    2020

  18. [22]

    Sebastian Riedel, Limin Yao, and Andrew McCallum. 2010. https://api.semanticscholar.org/CorpusID:2386383 Modeling relations and their mentions without labeled text . In ECML/PKDD

    2010

  19. [24]

    Yufei Tian, Tenghao Huang, Miri Liu, Derek Jiang, Alexander Spangher, Muhao Chen, Jonathan May, and Nanyun Peng. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.978 Are large language models capable of generating human-level narratives? In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 17659--17681, Miami, ...

    work page doi:10.18653/v1/2024.emnlp-main.978 2024

  20. [25]

    Ziqi Wang, Chen Zhu, Zhi Zheng, Xinhang Li, Tong Xu, Yongyi He, Qi Liu, Ying Yu, and Enhong Chen. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.183 Granular entity mapper: Advancing fine-grained multimodal named entity recognition and grounding . In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 3211--3226, Miami, Flo...

    work page doi:10.18653/v1/2024.findings-emnlp.183 2024

  21. [26]

    Maurice Weber, Daniel Fu, Quentin Anthony, Yonatan Oren, Shane Adams, Anton Alexandrov, Xiaozhong Lyu, Huu Nguyen, Xiaozhe Yao, Virginia Adams, Ben Athiwaratkun, Rahul Chalamala, Kezhen Chen, Max Ryabinin, Tri Dao, Percy Liang, Christopher Ré, Irina Rish, and Ce Zhang. 2024. https://arxiv.org/abs/2411.12372 Redpajama: an open dataset for training large la...

    arXiv 2024

  22. [27]

    Kevin Yang, Yuandong Tian, Nanyun Peng, and Dan Klein. 2022. https://doi.org/10.18653/v1/2022.emnlp-main.296 Re3: Generating longer stories with recursive reprompting and revision . In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 4393--4479, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics

    work page doi:10.18653/v1/2022.emnlp-main.296 2022

  23. [29]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    1972

  24. [30]

    Publications Manual , year = "1983", publisher =

    1983

  25. [31]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    work page doi:10.1145/322234.322243 1981

  26. [32]

    Scalable training of

    Andrew, Galen and Gao, Jianfeng , booktitle=. Scalable training of

  27. [33]

    Dan Gusfield , title =. 1997

    1997

  28. [34]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    2015

  29. [35]

    A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =

    Ando, Rie Kubota and Zhang, Tong , Issn =. A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data , Volume =. Journal of Machine Learning Research , Month = dec, Numpages =

  30. [36]

    Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence,

    Automatic Noisy Label Correction for Fine-Grained Entity Typing , author =. Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence,. 2022 , month =. doi:10.24963/ijcai.2022/599 , url =

    work page doi:10.24963/ijcai.2022/599 2022

  31. [37]

    2023 , eprint=

    TinyStories: How Small Can Language Models Be and Still Speak Coherent English? , author=. 2023 , eprint=

    2023

  32. [38]

    Proceedings of the AAAI Conference on Artificial Intelligence , author=

    Fine-Grained Entity Recognition , volume=. Proceedings of the AAAI Conference on Artificial Intelligence , author=. 2021 , month=. doi:10.1609/aaai.v26i1.8122 , abstractNote=

    work page doi:10.1609/aaai.v26i1.8122 2021

  33. [39]

    2016 , eprint=

    Context-Dependent Fine-Grained Entity Type Tagging , author=. 2016 , eprint=

    2016

  34. [40]

    2025 , eprint=

    Evaluating Creative Short Story Generation in Humans and Large Language Models , author=. 2025 , eprint=

    2025

  35. [41]

    ArXiv , year=

    Context-Dependent Fine-Grained Entity Type Tagging , author=. ArXiv , year=

  36. [42]

    doi:10.1016/j.heliyon.2024.e34262 Pamela Tierney and Steven M

    Xiaoyi Tang and Hongwei Chen and Daoyu Lin and Kexin Li , keywords =. Harnessing LLMs for multi-dimensional writing assessment: Reliability and alignment with human judgments , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.heliyon.2024.e34262 , url =

    work page doi:10.1016/j.heliyon.2024.e34262 2024

  37. [43]

    2014 , publisher=

    Handbook of Inter-Rater Reliability, 4th Edition: The Definitive Guide to Measuring The Extent of Agreement Among Raters , author=. 2014 , publisher=

    2014

  38. [44]

    ECML/PKDD , year=

    Modeling Relations and Their Mentions without Labeled Text , author=. ECML/PKDD , year=

  39. [45]

    CoRR , volume =

    Dan Gillick and Nevena Lazic and Kuzman Ganchev and Jesse Kirchner and David Huynh , title =. CoRR , volume =. 2014 , url =. 1412.1820 , timestamp =

    Pith/arXiv arXiv 2014

  40. [46]

    Zemel and Ruslan Salakhutdinov and Raquel Urtasun and Antonio Torralba and Sanja Fidler , title =

    Yukun Zhu and Ryan Kiros and Richard S. Zemel and Ruslan Salakhutdinov and Raquel Urtasun and Antonio Torralba and Sanja Fidler , title =. CoRR , volume =. 2015 , url =. 1506.06724 , timestamp =

    Pith/arXiv arXiv 2015

  41. [47]

    Liu , title =

    Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu , title =. Journal of Machine Learning Research , year =

  42. [48]

    2024 , eprint=

    RedPajama: an Open Dataset for Training Large Language Models , author=. 2024 , eprint=

    2024