Pith. sign in

REVIEW 3 major objections 6 minor 31 references

The Maskability Index, defined as the relative DepthRank improvement of masked over prefix prompting at five examples, predicts which template family will generate better knowledge completions, and a 30% threshold cleanly separates mask-fil

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 10:17 UTC pith:I72NGLY5

load-bearing objection A clean diagnostic idea undone by its own validation: one of the three map-phrasal relations behaves opposite to the predicted direction, and the paper ignores that tension. the 3 major comments →

arxiv 2607.20265 v1 pith:I72NGLY5 submitted 2026-07-22 cs.CL cs.AI

The Maskability Index: Predicting Task-Objective Alignment in Pretrained Language Models

Ahmad Pouramini , Mahsa Afsharzadeh This is my paper
classification cs.CL cs.AI MSC 68T5068Q32
keywords Maskability IndexDepthRankprompt-template alignmentknowledge base completionpretrained language modelsmasked vs prefix promptingfew-shot learningATOMIC2020
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Researchers often find that the same knowledge-base relation responds differently to masked 'fill in the blank' prompts versus prefix 'continue the sentence' prompts, depending on the pretrained model. This paper tries to make that dependence predictable with a single per-relation number, the Maskability Index (MI). MI measures how much closer to the top of the model's probability list the gold answer tokens move when the tail is masked, relative to when the model continues from a prefix. The paper argues that at five examples, MI separates relations with short, lexical tails (mask-filling: locations, attributes, object uses) from relations with long, compositional tails (map-phrasal: wants, needs, intents), and that this separation predicts which prompt family scores higher on T5-base under ROUGE and BERTScore. If this holds, MI gives a cheap way to choose prompts and adaptation strategies in low-resource knowledge base completion without running full evaluations.

Core claim

The paper's central claim is that template-objective alignment is a measurable relation-level property, and the Maskability Index captures it. For each relation, DepthRank is averaged over n=5 heads under two template families; MI is the relative reduction in DepthRank when the tail is masked instead of prefix-generated. Using a 30% threshold, six of the nine ATOMIC2020 relations studied (AtLocation, ObjectUse, CapableOf, HasProperty, FilledBy, xAttr) fall into a mask-filling group, and three (xIntent, xNeed, xWant) into a map-phrasal group. The paper reports that mask-filling relations achieve higher ROUGE and BERTScore under masked prompting, while map-phrasal relations do better under pre

What carries the argument

The central object is DepthRank, the average rank of each gold tail token in the model's sorted next-token probability distribution; lower DepthRank means the gold completion is easier to generate. MI is the derived ratio (DR_P − DR_MP)/DR_P, computed at a few-shot sample size of n=5, with a 30% threshold partitioning relations into mask-filling and map-phrasal. DepthRank converts the model's internal likelihood ordering into a per-relation alignment score; MI converts the difference between two template families into a binary decision rule that can be computed before any downstream evaluation.

Load-bearing premise

The claim rests on DepthRank differences measured with one specific set of verbalizer templates reflecting genuine masked-versus-prefix alignment, rather than surface effects such as tail length and token frequency (xAttr tails are short and high-frequency while xIntent tails start with the very common token 'to'), and the paper provides no normalization or alternative-template check for these confounds.

What would settle it

Construct new relations or re-verbalize existing ones so that tail length and token frequency are controlled across the two template families, compute MI at n=5, and check whether the 30% grouping still predicts which template gives higher held-out ROUGE/BERTScore; if the grouping reverses when lexical confounds are removed, MI is measuring lexical artifacts rather than objective alignment.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Practitioners can decide between masked and prefix templates for a new relation by computing DepthRank on five examples, instead of running full generation evaluations.
  • The observed grouping—six mask-filling relations, three map-phrasal—gives a concrete, testable map of ATOMIC2020 behavior under T5-base.
  • Because both ROUGE and BERTScore correlate strongly with each other across relations, validation of the grouping does not hinge on which semantic-fidelity metric is used.
  • The MI threshold at 30% is a stated operational rule; relations on either side of it are predicted to require different prompting strategies.

Where Pith is reading between the lines

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

  • The paper's own caveat about the frequent token 'to' suggests a testable extension: a frequency-normalized MI (per-token weighting by corpus frequency) should reduce or eliminate the grouping's dependence on lexical artifacts; if the grouping survives, MI is more likely to be about alignment.
  • One could probe generality by computing MI for the same relations with alternative verbalizer templates; if the mask-filling/map-phrasal assignment flips with template wording, the index is template-dependent rather than relation-intrinsic.
  • The paper frames MI as guiding 'adaptation strategies' but experiments only vary template families; a natural next step is to test whether MI also predicts when prompt-tuning versus fine-tuning will win.
  • Given the paper's own depth-motivation, MI may track a notion of relational depth; comparing MI against the earlier depth score on the same relations would show whether MI adds predictive information beyond depth alone.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper introduces a Maskability Index (MI) computed from the difference in DepthRank under masked versus prefix-style prompting, and uses it to partition nine ATOMIC2020 relations into Mask-Filling (MI>=30%) and Map-Phrasal groups. The central claim is that MI predicts which template family yields better downstream generation quality for a relation, with masked prompting better for Mask-Filling relations and prefix prompting better for Map-Phrasal relations. Validation is presented via group-level ROUGE/BERTScore tables for T5-base, supported by qualitative arguments about tail lexical properties.

Significance. If established, MI would be a cheap, model-intrinsic diagnostic for template selection in few-shot knowledge-base completion: it requires only the model's token-rank probabilities, not expensive generation metrics. The paper uses a standard benchmark (ATOMIC2020) and reports concrete per-relation numbers, which is commendable. However, the current evidence does not firmly support the headline predictive claim. The validation is a coarse group comparison with a post-hoc threshold, no per-relation correlation with downstream performance is reported, one Map-Phrasal relation directly contradicts the claimed direction, and the acknowledged token-frequency confound is not controlled. The paper's explicit discussion of limitations is a strength, but the limitations are central to the validity of the metric rather than minor caveats.

major comments (3)
  1. [§5.1/§6, Tables 2 and 4] The central validation claim 'map-phrasal relations perform better under prefix prompting' is contradicted by one of the three relations in that group. xWant is assigned to Map-Phrasal (MI=19.15%), but Table 4 reports ROUGE 0.35 under Masked Prompting versus 0.32 under Prompting, and BERTScore 0.50 under both. Thus the direction of the effect is wrong for a third of the Map-Phrasal group. Since the group contains only three relations, this is not a minor exception; the paper needs per-relation evidence, such as effect sizes for each relation or a correlation coefficient between MI and the performance gap, before claiming MI predicts template success.
  2. [§5.1/§6, Eq. (1) and Table 2] The threshold MI>=0.30 at n=5 is introduced in Eq. (1) and then used to partition the data; the validation in Section 6 reports only group averages for the resulting two groups. No correlation coefficient between MI and ROUGE/BERTScore is reported, no confidence intervals or significance tests accompany the group differences, and the threshold is selected post hoc rather than on an independent split. The smallest Map-Phrasal MI (xNeed, 3.92%) sits close to the threshold, so the grouping is fragile. The paper's statement that 'Both metrics exhibit a strong positive correlation (r > 0.9 across all relations)' refers to ROUGE versus BERTScore, not to MI versus downstream performance, and therefore does not support the abstract's claim of a positive correlation with generation quality.
  3. [§2 and §7, Table 1] The DepthRank measurements that define MI may be dominated by surface lexical properties rather than template–objective alignment. The manuscript concedes in §7 that 'very frequent tokens like "to" can bias DepthRank averages.' The illustrative case in Table 1 shows xIntent tails beginning with token 'to' (indices 1,53,1), while xAttr tails are single adjectives at ranks 1672/1136. Because xIntent and xNeed fall in the Map-Phrasal group and their tails systematically start with high-frequency tokens, the MI contrast may reflect token frequency and tail length rather than masked-versus-prefix alignment. Without frequency-normalized MI, tail-length controls, or a test on alternative templates, the paper has not excluded this confound as the driver of the grouping and the observed validation pattern.
minor comments (6)
  1. [Abstract and §6] The abstract claims MI 'correlates with downstream generation performance,' but the only correlation coefficient reported in §6 is between ROUGE and BERTScore (r>0.9). Clarify whether a correlation between MI and performance was actually computed, and if so, report it with confidence intervals.
  2. [Figure 1] Figure 1 shows average DepthRank as a function of sample size, but no error bars or variance measures are provided, and the definition of 'sample size' (number of heads, repeated subsamples, etc.) is not fully specified. This makes it difficult to assess whether the trends for individual relations are reliable.
  3. [§4] The description of T5-base as 'pretrained with a denoising objective and an auxiliary language modeling objective, followed by supervised fine-tuning on downstream reasoning templates' is not standard for T5-base, which is pretrained with a span-corruption objective. Please clarify or correct this model description.
  4. [Table 2] MI values are reported at n=5 without any measure of variability across subsamples. Given that xNeed (3.92%) lies near the 30% threshold, reporting standard errors or a range over random subsets would help assess the stability of the group assignments.
  5. [Appendix A, Table 5] The template examples for xAttr and isFilledBy appear to include full sentences before the relation phrase (e.g., 'PersonX teaches at a university. PersonX is seen as intelligent'). Clarify exactly how the encoder input is constructed for the masked and prompted conditions, including where the mask token is placed.
  6. [§5.1] Qualitative statements about 'short lexical tails' versus 'longer phrasal realizations' are not supported by quantitative tail-length statistics. Adding average tail token counts for the two MI groups would make the argument more precise.

Circularity Check

0 steps flagged

No circular derivation: MI and ROUGE/BERTScore are distinct measurements, and no fitted parameter or self-citation chain forces the claimed predictive relationship.

full rationale

The derivation chain is not circular. MI (Eq. 1) is defined from DepthRank token-rank differences between masked and prefix templates, while the Section 6 validation uses ROUGE and BERTScore on generated text. These are different measurements, so the reported correlation is not an identity forced by construction. The 30% threshold is an unconstrained grouping choice, but there is no evidence in the paper that it was fitted to the downstream ROUGE/BERTScore outcomes; an arbitrary threshold weakens external validity but does not make the prediction definitional. The only self-citation (Ref. [18], same first author) appears in a motivational sentence alongside UL2 [25] and is not load-bearing: the paper's own experiments, not the citation, carry the argument. The central empirical claim is nevertheless questionable for correctness reasons rather than circularity: Table 4 shows xWant (MI=19.15%, assigned to Map-Phrasal) has higher ROUGE under Masked Prompting (0.35) than Prompting (0.32), contradicting the group-level assertion that map-phrasal relations favor prefix prompting. Section 7 also concedes that frequent tokens like 'to' can bias DepthRank averages, a confound that affects the measurement but not the independence of the MI-versus-ROUGE comparison. These are falsification and robustness concerns, not circular reductions, so the circularity score is 0.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

The central claim rests on one hand-set threshold and several domain assumptions about what DepthRank measures; no new physical or ontological entities are posited. The main unearned input is the assumption that small-sample DepthRank gaps on manually chosen verbalizer templates reflect objective-template alignment rather than lexical confounds.

free parameters (2)
  • MI threshold (30%) = 0.30
    Hand-selected cutoff in Section 3.3 used to label relations Mask-Filling vs Map-Phrasal. No derivation is given; with xNeed at 3.92 and xWant at 19.15, a different threshold would change the map-phrasal group and the validation result.
  • Few-shot sample size n=5 = 5
    MI is computed at n=5 because differences are claimed to be more informative in low-resource settings, but this choice is not justified or varied in the validation.
axioms (4)
  • domain assumption DepthRank of gold tail tokens measures how accessible a relation is under a prompting style
    Invoked in Sections 2-3.2 to define the quantity MI is built from; if rank position is not a faithful proxy for generation quality, MI loses meaning.
  • domain assumption T5-base's denoising and auxiliary LM objectives make masked-vs-prefix a valid alignment contrast
    Section 4 asserts T5 was pretrained with denoising plus auxiliary LM and is 'suitable'; no evidence is given that this specific combination justifies the template-family comparison.
  • domain assumption Gold tails in ATOMIC2020 are the correct reference for measuring generation quality
    Used to compute both DepthRank and ROUGE/BERTScore; KBC references can have multiple valid tails, and scoring against a limited reference set may underestimate generation quality.
  • domain assumption Relation-level MI computed at n=5 transfers to the 100-head held-out set used for performance evaluation
    The paper computes MI on a small sample (Section 3.3) and validates on held-out sets; the stability of the small-sample estimate across the relation is not measured.

pith-pipeline@v1.3.0-alltime-deepseek · 7809 in / 11330 out tokens · 104590 ms · 2026-08-01T10:17:31.898176+00:00 · methodology

0 comments
read the original abstract

Large-scale pretrained language models such as T5 and BERT have demonstrated strong capabilities for generating structured knowledge. However, their performance depends on how closely the prompting strategy matches the objectives used during pretraining. We introduce the Maskability Index (MI), a quantitative metric that estimates whether a knowledge relation is better suited to masked-style prompting or prefix-style prompting in few-shot generation. MI is computed from differences in DepthRank scores between masked and unmasked templates, providing a principled measure of objective-template alignment. We evaluate MI on a diverse set of relations from the ATOMIC2020 knowledge base completion benchmark and show that it is positively correlated with downstream generation performance. These results indicate that MI can help select appropriate prompting templates and adaptation strategies for extracting relational knowledge from pretrained language models, especially in low-resource settings.

Figures

Figures reproduced from arXiv: 2607.20265 by Ahmad Pouramini, Mahsa Afsharzadeh.

Figure 1
Figure 1. Figure 1: Average DepthRank of different ATOMIC2020 relations as a function of the number of [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

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

31 extracted references · 7 linked inside Pith

  1. [1]

    Comet: Commonsense transformers for automatic knowledge graph con- struction

    Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chaitanya Malaviya, Asli Celikyilmaz, and Yejin Choi. Comet: Commonsense transformers for automatic knowledge graph con- struction. InACL, 2019

    2019

  2. [2]

    Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D. Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Lan- guage models are few-shot learners. InAdvances in Neural Information Processing Systems (NeurIPS), volume 33, pages 1877–1901, 2020. The Maskability Index: Predicting Task–Objective Alignmen...

    1901

  3. [3]

    Albert with knowledge graph encoder utilizing semantic similarity for commonsense question answering

    Byeongmin Choi, Yong-Sook Lee, Yeunwoong Kyung, and Eunchan Kim. Albert with knowledge graph encoder utilizing semantic similarity for commonsense question answering. arXiv preprint arXiv:2211.07065, 2022

    Pith/arXiv arXiv 2022

  4. [4]

    Estimating commonsense plau- sibility through semantic shifts.arXiv preprint arXiv:2502.13464, 2025

    Wanqing Cui, Keping Bi, Jiafeng Guo, and Xueqi Cheng. Estimating commonsense plau- sibility through semantic shifts.arXiv preprint arXiv:2502.13464, 2025

    Pith/arXiv arXiv 2025

  5. [5]

    Joe Davison, Joshua Feldman, and Alexander M. Rush. Commonsense knowledge mining from pretrained models. InEMNLP-IJCNLP, pages 1173–1178, 2019

    2019

  6. [6]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. InNAACL, 2019

    2019

  7. [7]

    Making pre-trained language models better few-shot learners

    Tianyu Gao, Adam Fisch, and Danqi Chen. Making pre-trained language models better few-shot learners. InProceedings of the 59th Annual Meeting of the Association for Com- putational Linguistics (ACL), volume 1, pages 3816–3830. Association for Computational Linguistics, 2021

    2021

  8. [8]

    Accent: An automatic event commonsense evaluation metric for open-domain dialogue systems

    Sarik Ghazarian, Yijia Shao, Rujun Han, Aram Galstyan, and Nanyun Peng. Accent: An automatic event commonsense evaluation metric for open-domain dialogue systems. In ACL, 2023

    2023

  9. [9]

    Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, and Yejin Choi

    Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, and Yejin Choi. Comet-atomic 2020: On symbolic and neural commonsense knowledge graphs. InProceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI), pages 6384–6392, 2021

    2020

  10. [10]

    Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, and Yejin Choi

    Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, and Yejin Choi. Comet-atomic 2020: On symbolic and neural commonsense knowledge graphs. InAAAI, 2020

    2020

  11. [11]

    Can you tell me how to improve my prompt? learning to rephrase prompts for language models

    Zhengbao Jiang, Frank F Xu, Jun Araki, and Graham Neubig. Can you tell me how to improve my prompt? learning to rephrase prompts for language models. InProceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5925–

  12. [12]

    Maieutic prompting: Logically consistent reasoning with recursive explanations

    Jaehun Jung, Lianhui Qin, Sean Welleck, Faeze Brahman, Chandra Bhagavatula, Ronan Le Bras, and Yejin Choi. Maieutic prompting: Logically consistent reasoning with recursive explanations. InEMNLP, pages 1266–1279, 2022

    2022

  13. [13]

    Smith, Yejin Choi, and Hannaneh Hajishirzi

    Jiacheng Liu, Wenya Wang, Dianzhuo Wang, Noah A. Smith, Yejin Choi, and Hannaneh Hajishirzi. Vera: A general-purpose plausibility estimation model for commonsense state- ments. InEMNLP, pages 1264–1287, 2023

    2023

  14. [14]

    Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing.arXiv preprint arXiv:2107.13586, 2021

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neu- big. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing.arXiv preprint arXiv:2107.13586, 2021

    Pith/arXiv arXiv 2021

  15. [15]

    Constrained text gen- eration with global guidance: Case study on commongen.arXiv preprint arXiv:2103.07170, 2021

    Yixian Liu, Liwen Zhang, Wenjuan Han, Yue Zhang, and Kewei Tu. Constrained text gen- eration with global guidance: Case study on commongen.arXiv preprint arXiv:2103.07170, 2021

    Pith/arXiv arXiv 2021

  16. [16]

    A robustly optimized bert pre-training approach with post-training

    Zhuang Liu, Wayne Lin, Ya Shi, and Jun Zhao. A robustly optimized bert pre-training approach with post-training. InChina National Conference on Chinese Computational Linguistics, pages 471–484. Springer, 2021. 10 Ahmad Pouramini, Mahsa Afsharizadeh

    2021

  17. [17]

    Language models as knowledge bases? InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing, pages 2463–2473, 2019

    Fabio Petroni, Tim Rockt¨ aschel, Sebastian Riedel, Patrick Lewis, Anton Bakhtin, Yuxiang Wu, and Alexander Miller. Language models as knowledge bases? InProceedings of the 2019 Conference on Empirical Methods in Natural Language Processing, pages 2463–2473, 2019

    2019

  18. [18]

    Matching tasks to objectives: Fine-tuning and prompt-tuning strategies for encoder-decoder pre-trained language models.Applied In- telligence, 54(20):9783–9810, Oct 2024

    Ahmad Pouramini and Hesham Faili. Matching tasks to objectives: Fine-tuning and prompt-tuning strategies for encoder-decoder pre-trained language models.Applied In- telligence, 54(20):9783–9810, Oct 2024

    2024

  19. [19]

    Language models are unsupervised multitask learners

    Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. Language models are unsupervised multitask learners. Technical report, OpenAI, 2019. OpenAI Blog/Technical report

    2019

  20. [20]

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

    2020

  21. [21]

    Smith, and Yejin Choi

    Maarten Sap, Ronan Le Bras, Emily Allaway, Chandra Bhagavatula, Nicholas Lourie, Han- nah Rashkin, Brendan Roof, Noah A. Smith, and Yejin Choi. Atomic: An atlas of machine commonsense for if-then reasoning. InAAAI, pages 3027–3035, 2019

    2019

  22. [22]

    Auto- prompt: Eliciting knowledge from language models with automatically generated prompts

    Taylor Shin, Yasaman Razeghi, Robert L Logan IV, Eric Wallace, and Sameer Singh. Auto- prompt: Eliciting knowledge from language models with automatically generated prompts. InProceedings of the 2020 Conference on Empirical Methods in Natural Language Process- ing (EMNLP), pages 4222–4235. Association for Computational Linguistics, 2020

    2020

  23. [23]

    Conceptnet 5.5: An open multilingual graph of general knowledge

    Robyn Speer, Joshua Chin, and Catherine Havasi. Conceptnet 5.5: An open multilingual graph of general knowledge. InAAAI, 2016

    2016

  24. [24]

    Entailer: Answering questions with faithful and truthful chains of reasoning

    Oyvind Tafjord, Bhavana Dalvi, and Peter Clark. Entailer: Answering questions with faithful and truthful chains of reasoning. InEMNLP, 2022

    2022

  25. [25]

    Ul2: Unifying lan- guage learning paradigms.arXiv preprint arXiv:2205.05131, 2022

    Yi Tay, Mostafa Dehghani, Vinh Q Tran, Xavier Garcia, Jason Wei, Xuezhi Wang, Hyung Won Chung, Siamak Shakeri, Dara Bahri, Tal Schuster, et al. Ul2: Unifying lan- guage learning paradigms.arXiv preprint arXiv:2205.05131, 2022

    Pith/arXiv arXiv 2022

  26. [26]

    Harnessing black-box control to boost com- monsense in lms’ generation

    Yufei Tian, Felix Zhang, and Nanyun Peng. Harnessing black-box control to boost com- monsense in lms’ generation. InEMNLP, 2023

    2023

  27. [27]

    Trinh and Quoc V

    Trieu H. Trinh and Quoc V. Le. A simple method for commonsense reasoning.arXiv preprint arXiv:1806.02847, 2018

    Pith/arXiv arXiv 2018

  28. [28]

    Retrieval augmentation for commonsense reasoning: A unified approach

    Wenhui Yu, Chenguang Zhu, Zhihan Zhang, Shuohang Wang, Zhuosheng Zhang, Yuwei Fang, and Meng Jiang. Retrieval augmentation for commonsense reasoning: A unified approach. InEMNLP, 2022

    2022

  29. [29]

    Alleviating the knowledge-language inconsistency: A study for deep commonsense knowledge.arXiv preprint arXiv:2105.13607, 2021

    Yi Zhang, Le Li, Yuyang Wu, Qiang Su, and Xiao Sun. Alleviating the knowledge-language inconsistency: A study for deep commonsense knowledge.arXiv preprint arXiv:2105.13607, 2021

    Pith/arXiv arXiv 2021

  30. [30]

    Large language models as commonsense knowl- edge for large-scale task planning

    Zirui Zhao, Wee Sun Lee, and David Hsu. Large language models as commonsense knowl- edge for large-scale task planning. InNeurIPS, 2024. The Maskability Index: Predicting Task–Objective Alignment in Pretrained Language Models 11 A Prompts for Knowledge Base Relations To convert knowledge-graph triples into natural language sentences, we used the relation-...

    2024

  31. [5936]

    Association for Computational Linguistics, 2020

    2020