Mitigating Gender Bias in Contextual Word Embeddings

Feng-Guang Su; Navya Yarrabelly; Vinay Damodaran

REVIEW 5 major objections 5 minor 30 references

Mitigating Gender Bias in Contextual Word Embeddings

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A training-time objective that blocks gendered words when predicting neutral nouns largely removes gender bias from BERT while preserving downstream performance, and the paper argues that stereotypical names, not gendered words, are the…

desk verdict The static NER-M+EGE result is the real story here; the BERT debiasing claim is not established by the current evaluation. read the letter →

arxiv 2411.12074 v1 pith:TXMLW3RP submitted 2024-11-18 cs.CL cs.LG

Navya Yarrabelly , Vinay Damodaran , Feng-Guang Su This is my paper · ORCID

classification cs.CLcs.LG

keywords genderbiascontextualembeddingsmaskedlanguagemodelingdebiasingBERTstaticwordnamedentitymaskingstereotypicalnames

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 sets out to show that gender bias in contextual embeddings can be reduced at training time by changing the masked-language-model objective, rather than by post-processing or heavy data augmentation. It proposes an objective in which the model predicts masked nouns without attending to gendered attribute words, and predicts masked attribute words without attending to nouns, plus a regularizer that pushes prediction scores for paired male and female tokens toward equality. The paper also investigates static embeddings and argues that the persistent clustering of stereotyped professions is driven mainly by co-occurring stereotypical names, not by gendered words themselves, and that masking those names while explicitly encoding gender categories brings profession clustering close to chance.

What carries the argument

The central object is a modified masked-language-model training procedure with two attention-mask strategies: when predicting a masked noun, the model is allowed to attend only to non-attribute tokens, and when predicting a masked attribute word, it may attend only to non-noun tokens. A regularizer adds the summed absolute difference of the pre-softmax prediction scores for paired male and female tokens, pushing the model to treat paired gendered words symmetrically for gender-neutral nouns. For static embeddings, the load-bearing mechanisms are an entity-mask corpus (NER-M) that replaces stereotypical male and female names with a token, and an explicit gender-encoding regularizer (EGE) that trains a CBOW model to classify center words as male, female, or neutral; when applied together, they separate professions from name-mediated gender clusters.

What would settle it

Train a linear or shallow classifier on the debiased BERT's representations of neutral nouns (for example, 'nurse') to predict the gender of an attribute word or named entity that appears elsewhere in the sentence; if the classifier achieves accuracy well above chance, gender information has survived the attention masking and the debiasing is superficial. For the static-embedding claim, run the Gonen-Goldberg clustering experiment on a corpus with names masked versus a corpus with gendered words removed but names intact; if clustering stays high when names remain, names are indeed the carrier.

Watch

Extended reading notes

Core claim

The central claim is that a regularized MLM objective can mitigate gender bias in BERT's contextual embeddings while keeping downstream performance intact, and that the main carrier of gender bias in static embeddings is stereotypical names rather than gendered words. The paper supports the first claim with pronoun-filling and occupation-filling experiments, lower stereotype scores on WinoBias, and retained accuracy on SST-2, CoLA, and QNLI. For static embeddings, it reports that combining NER-Masking of stereotypical names with an explicit gender-encoding regularizer lowers the Gonen-Goldberg profession-clustering accuracy to 0.5157, near chance, whereas removing gendered words alone leaves clustering at 0.5970.

Load-bearing premise

The contextual debiasing works only if blocking attention to gendered words prevents the model from encoding gender in the neutral noun's representation; if the remaining context still lets a probe recover the masked word's gender, the bias is hidden, not removed.

Editorial extensions

If this is right

BERT-style models can be debiased during continued pre-training without a large drop in downstream performance.
The proposed evaluation metrics give a direct, normatively motivated way to check whether a model is treating paired gendered words symmetrically in context.
For static embeddings, masking stereotypical names appears more effective than removing gendered words, pointing to named entities as the primary bias carrier.
Combining entity masking with explicit gender encoding reduces profession clustering to near chance, a result other debiasing methods do not reach.
Existing post-hoc debiasing methods fail to reduce profession clustering because they do not address the name-mediated transitive path of bias.

Reading between the lines

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

If stereotypical names are the primary bias carrier, then masking or rebalancing named entities in training corpora could become a general preprocessing step for other social biases, not just gender.
The contextual method's reliance on attention masks suggests a natural test of its depth: a probing classifier should not be able to recover the gender of a masked attribute word from the remaining context, a test the paper does not run.
The paper's own layer analysis hints that gender information may still be encoded in some BERT layers after debiasing, which would make the contextual method closer to suppression than elimination.
The proposed augmentation strategy, which swaps gendered labels and masks neutral words, may strip so much gender information that the model becomes unable to predict pronouns at all, a side effect observed in the paper's preliminary analysis.

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.

Desk Editor's Note

read the letter

Quick take: this paper has one genuinely interesting empirical result—NER-masking plus explicit gender encoding drops profession clustering accuracy to 0.5157 in static embeddings—and a new-looking regularized MLM objective for BERT. But the central claim that the objective debiases contextual embeddings is not backed by the evidence as written. The intrinsic evaluation largely measures what the regularizer optimizes, and no probe tests whether gender is still recoverable from unmasked context at inference. I'd send it to review only with the expectation of major revision or rejection; the static result is worth chasing but the contextual half needs real work.

What's new: the attention-masked MLM with a logit-difference regularizer (Eq. 1) is not in the cited prior work. The combination of NER-masking and explicit gender encoding for static embeddings is also new, and the near-chance clustering is a clean demonstration that names can carry more bias signal than gendered words. The paper also does the right thing by checking downstream tasks (SST-2, CoLA, QNLI) and by engaging directly with the Gonen–Goldberg critique.

Soft spots, in order of severity. First, the contextual evaluation is partly circular: the main metric—probability of he/she given an occupation, and vice versa—is precisely what the regularizer drives toward equality for gender-neutral words. The paper acknowledges this alignment but doesn't treat it as a limitation. Second, the training-time attention mask doesn't guarantee inference-time debiasing; gender information can route through other tokens. The paper invokes Gonen–Goldberg for static embeddings but runs no probing or recovery experiment for BERT. Third, the BERT continued pre-training is underspecified: no base model, steps, learning rate, loss weight, or compute. Fourth, reproducibility suffers from missing code/data and unfinished placeholders ('Figure ??', 'sec. ??', 'Assignment 3'). Fifth, the 'empirical proof' that names are the main source of static bias lacks a control experiment for general token masking—masking any frequent entity type might break spurious correlations.

The static experiment is the strongest part, and the NER-M+EGE result deserves follow-up. But as a standalone paper, the contextual claims overreach. I'd recommend that a serious editor send it to referees—there's enough substance to warrant detailed comments—but the outcome should be reject-and-resubmit or major revision, not acceptance.

Referee Report

5 major / 5 minor

Summary. The manuscript proposes two families of debiasing methods. For contextual embeddings (BERT), it introduces a regularized masked-language-model objective (Eq. 1) with attention masks that prevent the model from using attribute words when predicting nouns and vice versa, plus a gender-prediction data-augmentation strategy. It evaluates these with pronoun/occupation prediction tasks, CrowS-pairs-style scoring, SEAT, and downstream GLUE tasks. For static embeddings, it proposes gender-neutral corpus rewriting, explicit gender encoding (EGE), and masking stereotypical names (NER-M); the headline static result is that NER-M plus EGE reduces Gonen-Goldberg clustering accuracy from about 0.58 to 0.5157. The paper also argues that stereotypical names, not gendered words, are the main source of static gender bias.

Significance. If correct, the regularized MLM would be a practical training-time debiaser for BERT, and the NER-M+EGE result would provide a striking confirmation of the names-as-carriers hypothesis. The paper is honest about some limitations and makes a falsifiable prediction in the clustering experiment. However, the contextual claim is not yet supported because the main intrinsic metric is essentially the training objective and no recovery or probing experiment is reported; the static causal claim is undercut by the ablation pattern. With additional controls and targeted experiments, the ideas could become a contribution, but in the present form the central claims outrun the evidence.

major comments (5)

[Section 3.2.1, Eq. (1), Fig. 1; Section 3.4.1] The first intrinsic evaluation is circular with respect to the training objective. Eq. (1) regularizes the model so that f(x_i) and f(y_i) are close for paired gender tokens, and Fig. 2 reports exactly these prediction scores for he/she, him/her, his/her across occupations. The attention mask in Fig. 1 is applied only during continued pre-training; at inference the model is unmasked, so gender information can be rerouted through names, other nouns, adjectives, or verbs. The paper itself invokes the Gonen-Goldberg recoverability critique for static embeddings in Section 4.1 and shows in Tables 7 and 8 that names remain strong gender carriers, yet no probing or recovery experiment is run on BERT. Without a probe (e.g., training a gender classifier on the contextual representations of a masked occupation from the Winogender templates), the abstract's claim that the objective eliminates the dependencies is unsupported.
[Section 4.4.1, Table 4] The static claim that names are the main source of bias is not supported by the ablations. NER masking alone yields 0.5839 ± 0.0356, no better than the original 0.5802 ± 0.0282; EGE alone yields 0.5743 ± 0.0106; only the combination NER-M+EGE drops to 0.5157 ± 0.0050. This is an interactive effect, not evidence that names alone carry the bias. The conclusion requires (i) a random-token masking baseline matched for the number of masked tokens, (ii) ablations that mask stereotypical names while keeping gendered words, and (iii) a significance test across the 15 runs. As reported, the experiments do not isolate names or gendered words as the causal carrier.
[Section 3.4.3, Table 3] The SEAT analysis is difficult to verify and the conclusion that SEAT is not an effective measure does not follow from the displayed numbers. The column header 'BERT Debiased-bert ours SENT-DEBIAS BERT' is malformed; rows such as C6b and C7b appear to show p-values below 0.05 for the debiased model (0.03 and 0.04), yet the text says that 4/6 sets are not statistically significant. Because the paper uses this conclusion to justify abandoning SEAT in favor of the proposed metrics, the table must be unambiguous about which model each effect size and p-value belong to, and the counting must be consistent with the table.
[Section 3.2.1 and Section 3.3] The continued pre-training experiments are not reproducible. There is no report of the BERT variant, number of training steps or epochs, batch size, learning rate, mask probability, the exact attribute-word list, or the weight of the regularizer in Eq. (1). Section 3.3 names BookCorpus but gives no training schedule. Without these details, the reader cannot assess whether the regularizer dominates the masked-LM loss or how much the debiasing costs in model quality. This is especially important because the paper's contextual method is the primary contribution.
[Section 3.4.2, Table 1] The bias-score reductions on CrowS-pairs-style evaluation are reported without any uncertainty or significance testing; for example, Type1 test values 53.54 vs 51.52 and Type2 test values 60.61 vs 54.04 are presented as improvements with no error bars, number of runs, or paired test. Given the known variance of template-based stereotype scores, these differences may not be reliable. The caption and text should also state consistently that lower values are better, since Table 1 already says that higher numbers indicate higher model bias.

minor comments (5)

[Appendix A.1.1 and A.1.2] There are unresolved cross-references: Section A.1.1 refers to 'sec. ??' and Appendix A.1.2 refers to 'Figure ??(a)' and 'Figure ??(b)', which should be resolved before submission.
[Section 4.4.2, Section 1, Appendix A.3.3, Table 7] There are several typos: 'intuion' in Section 4.4.2, 'debasing' in Section 1, 'smenatics' in Appendix A.3.3, and 'explicity' in the Table 7 caption.
[Section 3.4.3] The text mentions 'shown in Assignment 3,' which appears to refer to an external course assignment; this should be removed or replaced with a citation or an appendix section.
[Section 3.4.1] The criterion that target occupations appear 'in the top 30000 predicted words' needs a motivation and a statement of the BERT vocabulary size; otherwise the effective test set is unclear.
[Table 2] The downstream task results are reported without standard deviations across fine-tuning runs; the claimed 0.8% gain on SST-2 (93.4 vs 92.6) is within typical run-to-run variability for GLUE fine-tuning.

Circularity Check

1 steps flagged · score 6.0 of 10

Contextual debiasing is partially validated by a metric that restates the regularizer; the static-embedding name-carrier result is empirical and not circular.

self definitional [Abstract; Sec. 3.2.1 Eq. (1); Sec. 3.4.1 Fig. 2]
"Since previous works on measuring bias in contextual embeddings lack in normative reasoning, we also propose novel evaluation metrics that are straight-forward and aligned with our motivations in debiasing. ... Given pairs of gendered tokens TA = {(x1, y1), (x2, y2), ...(xs, ys)} where xi represents male tokens while yi represents the corresponding female tokens, we propose a newly-designed regularizer when calculating the cross-entropy losses of gender-neutral tokens."

Eq. (1) is applied when computing cross-entropy losses of gender-neutral tokens: for each masked gender-neutral noun it penalizes the difference between the LM's prediction scores f(x_i) and f(y_i) for paired male/female attribute words. Fig. 2's intrinsic-bias metric masks he/she pronouns in occupation templates and declares the model less biased when the predicted probabilities of he and she are closer together. Thus the headline quantitative evidence for "largely mitigates gender bias" measures the same paired-attribute score difference the regularizer was constructed to drive to zero. This is a training-objective restatement rather than an independent probe of the deployed unmasked model.

full rationale

The strongest circular step is the contextual intrinsic evaluation: the paper's own abstract says the new metrics are "aligned with our motivations," and Eq. (1)'s regularizer directly equalizes the he/she prediction scores that Fig. 2 then presents as evidence of debiasing. This makes the central BERT mitigation claim, insofar as it rests on Figs. 2–3, partially a check of the training loss. However, the paper also reports an external CrowS-pairs stereotype score (Table 1), downstream GLUE-style tasks (Table 2), and a layerwise probing analysis, so the claim is not entirely forced. The static-embedding section is empirical rather than circular: NER-M + EGE's near-chance clustering accuracy and the nanny/molly neighbour tables are external quantitative results, not restatements of the training objective. There is no load-bearing self-citation chain or uniqueness theorem imported from the authors. The recoverability objection raised by the skeptical reader (gender may be routed through names or other context tokens) is a correctness gap rather than a circularity reduction; the paper's own Tables 7–8 show names remain gender carriers after debiasing, which supports that concern. Overall score 6 reflects one by-construction reduction in the principal intrinsic evaluation, while acknowledging the independent benchmarks prevent a higher score.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The methods rely on several hand-chosen hyperparameters (regularization strength, masking percentages, name frequency cutoff) and on operational definitions of gender-neutrality and gender direction. The binary-gender framing and the assumption that NER masking is a neutral intervention are load-bearing and untested.

free parameters (3)

EGE regularizer weight = 0.5
Set as a constant in Section 4.3; no sensitivity analysis or ablation reported.
Gender-neutral word masking percentage (Strategy 1) = 40%
Chosen in Section 3.2.2; no sweep.
Name frequency cutoff for stereotypical name list = >10,000 occurrences since 1880
Used in Section 4.3 to curate the NER-mask list; arbitrary and untested.

assumptions (4)

domain assumption Gender is binary and captured by a predefined list of male/female word pairs (TA).
Used in Eq. (1) and in the EGE regularizer (Section 3.2.1, 4.2.2); excludes non-binary and context-dependent genders.
domain assumption All nouns except attribute words are gender-neutral.
Operational definition in Section 3.2.1; ignores nouns with inherent gender and polysemy.
domain assumption Masking stereotypical names with an entity token removes gender information while preserving other semantics.
Underpins the NER-M method and the causal claim in Section 4.2.3; no control for general token-masking effects.
domain assumption The gender direction computed via PCA on paired names (Dev and Phillips 2019) is the correct bias measure.
Used for direct bias and clustering evaluations in Section 4.4.1 and A.3.2.

invented entities (1)

Entity mask token for names
purpose: Replaces stereotypical names during CBOW training to prevent name-gender co-occurrence from biasing profession embeddings.
This is a preprocessing token, not a new physical or conceptual entity; its sole evidence is the in-paper clustering result.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Mitigating Gender Bias in Contextual Word Embeddings." pith.science (2026). https://pith.science/paper/TXMLW3RP

@misc{pith2026241112074,
  author       = {Pith},
  title        = {Pith review of: Mitigating Gender Bias in Contextual Word Embeddings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TXMLW3RP}},
  note         = {Machine review of arXiv:2411.12074}
}

read the original abstract

Word embeddings have been shown to produce remarkable results in tackling a vast majority of NLP related tasks. Unfortunately, word embeddings also capture the stereotypical biases that are prevalent in society, affecting the predictive performance of the embeddings when used in downstream tasks. While various techniques have been proposed \cite{bolukbasi2016man, zhao2018learning} and criticized\cite{gonen2019lipstick} for static embeddings, very little work has focused on mitigating bias in contextual embeddings. In this paper, we propose a novel objective function for MLM(Masked-Language Modeling) which largely mitigates the gender bias in contextual embeddings and also preserves the performance for downstream tasks. Since previous works on measuring bias in contextual embeddings lack in normative reasoning, we also propose novel evaluation metrics that are straight-forward and aligned with our motivations in debiasing. We also propose new methods for debiasing static embeddings and provide empirical proof via extensive analysis and experiments, as to why the main source of bias in static embeddings stems from the presence of stereotypical names rather than gendered words themselves. All experiments and embeddings studied are in English, unless otherwise specified.\citep{bender2011achieving}.

Figures

Figures reproduced from arXiv: 2411.12074 by the authors.

**Figure 2.** The predicted probabilities of gendered pronouns (he/she), object pronouns (him/her), or possessive [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The predicted probabilities of occupations based on similar contexts with different gendered pro [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (7 more)

**Figure 4.** Figure 4: Classification accuracy for each layer in BERT. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 5.** Figure 5: The gender clustering visualization for (a) [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: The word cloud visualizations for the neigh [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]

**Figure 7.** Figure 7: TSNE visualizations of the vectors during [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]

**Figure 8.** Figure 8: TSNE plot for neighbours of nurse (color [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]

**Figure 9.** Figure 9: Gender clustering of (a) Original, (b) NER-M + EGE embeddings. Notice the change in clusters for nurse and the increase in separation between hairdresser and nanny [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]

**Figure 10.** Figure 10: The word cloud visualizations for the word [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 6 canonical work pages

[1]

Emily M Bender. 2011. On achieving and evaluating language-independence in nlp. Linguistic Issues in Language Technology, 6(3):1--26

work page 2011
[2]

Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems, 29:4349--4357

work page 2016
[3]

Bryson, and Arvind Narayanan

Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. https://doi.org/10.1126/science.aal4230 Semantics derived automatically from language corpora contain human-like biases . Science, 356(6334):183–186

doi:10.1126/science.aal4230 2017
[4]

Sunipa Dev and Jeff Phillips. 2019. http://arxiv.org/abs/1901.07656 Attenuating bias in word vectors

work page Pith review arXiv 2019
[5]

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
[6]

Lucas Dixon, John Li, Jeffrey Sorensen, Nithum Thain, and Lucy Vasserman. 2018. Measuring and mitigating unintended bias in text classification. In Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, pages 67--73

2018
[7]

Hila Gonen and Yoav Goldberg. 2019. Lipstick on a pig: Debiasing methods cover up systematic gender biases in word embeddings but do not remove them. arXiv preprint arXiv:1903.03862

arXiv 2019
[8]

Eduard Hovy, Mitchell Marcus, Martha Palmer, Lance Ramshaw, and Ralph Weischedel. 2006. Ontonotes: The 90 In Proceedings of the Human Language Technology Conference of the NAACL, Companion Volume: Short Papers, NAACL-Short '06, page 57–60, USA. Association for Computational Linguistics

work page 2006

Show all 30 references

[9]

Vaibhav Kumar, Tenzin Singhay Bhotia, Vaibhav Kumar, and Tanmoy Chakraborty. 2020. http://arxiv.org/abs/2006.01938 Nurse is closer to woman than surgeon? mitigating gender-biased proximities in word embeddings

2020
[10]

Keita Kurita, Nidhi Vyas, Ayush Pareek, Alan W Black, and Yulia Tsvetkov. 2019. Measuring bias in contextualized word representations. arXiv preprint arXiv:1906.07337

2019 arXiv
[11]

Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, and Radu Soricut. 2020. http://arxiv.org/abs/1909.11942 Albert: A lite bert for self-supervised learning of language representations

2020 arXiv
[12]

Paul Pu Liang, Irene Mengze Li, Emily Zheng, Yao Chong Lim, Ruslan Salakhutdinov, and Louis-Philippe Morency. 2020. Towards debiasing sentence representations. arXiv preprint arXiv:2007.08100

2020 arXiv
[13]

Laurens Van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-sne. Journal of machine learning research, 9(11)

2008
[14]

Chandler May, Alex Wang, Shikha Bordia, Samuel R Bowman, and Rachel Rudinger. 2019. On measuring social biases in sentence encoders. arXiv preprint arXiv:1903.10561

2019 arXiv
[15]

Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781

2013 arXiv
[16]

Nikita Nangia, Clara Vania, Rasika Bhalerao, and Samuel R. Bowman. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.154 C row S -pairs: A challenge dataset for measuring social biases in masked language models . In Proceedings of the 2020 Conference on Empirical Methods in Na...

2020 doi
[17]

Jeffrey Pennington, Richard Socher, and Christopher Manning. 2014. https://doi.org/10.3115/v1/D14-1162 G love: Global vectors for word representation . In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing ( EMNLP ) , pages 1532--1543, Doha,...

2014 doi
[18]

ME Peters, M Neumann, M Iyyer, M Gardner, C Clark, K Lee, and L Zettlemoyer. 2018. Deep contextualized word representations. arxiv 2018. arXiv preprint arXiv:1802.05365

2018 arXiv
[19]

Rachel Rudinger, Jason Naradowsky, Brian Leonard, and Benjamin Van Durme . 2018. Gender bias in coreference resolution. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, New Orlean...

2018
[20]

Richard Socher, Alex Perelygin, Jean Wu, Jason Chuang, Christopher D Manning, Andrew Y Ng, and Christopher Potts. 2013. Recursive deep models for semantic compositionality over a sentiment treebank. In Proceedings of the 2013 conference on empirical methods in natural language...

2013
[21]

Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R Bowman. 2018. Glue: A multi-task benchmark and analysis platform for natural language understanding. arXiv preprint arXiv:1804.07461

2018 arXiv
[22]

Tianlu Wang, Xi Victoria Lin, Nazneen Fatema Rajani, Bryan McCann, Vicente Ordonez, and Caiming Xiong. 2020. Double-hard debias: Tailoring word embeddings for gender bias mitigation. arXiv preprint arXiv:2005.00965

2020 arXiv
[23]

Alex Warstadt, Amanpreet Singh, and Samuel R Bowman. 2019. Neural network acceptability judgments. Transactions of the Association for Computational Linguistics, 7:625--641

2019
[24]

Kellie Webster, Xuezhi Wang, Ian Tenney, Alex Beutel, Emily Pitler, Ellie Pavlick, Jilin Chen, and Slav Petrov. 2020. Measuring and reducing gendered correlations in pre-trained models. arXiv preprint arXiv:2010.06032

2020 arXiv
[25]

Jieyu Zhao, Tianlu Wang, Mark Yatskar, Ryan Cotterell, Vicente Ordonez, and Kai-Wei Chang. 2019. Gender bias in contextualized word embeddings. arXiv preprint arXiv:1904.03310

2019 arXiv
[26]

Jieyu Zhao, Tianlu Wang, Mark Yatskar, Vicente Ordonez, and Kai-Wei Chang. 2018 a . Gender bias in coreference resolution: Evaluation and debiasing methods. arXiv preprint arXiv:1804.06876

2018 arXiv
[27]

Jieyu Zhao, Yichao Zhou, Zeyu Li, Wei Wang, and Kai-Wei Chang. 2018 b . Learning gender-neutral word embeddings. arXiv preprint arXiv:1809.01496

2018 arXiv
[28]

Yukun Zhu, Ryan Kiros, Rich Zemel, Ruslan Salakhutdinov, Raquel Urtasun, Antonio Torralba, and Sanja Fidler. 2015. Aligning books and movies: Towards story-like visual explanations by watching movies and reading books. In The IEEE International Conference on Computer Vision (ICCV)

2015
[29]

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...
[30]

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 12, 2026 · model on record in the stance chip above.

[1] [1]

Emily M Bender. 2011. On achieving and evaluating language-independence in nlp. Linguistic Issues in Language Technology, 6(3):1--26

work page 2011

[2] [2]

Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems, 29:4349--4357

work page 2016

[3] [3]

Bryson, and Arvind Narayanan

Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. https://doi.org/10.1126/science.aal4230 Semantics derived automatically from language corpora contain human-like biases . Science, 356(6334):183–186

doi:10.1126/science.aal4230 2017

[4] [4]

Sunipa Dev and Jeff Phillips. 2019. http://arxiv.org/abs/1901.07656 Attenuating bias in word vectors

work page Pith review arXiv 2019

[5] [5]

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

[6] [6]

Lucas Dixon, John Li, Jeffrey Sorensen, Nithum Thain, and Lucy Vasserman. 2018. Measuring and mitigating unintended bias in text classification. In Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, pages 67--73

2018

[7] [7]

Hila Gonen and Yoav Goldberg. 2019. Lipstick on a pig: Debiasing methods cover up systematic gender biases in word embeddings but do not remove them. arXiv preprint arXiv:1903.03862

arXiv 2019

[8] [8]

Eduard Hovy, Mitchell Marcus, Martha Palmer, Lance Ramshaw, and Ralph Weischedel. 2006. Ontonotes: The 90 In Proceedings of the Human Language Technology Conference of the NAACL, Companion Volume: Short Papers, NAACL-Short '06, page 57–60, USA. Association for Computational Linguistics

work page 2006