Pith. sign in

REVIEW 5 major objections 5 minor 41 references

From Detection to Mitigation: Addressing Gender Bias in Chinese Texts via Efficient Tuning and Voting-Based Rebalancing

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A LoRA-tuned Qwen2.5 model with class-balanced expert voting detects, classifies, and mitigates Chinese gender bias, scoring 47.90% on the shared task and finishing fourth.

desk verdict Read this as a shared-task system description, not a methods paper: the ranked results are real, but the effectiveness claim is undercut by the Prompt baseline and missing ablations. read the letter →

arxiv 2509.07889 v1 pith:COGHDEWV submitted 2025-09-09 cs.CL

Chengyan Wu , Yiqiang Cai , Yufei Cheng , Yun Xue This is my paper · ORCID
classification cs.CL
keywords genderbiasChineseNLPLoRAlargelanguagemodelstextclassificationrewritingclassimbalancemajorityvoting
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 tries to establish that a complete gender-bias handling pipeline for Chinese can be built by efficiently adapting a general instruction-tuned LLM instead of training from scratch. Starting from the CORGI-PM corpus, the authors fine-tune Qwen2.5-7B-Instruct with LoRA, rebalance the strongly skewed training data by recombining biased and non-biased sentences into roughly 1:1 subsets, and train six expert models. For detection and classification, the experts' predictions are combined by majority vote; for rewriting, the model samples de-biased outputs at several temperatures. On the official evaluation the method reaches 47.90% average score and fourth place, with detection F1 of 0.720. A sympathetic reader would take this as evidence that lightweight LoRA plus data recombination is a workable recipe for practical bias-detection tools.

What carries the argument

The mechanism is a pipeline with three coordinated pieces. LoRA keeps the Qwen2.5-7B-Instruct backbone frozen and inserts small trainable low-rank matrices, so each subtask-specific expert adapts with few parameters. For classification, the 21,418 non-biased training sentences are split into five subsets and randomly combined with the 4,172 biased sentences at about 1:1 ratio, producing six experts; the final prediction is the majority answer when more than three of the six agree. For rewriting, generation temperatures of 0.01, 0.1, and 0.3 are sampled to produce stylistic variation, and the official overlap metrics are used to assess rewrites. A unified prompt with role, task description, and examples is shared across subtasks, and the bias-type prediction from subtask 2 is inserted into the rewriting prompt.

What would settle it

Run a human-annotation or independent-classifier study on the system's rewritten outputs: if sentences with high BLEU/METEOR/ROUGE-L are still judged to contain gender bias, the metric-based mitigation claim is refuted. A second decisive check is an ablation comparing the six-expert voting pipeline with a single expert trained on all data, using the development set's F1 scores.

Watch

Extended reading notes

Core claim

The central claim is that LoRA-based fine-tuning of a general-purpose instruction-tuned LLM, together with class-balanced data recombination and six-expert majority voting, is sufficient for competitive sentence-level gender-bias detection, classification, and mitigation in Chinese. On the official test set the model detects bias with F1 0.720, ranking second on that subtask; classifies bias type with F1 0.453; and scores 0.265 on rewriting as measured by the average of BLEU, METEOR, and ROUGE-L against reference rewrites. The authors attribute the detection result to their LoRA and learning-rate configuration, explain the classification gap by missing within-category imbalance and reuse of detection-oriented hyperparameters, and describe the temperature sweep on rewriting as exploratory rather than exhaustive. The overall result, an average of 47.90% and fourth place, is offered as confirmation that the proposed pipeline works in practice.

Load-bearing premise

The load-bearing premise is that measuring a rewrite's lexical overlap with one reference de-biased version through BLEU, METEOR, and ROUGE-L actually tells whether gender bias was removed; if that metric is not measuring bias removal, the subtask-3 scores do not support the mitigation claim.

Editorial extensions

If this is right

  • If these results hold, a frozen 7-billion-parameter LLM with LoRA adapters is enough for usable Chinese gender-bias detection, lowering the hardware and data barrier for fairness tooling.
  • The rebalancing-and-voting design implies that class imbalance can be handled by training several moderately balanced experts rather than by carefully reweighting a single model.
  • Since the three temperatures tested gave nearly identical mitigation scores, the paper's own results suggest temperature variation is not the main driver of rewrite quality in that range.
  • The authors' statement that they reused subtask-1 hyperparameters for subtask-2 implies the classification result underestimates what the same pipeline could achieve with task-specific tuning.
  • Because the rewriting prompt consumes subtask-2 bias-type predictions, classification errors propagate into mitigation, so improving the classifier should directly improve rewriting relevance.

Reading between the lines

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

  • A human-annotation comparison would probably reveal that lexical-overlap metrics reward surface similarity to a reference rewrite rather than actual de-biasing, so the subtask-3 score should be read as a proxy, not as proof of bias removal.
  • The paper never ablates the voting mechanism, so a single-expert baseline on the same full training set would separate the contribution of the ensemble from the contribution of simply using more data.
  • The flat temperature results suggest that a broader sweep or an adaptive sampling strategy might find gains the current design misses, or alternatively that the pipeline is already robust to this hyperparameter.
  • Since none of the mechanism components are gender-specific, the same recombination-and-voting recipe could be transferred to other bias axes such as occupation, race, or age by swapping the annotated corpus.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. This paper is a system description for NLPCC-2025 Shared Task 7 on sentence-level gender bias detection, classification, and mitigation in Chinese. The authors fine-tune Qwen2.5-7B-Instruct with LoRA, rebalance the training data by randomly combining biased and unbiased instances into subsets, train six expert models on these subsets, and apply majority voting for subtasks 1 and 2. For subtask 3, they vary the decoding temperature across {0.01, 0.1, 0.3}. On the official leaderboard the system achieves an average score of 0.479 and ranks fourth, behind the organizer's 'Prompt' baseline (0.502). The paper claims that the results 'demonstrate the effectiveness' of the proposed approach.

Significance. If the effectiveness claim were adequately supported, the paper would provide a practical recipe for adapting a 7B LLM to Chinese gender-bias tasks with modest compute. Its strengths are that the leaderboard scores are externally measured, the work uses a real annotated corpus (CORGI-PM), and the LoRA-based design is parameter-efficient. However, the central claim is not yet established: the proposed system underperforms the Prompt baseline on average and on subtask 2, and no ablation isolates the contribution of data recombination, majority voting, or multi-temperature sampling. The paper's own reported temperature sweep shows near-identical scores. The paper is a reasonable system-description draft, but the load-bearing evidence for 'effectiveness' is missing.

major comments (5)
  1. [§4.4, Table 2] The abstract and conclusion claim that the results 'demonstrate the effectiveness of the proposed approach,' but Table 2 shows that the organizer's Prompt baseline achieves a higher average score (0.502) than the proposed system (0.479), and on subtask 2 the gap is 0.505 vs. 0.453. A leaderboard rank alone does not establish effectiveness relative to a no-fine-tuning baseline. The authors should either temper the claim to 'competitive performance' or provide a controlled comparison on the dev set, e.g., the same Qwen2.5-7B-Instruct model with zero-shot prompting, with LoRA fine-tuning without recombination/voting, and with the full proposed pipeline.
  2. [§3.4] The majority voting component is asserted to 'enhance the generalization capability and stability of the model,' but no ablation is reported. There is no comparison of a single expert, a small ensemble, or a non-voting aggregation on the same dev set. Without such an ablation, the benefit of six-expert majority voting is unsubstantiated. Additionally, the tie-breaking procedure is underspecified: the sentence 'retains the original confidence scores for downstream processing' does not explain how a final prediction is produced when no majority of more than three experts is reached.
  3. [§3.3, Table 4] The multi-temperature sampling mechanism is claimed to improve bias mitigation, yet Table 4 shows that varying the temperature changes the subtask 3 score only from 0.394 to 0.392 over {0.01, 0.1, 0.3}. This is within noise and does not demonstrate any benefit. Moreover, the paper does not describe how multiple temperatures are combined: Section 3.3 says texts are generated with different styles by varying temperature, but Table 4 reports each temperature independently, not an ensemble. If multi-temperature sampling is intended as an ensemble, the integration procedure must be specified and its effect measured; otherwise the claim should be removed.
  4. [§4.3, Table 6] The evaluation of subtask 3 uses BLEU, METEOR, and ROUGE-L against a single reference rewrite (reported in Table 6 as BLEU 0.009, METEOR 0.391, ROUGE-L 0.367). These lexical-overlap metrics measure how closely the output resembles the reference text, not whether gender bias has actually been removed. A system that produces fluent but still-biased rewrites could score well if the reference is lexically close, while a genuinely debiased paraphrase could score poorly. The subtask 3 scores therefore do not support the 'mitigation' part of the effectiveness claim. At minimum, the authors should include qualitative examples or a bias-specific evaluation (e.g., classification of the rewritten text as biased or unbiased by a separate model) to validate that the rewrites remove bias.
  5. [§4.4] The paper explicitly states, 'we do not submit our best-performing configuration, indicating that our model still has room for further improvement.' This is an important limitation: the reported leaderboard score is not the system's best result, which further weakens the conclusion that the approach is effective as described. The authors should report their dev-set results for the configurations they actually compare, including the best configuration, rather than relying on the submitted run.
minor comments (5)
  1. [Abstract and §5] The phrase 'demonstrating the effectiveness of the proposed approach' appears in both the abstract and conclusion, but the evidence presented does not support such a strong claim. Please rephrase to describe the results as 'competitive' or 'among the top-performing systems' until controlled comparisons are provided.
  2. [§3.4, References [22,33,34,35]] References [33] and [34] are the authors' own previous work, as indicated by the author list. If these are self-citations, this should be disclosed in the text; otherwise the citation practice is unclear. Also, references [36] and [37] appear to be the same CORGI-PM paper duplicated; merge them.
  3. [§4.2] The use of 'multi-temperature sampling' is misleading: the experiments in Table 4 evaluate three individual temperature values, not a sampling ensemble. If the intended method is to sample multiple outputs at each temperature and then select or aggregate, that procedure should be described; otherwise use 'temperature sweep' instead of 'sampling'.
  4. [Figure 1 and Table 1] Figure 1 labels the data subsets as 'Data A, B, C, N' but the caption does not explain that N corresponds to the six experts; the figure would benefit from a clearer mapping between the subsets and the expert models. Also, 'T able 1' and 'T able 2' in the text contain stray spaces.
  5. [§4.4] The discussion of hyperparameter tuning says 'our optimization in LoRA design and learning rate scheduling confirms the practical value of our efficient parameter search strategy,' but no learning-rate or LoRA-rank ablation is reported. Either provide the supporting results or soften this sentence.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the shared-task result rests on an external leaderboard evaluation, not on a self-referential derivation.

full rationale

This paper is a shared-task system description, not a derivation whose conclusions are encoded in its premises. Its central claim is an external leaderboard score (47.90%, rank 4) obtained by fine-tuning Qwen2.5-7B-Instruct with LoRA, rebalanced data subsets, majority voting, and multi-temperature sampling. The training data (CORGI-PM) is distinct from the held-out test set, and the reported F1 and BLEU/METEOR/ROUGE-L scores come from the shared-task evaluation rather than from any equation in the paper that takes its own outputs as inputs. No fitted parameter is renamed as a prediction, and no claimed prediction is equivalent by construction to a fitted value. The self-citations [33,34] in Section 3.4 are used only as contextual examples that voting is a known ensemble strategy; they are accompanied by external citations [22,35,1,5] and do not carry the system's empirical result. The paper's own admissions in Section 4.4 — the competition allowed only three final submissions, only three temperatures were explored, the best configuration was not submitted, and no voting ablation is reported — weaken the strength of the 'effectiveness' claim, as does the comparison with the Prompt baseline in Table 2, where the proposed system's average score (0.479) is below Prompt (0.502). These are evidence-quality and validity concerns, not circularity: per the hard rules, a missing ablation or a baseline that is not beaten is not a derivation loop. No circular step can be exhibited with a specific reduction from any equation or fitted quantity. The honest finding is therefore no significant circularity.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The system rests on standard NLP engineering choices: LoRA hyperparameters, data re-sampling ratios, ensemble size, voting threshold, and sampling temperatures. None of these are derived or validated by controlled experiments. The evaluation additionally assumes the CORGI-PM annotations and the official leaderboard metrics are valid, including the use of BLEU/METEOR/ROUGE-L as a proxy for successful bias mitigation. No new entities are introduced.

free parameters (7)
  • learning rate = 3e-4
    Used for all LoRA fine-tuning; no search or sensitivity analysis is reported (Section 4.2).
  • training epochs = 4
    Fixed for all subtasks; the authors state they reused the same configuration without task-specific tuning (Section 4.4).
  • inference temperature = {0.01, 0.1, 0.3}
    Chosen by hand for subtask 3; Table 4 shows changes of about 0.001 in the subtask-3 score across these values.
  • voting threshold = more than 3 of 6 experts
    Majority-voting rule defined in Section 3.4; no sensitivity analysis is provided.
  • number of expert models = 6
    Ensemble size in Section 3.4; no comparison with other ensemble sizes is reported.
  • data recombination ratio = approximately 1:1 biased to non-biased
    Described in Section 3.3; the exact ratio varies by random split and is not reported.
  • LoRA rank and alpha = not reported
    Central to the LoRA adaptation and needed for reproduction, but omitted from Section 4.2.
assumptions (4)
  • domain assumption CORGI-PM annotations are high-quality and correctly label gender bias in Chinese sentences.
    The paper trains and evaluates on CORGI-PM via the shared task (Section 4.1) without independently validating label quality.
  • domain assumption The official shared-task test labels and evaluation metrics are a valid measure of detection, classification, and mitigation quality.
    All headline scores come from the NLPCC-2025 leaderboard (Tables 2-6); if the test set or metrics are flawed, the central results are not interpretable.
  • domain assumption Training multiple experts on randomly recombined balanced subsets and taking a majority vote improves classification robustness.
    Asserted in Section 3.4 as the motivation for the ensemble, but no ablation compares it with a single expert.
  • domain assumption Multi-temperature sampling produces better or more diverse de-biased rewrites.
    Sections 3.4 and 4.4 claim improved diversity, yet Table 4 shows almost no score difference across temperatures.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Detection to Mitigation: Addressing Gender Bias in Chinese Texts via Efficient Tuning and Voting-Based Rebalancing." pith.science (2026). https://pith.science/paper/COGHDEWV

@misc{pith2026250907889,
  author       = {Pith},
  title        = {Pith review of: From Detection to Mitigation: Addressing Gender Bias in Chinese Texts via Efficient Tuning and Voting-Based Rebalancing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/COGHDEWV}},
  note         = {Machine review of arXiv:2509.07889}
}
read the original abstract

This paper presents our team's solution to Shared Task 7 of NLPCC-2025, which focuses on sentence-level gender bias detection and mitigation in Chinese. The task aims to promote fairness and controllability in natural language generation by automatically detecting, classifying, and mitigating gender bias. To address this challenge, we adopt a fine-tuning approach based on large language models (LLMs), efficiently adapt to the bias detection task via Low-Rank Adaptation (LoRA). In terms of data processing, we construct a more balanced training set to alleviate class imbalance and introduce heterogeneous samples from multiple sources to enhance model generalization. For the detection and classification sub-tasks, we employ a majority voting strategy that integrates outputs from multiple expert models to boost performance. Additionally, to improve bias generation detection and mitigation, we design a multi-temperature sampling mechanism to capture potential variations in bias expression styles. Experimental results demonstrate the effectiveness of our approach in bias detection, classification, and mitigation. Our method ultimately achieves an average score of 47.90%, ranking fourth in the shared task.

Figures

Figures reproduced from arXiv: 2509.07889 by the authors.

Figure 1
Figure 1. The overall framework of the proposed method. From left to right, it illustrates the specific implementations of subtasks 1, 2, and 3. beliefs. The output is a multi-hot vector indicating the presence of each bias type (e.g.,[1, 0, 0] indicates the presence of the AC bias type.). Bias Mitigation. Given a biased sentence, the goal is to generate a revised version that mitigates or removes the gender bias while preser… view at source ↗
Figure 2
Figure 2. The prompts used in the LLMs. unbiased instances is randomly split into five subsets. Each subset is then ran￾domly combined with the biased data at approximately a 1:1 ratio to construct diverse training sets. In addition to preserving the completeness of the data, we also create a combined dataset that includes all biased and unbiased instances without any splitting. For subtask 3, we generate texts with different… view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 18 canonical work pages

  1. [1]

    Heliyon10(19) (2024)

    Agbesi, V.K., Chen, W., Yussif, S.B., Ukwuoma, C.C., Gu, Y.H., Al-Antari, M.A.: Mutcelm: An optimal multi-textcnn-based ensemble learning for text classification. Heliyon10(19) (2024)

    work page 2024

  2. [2]

    In: Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summa- rization

    Banerjee, S., Lavie, A.: Meteor: An automatic metric for mt evaluation with im- proved correlation with human judgments. In: Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summa- rization. pp. 65–72 (2005)

    2005

  3. [3]

    arXiv preprint arXiv:2005.14050 (2020)

    Blodgett, S.L., Barocas, S., Daumé III, H., Wallach, H.: Language (technology) is power: A critical survey of" bias" in nlp. arXiv preprint arXiv:2005.14050 (2020)

    arXiv 2020

  4. [4]

    Advances in neural information processing systems29(2016)

    Bolukbasi, T., Chang, K.W., Zou, J.Y., Saligrama, V., Kalai, A.T.: Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems29(2016)

    work page 2016

  5. [5]

    Journal of Computer Technology and Software3(6) (2024)

    Cang, Y., Yang, W., Sun, D., Ye, Z., Zheng, Z.: Albert-driven ensemble learning for medical text classification. Journal of Computer Technology and Software3(6) (2024)

    work page 2024

  6. [6]

    Catena 151, 147–160 (2017) Title Suppressed Due to Excessive Length 11

    Chen, W., Xie, X., Wang, J., Pradhan, B., Hong, H., Bui, D.T., Duan, Z., Ma, J.: A comparative study of logistic model tree, random forest, and classification and regression tree models for spatial prediction of landslide susceptibility. Catena 151, 147–160 (2017) Title Suppressed Due to Excessive Length 11

    work page 2017

  7. [7]

    Master’s thesis, University of Waterloo (2015)

    Chen, Y.: Convolutional neural network for sentence classification. Master’s thesis, University of Waterloo (2015)

    work page 2015

  8. [8]

    arXiv preprint arXiv:2407.10759 (2024)

    Chu, Y., Xu, J., Yang, Q., Wei, H., Wei, X., Guo, Z., Leng, Y., Lv, Y., He, J., Lin, J., et al.: Qwen2-audio technical report. arXiv preprint arXiv:2407.10759 (2024)

    arXiv 2024

Show all 41 references
  1. [9]

    Nature Machine Intelligence1(11), 495–496 (2019)

    Costa-Jussà, M.R.: An analysis of gender bias studies in natural language process- ing. Nature Machine Intelligence1(11), 495–496 (2019)

    2019

  2. [10]

    In: Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers)

    Devlin, J., Chang, M.W., Lee, K., Toutanova, K.: Bert: Pre-training of deep bidi- rectional transformers for language understanding. In: Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologi...

    2019

  3. [11]

    Proceedings of the National Academy of Sciences115(16), E3635–E3644 (2018)

    Garg, N., Schiebinger, L., Jurafsky, D., Zou, J.: Word embeddings quantify 100 years of gender and ethnic stereotypes. Proceedings of the National Academy of Sciences115(16), E3635–E3644 (2018)

    2018

  4. [12]

    arXiv preprint arXiv:1901.11333 (2019)

    Jin, Z., Jin, D., Mueller, J., Matthews, N., Santus, E.: Imat: Unsupervised text attribute transfer via iterative matching and translation. arXiv preprint arXiv:1901.11333 (2019)

    2019 arXiv

  5. [13]

    arXiv preprint arXiv:1804.06437 (2018)

    Li, J., Jia, R., He, H., Liang, P.: Delete, retrieve, generate: a simple approach to sentiment and style transfer. arXiv preprint arXiv:1804.06437 (2018)

    2018 arXiv

  6. [14]

    arXiv preprint arXiv:2101.00190 (2021)

    Li, X.L., Liang, P.: Prefix-tuning: Optimizing continuous prompts for generation. arXiv preprint arXiv:2101.00190 (2021)

    2021 arXiv

  7. [15]

    In: Text sum- marization branches out

    Lin, C.Y.: Rouge: A package for automatic evaluation of summaries. In: Text sum- marization branches out. pp. 74–81 (2004)

    2004

  8. [16]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Liu, Q., Qin, J., Ye, W., Mou, H., He, Y., Wang, K.: Adaptive prompt routing for arbitrary text style transfer with pre-trained language models. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 38, pp. 18689–18697 (2024)

    2024

  9. [17]

    arXiv preprint arXiv:2110.07602 (2021)

    Liu, X., Ji, K., Fu, Y., Tam, W.L., Du, Z., Yang, Z., Tang, J.: P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks. arXiv preprint arXiv:2110.07602 (2021)

    2021 arXiv

  10. [18]

    Logic, language, and security: essays dedicated to Andre Scedrov on the occasion of his 65th birthday pp

    Lu, K., Mardziel, P., Wu, F., Amancharla, P., Datta, A.: Gender bias in neural natural language processing. Logic, language, and security: essays dedicated to Andre Scedrov on the occasion of his 65th birthday pp. 189–202 (2020)

    2020

  11. [19]

    Neural Computation 9(8), 1735 (1997)

    Memory, L.S.T.: Sepp hochreiter and jürgen schmidhuber. Neural Computation 9(8), 1735 (1997)

    1997

  12. [20]

    arXiv preprint arXiv:1301.3781 (2013)

    Mikolov, T., Chen, K., Corrado, G., Dean, J.: Efficient estimation of word repre- sentations in vector space. arXiv preprint arXiv:1301.3781 (2013)

    2013 arXiv

  13. [21]

    Proceedings of the national academy of sciences109(41), 16474–16479 (2012)

    Moss-Racusin, C.A., Dovidio, J.F., Brescoll, V.L., Graham, M.J., Handelsman, J.: Science faculty’s subtle gender biases favor male students. Proceedings of the national academy of sciences109(41), 16474–16479 (2012)

    2012

  14. [22]

    Expert Systems with Applications62, 1–16 (2016)

    Onan, A., Korukoğlu, S., Bulut, H.: A multiobjective weighted voting ensemble classifier based on differential evolution algorithm for text sentiment classification. Expert Systems with Applications62, 1–16 (2016)

    2016

  15. [23]

    In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics

    Papineni, K., Roukos, S., Ward, T., Zhu, W.J.: Bleu: a method for automatic evaluation of machine translation. In: Proceedings of the 40th annual meeting of the Association for Computational Linguistics. pp. 311–318 (2002)

    2002

  16. [24]

    In: Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP)

    Pennington, J., Socher, R., Manning, C.D.: Glove: Global vectors for word repre- sentation. In: Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP). pp. 1532–1543 (2014)

    2014

  17. [25]

    Journal of machine learning research21(140), 1–67 (2020) 12 F

    Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., Liu, P.J.: Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research21(140), 1–67 (2020) 12 F. Author et al

    2020

  18. [26]

    arXiv preprint arXiv:2109.03910 (2021)

    Reif, E., Ippolito, D., Yuan, A., Coenen, A., Callison-Burch, C., Wei, J.: A recipe for arbitrary text style transfer with large language models. arXiv preprint arXiv:2109.03910 (2021)

    2021 arXiv

  19. [27]

    Schütze, H., Manning, C.D., Raghavan, P.: Introduction to information retrieval, vol. 39. Cambridge University Press Cambridge (2008)

    2008

  20. [28]

    Advances in neural information processing systems27(2014)

    Sutskever, I., Vinyals, O., Le, Q.V.: Sequence to sequence learning with neural networks. Advances in neural information processing systems27(2014)

    2014

  21. [29]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    2017

  22. [30]

    arXiv preprint arXiv:2310.00035 (2023)

    Wang, X., Aitchison, L., Rudolph, M.: Lora ensembles for large language model fine-tuning. arXiv preprint arXiv:2310.00035 (2023)

    2023 arXiv

  23. [31]

    Transactions of the Association for Com- putational Linguistics6, 605–617 (2018)

    Webster, K., Recasens, M., Axelrod, V., Baldridge, J.: Mind the gap: A balanced corpus of gendered ambiguous pronouns. Transactions of the Association for Com- putational Linguistics6, 605–617 (2018)

    2018

  24. [32]

    Advances in neural information processing systems35, 24824–24837 (2022)

    Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E., Le, Q.V., Zhou, D., et al.: Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems35, 24824–24837 (2022)

    2022

  25. [33]

    In: China Health Information Processing Conference

    Wu, C., Fang, W., Dai, F., Yin, H.: A model ensemble approach with llm for chinese text classification. In: China Health Information Processing Conference. pp. 214–230. Springer (2023)

    2023

  26. [34]

    In: China Health Information Processing Conference

    Wu, C., Lin, Z., Fang, W., Huang, Y.: A medical diagnostic assistant based on llm. In: China Health Information Processing Conference. pp. 135–147. Springer (2023)

    2023

  27. [35]

    arXiv preprint arXiv:2406.15477 (2024)

    Yin, K., Liu, C., Mostafavi, A., Hu, X.: Crisissense-llm: Instruction fine-tuned large language model for multi-label social media text classification in disaster informatics. arXiv preprint arXiv:2406.15477 (2024)

    2024 arXiv

  28. [37]

    Zhang, G., Li, Y., Wu, Y., Zhang, L., Lin, C., Geng, J., Wang, S., Fu, J.: Corgi-pm: A chinese corpus for gender bias probing and mitigation (2023), https://arxiv.org/abs/2301.00395

    2023 arXiv

  29. [38]

    arXiv preprint arXiv:1808.07894 (2018)

    Zhang, Z., Ren, S., Liu, S., Wang, J., Chen, P., Li, M., Zhou, M., Chen, E.: Style transfer as unsupervised machine translation. arXiv preprint arXiv:1808.07894 (2018)

    2018 arXiv

  30. [39]

    arXiv preprint arXiv:1707.09457 (2017)

    Zhao, J., Wang, T., Yatskar, M., Ordonez, V., Chang, K.W.: Men also like shop- ping: Reducing gender bias amplification using corpus-level constraints. arXiv preprint arXiv:1707.09457 (2017)

    2017 arXiv

  31. [40]

    arXiv preprint arXiv:1804.06876 (2018)

    Zhao, J., Wang, T., Yatskar, M., Ordonez, V., Chang, K.W.: Gender bias in coreference resolution: Evaluation and debiasing methods. arXiv preprint arXiv:1804.06876 (2018)

    2018 arXiv

  32. [41]

    arXiv preprint arXiv:1511.08630 (2015)

    Zhou, C., Sun, C., Liu, Z., Lau, F.: A c-lstm neural network for text classification. arXiv preprint arXiv:1511.08630 (2015)

    2015 arXiv

  33. [42]

    In: Proceedings of the 1st ACM Workshop on Large AI Systems and Models with Privacy and Safety Analysis

    Zhu, K., Wang, J., Zhou, J., Wang, Z., Chen, H., Wang, Y., Yang, L., Ye, W., Zhang, Y., Gong, N., et al.: Promptrobust: Towards evaluating the robustness of large language models on adversarial prompts. In: Proceedings of the 1st ACM Workshop on Large AI Systems and Models wit...

    2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.