<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.3 20210610//EN" "JATS-journalpublishing1-3.dtd">
<article article-type="research-article" dtd-version="1.3" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xml:lang="ru"><front><journal-meta><journal-id journal-id-type="publisher-id">ntv</journal-id><journal-title-group><journal-title xml:lang="ru">Научно-технический вестник информационных технологий, механики и оптики</journal-title><trans-title-group xml:lang="en"><trans-title>Scientific and Technical Journal of Information Technologies, Mechanics and Optics</trans-title></trans-title-group></journal-title-group><issn pub-type="ppub">2226-1494</issn><issn pub-type="epub">2500-0373</issn><publisher><publisher-name>Университет ИТМО</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.17586/2226-1494-2024-24-6-1024-1034</article-id><article-id custom-type="elpub" pub-id-type="custom">ntv-409</article-id><article-categories><subj-group subj-group-type="heading"><subject>Research Article</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="ru"><subject>КОМПЬЮТЕРНЫЕ СИСТЕМЫ И ИНФОРМАЦИОННЫЕ ТЕХНОЛОГИИ</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="en"><subject>COMPUTER SCIENCE</subject></subj-group></article-categories><title-group><article-title>Улучшение вопросно-ответных систем в области программирования с дообучением языковых моделей на структурированных разнородных данных онлайн-форумов</article-title><trans-title-group xml:lang="en"><trans-title>Improving question answering in programming domain with pretrained language model finetuning using structured diverse online forum data</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0000-0003-3705-0047</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Горбатовский</surname><given-names>А. В.</given-names></name><name name-style="western" xml:lang="en"><surname>Gorbatovski</surname><given-names>A. V.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Горбатовский Алексей Валерьевич - студент,</p><p>Санкт-Петербург, 197101</p></bio><bio xml:lang="en"><p>Alexey V. Gorbatovski - Student,</p><p>Saint Petersburg, 197101</p></bio><email xlink:type="simple">gorbatoski@itmo.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0009-0007-8396-2801</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Разин</surname><given-names>А. Д.</given-names></name><name name-style="western" xml:lang="en"><surname>Razin</surname><given-names>A. D.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Разин Александр Дмитриевич - студент,</p><p>Санкт-Петербург, 197101</p></bio><bio xml:lang="en"><p>Alexandr D. Razin - Student,</p><p>Saint Petersburg, 197101</p></bio><email xlink:type="simple">adrazin@itmo.ru</email><xref ref-type="aff" rid="aff-1"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0000-0001-7984-4721</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Алиев</surname><given-names>А. А.</given-names></name><name name-style="western" xml:lang="en"><surname>Aliev</surname><given-names>A. A.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Алиев Артем Александрович - старший преподаватель,</p><p>Санкт-Петербург, 199034</p></bio><bio xml:lang="en"><p>Artem A. Aliev - Senior Lecturer, </p><p>Saint Petersburg, 199034</p></bio><email xlink:type="simple">artem.aliev@gmail.com</email><xref ref-type="aff" rid="aff-2"/></contrib><contrib contrib-type="author" corresp="yes"><contrib-id contrib-id-type="orcid">https://orcid.org/0000-0001-8828-4615</contrib-id><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Ковальчук</surname><given-names>С. В.</given-names></name><name name-style="western" xml:lang="en"><surname>Kovalchuk</surname><given-names>S. V.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Ковальчук Сергей Валерьевич - кандидат технических наук, доцент,</p><p>Санкт-Петербург, 197101</p></bio><bio xml:lang="en"><p>Sergey V. Kovalchuk - PhD, Associate Professor,</p><p>Saint Petersburg, 197101</p></bio><email xlink:type="simple">kovalchuk@itmo.ru</email><xref ref-type="aff" rid="aff-1"/></contrib></contrib-group><aff-alternatives id="aff-1"><aff xml:lang="ru"><institution>Университет ИТМО</institution><country>Россия</country></aff><aff xml:lang="en"><institution>ITMO University</institution><country>Russian Federation</country></aff></aff-alternatives><aff-alternatives id="aff-2"><aff xml:lang="ru"><institution>Санкт-Петербургский государственный университет</institution><country>Россия</country></aff><aff xml:lang="en"><institution>St. Petersburg State University (SPbSU)</institution><country>Russian Federation</country></aff></aff-alternatives><pub-date pub-type="collection"><year>2024</year></pub-date><pub-date pub-type="epub"><day>29</day><month>12</month><year>2024</year></pub-date><volume>24</volume><issue>6</issue><fpage>1024</fpage><lpage>1034</lpage><permissions><copyright-statement>Copyright &amp;#x00A9; Горбатовский А.В., Разин А.Д., Алиев А.А., Ковальчук С.В., 2024</copyright-statement><copyright-year>2024</copyright-year><copyright-holder xml:lang="ru">Горбатовский А.В., Разин А.Д., Алиев А.А., Ковальчук С.В.</copyright-holder><copyright-holder xml:lang="en">Gorbatovski A.V., Razin A.D., Aliev A.A., Kovalchuk S.V.</copyright-holder><license xml:lang="ru" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>Данная работа распространяется под лицензией Creative Commons Attribution 4.0.</license-p></license><license xml:lang="en" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>This work is licensed under a Creative Commons Attribution 4.0 License.</license-p></license></permissions><self-uri xlink:href="https://ntv.elpub.ru/jour/article/view/409">https://ntv.elpub.ru/jour/article/view/409</self-uri><abstract><sec><title>Введение</title><p>Введение. Тематические вопросно-ответные онлайн-форумы предметных сообществ, такие как Stack Overflow, сегодня становятся незаменим инструментом разработчиков программного обеспечения. Форумы обеспечивают быстрый и эффективный поиск решений и оперативный отклик сообщества. Современные большие языковые модели, обучаемые, в том числе, на данных таких форумов, обладают потенциалом для автоматизации ответов на тематические вопросы. Но часто языковые модели демонстрируют существенную ограниченность в сложных областях, таких как программирование из-за разнородности области и контекстов задаваемых вопросов.</p></sec><sec><title>Метод</title><p>Метод. В работе представлен подход к решению проблемы разнородных данных на основе структурирования данных сложной предметной области. На первом этапе предлагается декомпозиция доступных данных форумов с выделением тематических подмножеств. Далее, для отдельных тематик происходит дообучение моделей, применяя обучение с подкреплением с человеческой обратной связью (Reinforcement Learning with Human Feedback, RLHF) с использованием пользовательских оценок доступных в данных. Для управления ансамблем дообученных моделей используется классификация вопросов с последующим выбором соответствующей модели.</p></sec><sec><title>Основные результаты</title><p>Основные результаты. Экспериментальные исследования были проведены на подмножестве вопросов, связанных с Python, из Stack Overflow, с использованием модели Llama 7B в качестве базовой языковой модели. Результаты исследований показали, что путем классификации вопросов возможно повысить производительность модели до +22,5 % по метрике Rouge. Кроме того, включение RLHF привело к дополнительному улучшению до +11,2 %. Для валидации этих результатов выполнена экспертная оценка сгенерированных ответов, которая подтвердила эффективность представленного подхода.</p></sec><sec><title>Обсуждение</title><p>Обсуждение. Исследование показывает, что путем структурирования данных онлайн-форумов и обработки неявной обратной связи возможно значительно улучшить производительность больших языковых моделей в таких сложных областях, характеризующихся высокой неоднородностью, как разработка программного обеспечения.</p></sec></abstract><trans-abstract xml:lang="en"><p>Today, Community Question Answering (CQA) forums such as Stack Overflow are becoming an irreplaceable tool for software developers, providing fast and efficient solution search and prompt community response. Although modern Pretrained Language Models (PLMs), also trained including on data from such forums, have the potential to automate answering of domain-specific questions, they often show significant limitations in complex domains such as programming due to the heterogeneity of the domain and variety in contexts of the questions being asked. In our study, we propose an approach to solving this problem based on structuring data in a complex domain. The first stage includes decomposing available forum data with the selection of thematic subsets. Next, for individual topics, models are finetuned using Reinforcement Learning with Human Feedback (RLHF) using the voting available in the forum data. Finally, to manage the ensemble of finetuned models, question classification is used with subsequent selection of the appropriate model. Experimental studies were conducted on a subset of Python-related questions from Stack Overflow, using the Llama 7B model as the base language model. Experimental studies were conducted on a subset of Python-related questions from Stack Overflow forum using the Llama 7B model as a base PLM. The results of the studies showed that by classifying questions we can improve the model performance up to +22.5 % on the Rouge metric. Moreover, the inclusion of RLHF resulted in an additional improvement of up to +11.2 %. To validate these results, we performed human evaluation of the generated responses, which confirmed the effectiveness of our approach. This study shows that by structuring community data and processing implicit feedback, we can significantly improve PLM performance in CQA tasks in complex domains characterized by high heterogeneity, such as software development.</p></trans-abstract><kwd-group xml:lang="ru"><kwd>вопросно-ответные системы</kwd><kwd>обработка естественного языка</kwd><kwd>генерация естественного языка</kwd><kwd>предобученные языковые модели</kwd><kwd>большие языковые модели</kwd><kwd>дообучение</kwd><kwd>разработка программного обеспечения</kwd></kwd-group><kwd-group xml:lang="en"><kwd>question answering</kwd><kwd>natural language processing</kwd><kwd>natural language generation</kwd><kwd>pretrained language models</kwd><kwd>large language models</kwd><kwd>finetuning</kwd><kwd>software development</kwd></kwd-group><funding-group><funding-statement xml:lang="ru">Исследование выполнено за счет гранта Российского научного фонда № 24-11-00272, https://rscf.ru/ project/24-11-00272/.</funding-statement></funding-group></article-meta></front><back><ref-list><title>References</title><ref id="cit1"><label>1</label><citation-alternatives><mixed-citation xml:lang="ru">Koubaa A., Boulila W., Ghouti L., Alzahem A., Latif S. Exploring ChatGPT capabilities and limitations: A survey // IEEE Access. 2023. V. 11. P. 118698–118721. https://doi.org/10.1109/access.2023.3326474</mixed-citation><mixed-citation xml:lang="en">Koubaa A., Boulila W., Ghouti L., Alzahem A., Latif S. Exploring ChatGPT capabilities and limitations: A survey. IEEE Access, 2023, vol. 11, pp. 118698–118721. https://doi.org/10.1109/access.2023.3326474</mixed-citation></citation-alternatives></ref><ref id="cit2"><label>2</label><citation-alternatives><mixed-citation xml:lang="ru">Dave T., Athaluri S.A., Singh S. ChatGPT in medicine: an overview of its applications, advantages, limitations, future prospects, and ethical considerations // Frontiers in Artificial Intelligence. 2023. V. 6. https://doi.org/10.3389/frai.2023.1169595</mixed-citation><mixed-citation xml:lang="en">Dave T., Athaluri S.A., Singh S. ChatGPT in medicine: an overview of its applications, advantages, limitations, future prospects, and ethical considerations. Frontiers in Artificial Intelligence, 2023, vol. 6. https://doi.org/10.3389/frai.2023.1169595</mixed-citation></citation-alternatives></ref><ref id="cit3"><label>3</label><citation-alternatives><mixed-citation xml:lang="ru">Hadi M.U., Tashi Q.A., Qureshi R., Shah A., Muneer A., Irfan M., Zafar A., Shaikh M.B., Akhtar N., Hassan S.Z., Shoman M., Wu J., Mirjalili S., Shah M. Large language models: a comprehensive survey of its applications, challenges, limitations, and future prospects // Authorea Preprints. 2023.</mixed-citation><mixed-citation xml:lang="en">Hadi M.U., Tashi Q.A., Qureshi R., Shah A., Muneer A., Irfan M., Zafar A., Shaikh M.B., Akhtar N., Hassan S.Z., Shoman M., Wu J., Mirjalili S., Shah M. Large language models: a comprehensive survey of its applications, challenges, limitations, and future prospects. Authorea Preprints, 2023.</mixed-citation></citation-alternatives></ref><ref id="cit4"><label>4</label><citation-alternatives><mixed-citation xml:lang="ru">Bird C., Ford D., Zimmermann T., Forsgren N., Kalliamvakou E., Lowdermilk T., Gazit I. Taking Flight with Copilot: Early insights and opportunities of AI-powered pair-programming tools // Queue. 2022. V. 20. N 6. P. 35–57. https://doi.org/10.1145/3582083</mixed-citation><mixed-citation xml:lang="en">Bird C., Ford D., Zimmermann T., Forsgren N., Kalliamvakou E., Lowdermilk T., Gazit I. Taking Flight with Copilot: Early insights and opportunities of AI-powered pair-programming tools. Queue, 2022, vol. 20, no. 6, pp. 35–57. https://doi.org/10.1145/3582083</mixed-citation></citation-alternatives></ref><ref id="cit5"><label>5</label><citation-alternatives><mixed-citation xml:lang="ru">Ernst N.A., Bavota G. AI-Driven Development Is Here: Should You Worry? // IEEE Software. 2022. V. 39. N 2. P. 106–110. https://doi.org/10.1109/ms.2021.3133805</mixed-citation><mixed-citation xml:lang="en">Ernst N.A., Bavota G. AI-Driven Development Is Here: Should You Worry? IEEE Software, 2022, vol. 39, no. 2, pp. 106–110. https://doi.org/10.1109/ms.2021.3133805</mixed-citation></citation-alternatives></ref><ref id="cit6"><label>6</label><citation-alternatives><mixed-citation xml:lang="ru">Liang J.T., Yang C., Myers B.A. A large-scale survey on the usability of AI programming assistants: successes and challenges // Proc. Of the 46th IEEE/ACM International Conference on Software Engineering (ICSE ‘24). 2024. P. 605–617. https://doi.org/10.1145/3597503.3608128</mixed-citation><mixed-citation xml:lang="en">Liang J.T., Yang C., Myers B.A. A large-scale survey on the usability of AI programming assistants: successes and challenges. Proc. of the 46th IEEE/ACM International Conference on Software Engineering (ICSE ‘24) , 2024, pp. 605–617. https://doi.org/10.1145/3597503.3608128</mixed-citation></citation-alternatives></ref><ref id="cit7"><label>7</label><citation-alternatives><mixed-citation xml:lang="ru">Roy P.K., Saumya S., Singh J.P., Banerjee S., Gutub A. Analysis of community question-answering issues via machine learning and deep learning: State-of-the-art review // CAAI Transactions on Intelligence Technology. 2023. V. 8. N 1. P. 95–117. https://doi.org/10.1049/cit2.12081</mixed-citation><mixed-citation xml:lang="en">Roy P.K., Saumya S., Singh J.P., Banerjee S., Gutub A. Analysis of community question-answering issues via machine learning and deep learning: State-of-the-art review. CAAI Transactions on Intelligence Technology, 2023, vol. 8, no. 1, pp. 95–117. https://doi.org/10.1049/cit2.12081</mixed-citation></citation-alternatives></ref><ref id="cit8"><label>8</label><citation-alternatives><mixed-citation xml:lang="ru">Blanco G., Pérez-López R., Fdez-Riverola F., Lourenço A.M.G. Understanding the social evolution of the Java community in Stack Overflow: A 10-year study of developer interactions // Future Generation Computer Systems. 2020. V. 105. P. 446–454. https://doi.org/10.1016/j.future.2019.12.021</mixed-citation><mixed-citation xml:lang="en">Blanco G., Pérez-López R., Fdez-Riverola F., Lourenço A.M.G. Understanding the social evolution of the Java community in Stack Overflow: A 10-year study of developer interactions. Future Generation Computer Systems, 2020, vol. 105, pp. 446–454. https://doi.org/10.1016/j.future.2019.12.021</mixed-citation></citation-alternatives></ref><ref id="cit9"><label>9</label><citation-alternatives><mixed-citation xml:lang="ru">Beyer S., Macho C., Di Penta M., Pinzger M. What kind of questions do developers ask on Stack Overflow? A comparison of automated approaches to classify posts into question categories // Empirical Software Engineering. 2020. V. 25. N 3. P. 2258–2301. https://doi.org/10.1007/s10664-019-09758-x</mixed-citation><mixed-citation xml:lang="en">Beyer S., Macho C., Di Penta M., Pinzger M. What kind of questions do developers ask on Stack Overflow? A comparison of automated approaches to classify posts into question categories. Empirical Software Engineering, 2020, vol. 25, no. 3, pp. 2258–2301. https://doi.org/10.1007/s10664-019-09758-x</mixed-citation></citation-alternatives></ref><ref id="cit10"><label>10</label><citation-alternatives><mixed-citation xml:lang="ru">Yuan S., Qin H., Gu X., Shen B. Clean and learn: Improving robustness to spurious solutions in API question answering // International Journal of Software Engineering and Knowledge Engineering. 2022. V. 32. N 7. P. 1101–1123. https://doi.org/10.1142/s0218194022500449</mixed-citation><mixed-citation xml:lang="en">Yuan S., Qin H., Gu X., Shen B. Clean and learn: Improving robustness to spurious solutions in API question answering. International Journal of Software Engineering and Knowledge Engineering, 2022, vol. 32, no. 7, pp. 1101–1123. https://doi.org/10.1142/s0218194022500449</mixed-citation></citation-alternatives></ref><ref id="cit11"><label>11</label><citation-alternatives><mixed-citation xml:lang="ru">Kovalchuk S., Lomshakov V., Aliev A. Human perceiving behavior modeling in evaluation of code generation models // Proc. of the 2nd Workshop on Natural Language Generation, Evaluation, and Metrics (GEM). 2022. P. 287–294. https://doi.org/10.18653/v1/2022.gem-1.24</mixed-citation><mixed-citation xml:lang="en">Kovalchuk S., Lomshakov V., Aliev A. Human perceiving behavior modeling in evaluation of code generation models. Proc. of the 2nd Workshop on Natural Language Generation, Evaluation, and Metrics (GEM), 2022, pp. 287–294. https://doi.org/10.18653/v1/2022.gem1.24</mixed-citation></citation-alternatives></ref><ref id="cit12"><label>12</label><citation-alternatives><mixed-citation xml:lang="ru">Nakano R., Hilton J., Balaji S., Wu J., Ouyang L., Kim C., Hesse C., Jain S., Kosaraju V., Saunders W., Jiang X., Cobbe K., Eloundou T., Krueger G., Button K., Knight M., Chess B., Schulman J. WebGPT: Browser-assisted question-answering with human feedback // arXiv. 2022. arXiv:2112.09332. 2021. https://doi.org/10.48550/arXiv.2112.09332</mixed-citation><mixed-citation xml:lang="en">Nakano R., Hilton J., Balaji S., Wu J., Ouyang L., Kim C., Hesse C., Jain S., Kosaraju V., Saunders W., Jiang X., Cobbe K., Eloundou T., Krueger G., Button K., Knight M., Chess B., Schulman J. WebGPT: Browser-assisted question-answering with human feedback. arXiv, 2022, arXiv:2112.09332. 2021. https://doi.org/10.48550/arXiv.2112.09332</mixed-citation></citation-alternatives></ref><ref id="cit13"><label>13</label><citation-alternatives><mixed-citation xml:lang="ru">Casper S., Davies X., Shi C. Open problems and fundamental limitations of reinforcement learning from human feedback // arXiv. 2023. arXiv:2307.15217. https://doi.org/10.48550/arXiv.2307.15217</mixed-citation><mixed-citation xml:lang="en">Casper S., Davies X., Shi C. Open problems and fundamental limitations of reinforcement learning from human feedback. arXiv, 2023, arXiv:2307.15217. https://doi.org/10.48550/arXiv.2307.15217</mixed-citation></citation-alternatives></ref><ref id="cit14"><label>14</label><citation-alternatives><mixed-citation xml:lang="ru">Gorbatovski A., Kovalchuk S. Bayesian networks for named entity prediction in programming community question answering // Lecture Notes in Computer Science. 2023. V. 14074. P. 282–289. https://doi.org/10.1007/978-3-031-36021-3_28</mixed-citation><mixed-citation xml:lang="en">Gorbatovski A., Kovalchuk S. Bayesian networks for named entity prediction in programming community question answering. Lecture Notes in Computer Science, 2023, vol. 14074, pp. 282–289. https://doi.org/10.1007/978-3-031-36021-3_28</mixed-citation></citation-alternatives></ref><ref id="cit15"><label>15</label><citation-alternatives><mixed-citation xml:lang="ru">Rvanova L., Kovalchuk S. Automatic structuring of topics for natural language generation in community question answering in programming domain // Lecture Notes in Computer Science. 2023. V. 14074. P. 322–329. https://doi.org/10.1007/978-3-031-36021-3_33</mixed-citation><mixed-citation xml:lang="en">Rvanova L., Kovalchuk S. Automatic structuring of topics for natural language generation in community question answering in programming domain. Lecture Notes in Computer Science, 2023, vol. 14074, pp. 322–329. https://doi.org/10.1007/978-3-031-36021-3_33</mixed-citation></citation-alternatives></ref><ref id="cit16"><label>16</label><citation-alternatives><mixed-citation xml:lang="ru">Hu E.J., Shen Y., Wallis P., Allen-Zhu Z., Li Y., Wang S., Wang L., Chen W. LoRA: Low-rank adaptation of large language models // International Conference on Learning Representations (ICLR). 2022.</mixed-citation><mixed-citation xml:lang="en">Hu E.J., Shen Y., Wallis P., Allen-Zhu Z., Li Y., Wang S., Wang L., Chen W. LoRA: Low-rank adaptation of large language models. International Conference on Learning Representations (ICLR), 2022.</mixed-citation></citation-alternatives></ref><ref id="cit17"><label>17</label><citation-alternatives><mixed-citation xml:lang="ru">Stiennon N., Ouyang L., Wu J., Ziegler D.M., Lowe R., Voss C., Radford A., Amodei D., Christiano P. Learning to summarize from human feedback // Advances in Neural Information Processing Systems. 2020. V. 33. P. 3008–3021.</mixed-citation><mixed-citation xml:lang="en">Stiennon N., Ouyang L., Wu J., Ziegler D.M., Lowe R., Voss C., Radford A., Amodei D., Christiano P. Learning to summarize from human feedback. Advances in Neural Information Processing Systems, 2020, vol. 33, pp. 3008–3021.</mixed-citation></citation-alternatives></ref><ref id="cit18"><label>18</label><citation-alternatives><mixed-citation xml:lang="ru">Ouyang L., Wu J., Jiang X., Almeida D., Wainwright C.L., Mishkin P., Zhang C., Agarwal S., Slama K., Ray A., Schulman J., Hilton J., Kelton F., Miller L., Simens M., Askell A., Welinder P., Christiano P., Leike J., Lowe R. Training language models to follow instructions with human feedback // Advances in Neural Information Processing Systems. 2022. V. 35. P. 27730–27744.</mixed-citation><mixed-citation xml:lang="en">Ouyang L., Wu J., Jiang X., Almeida D., Wainwright C.L., Mishkin P., Zhang C., Agarwal S., Slama K., Ray A., Schulman J., Hilton J., Kelton F., Miller L., Simens M., Askell A., Welinder P., Christiano P., Leike J., Lowe R. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 2022, vol. 35, pp. 27730–27744.</mixed-citation></citation-alternatives></ref><ref id="cit19"><label>19</label><citation-alternatives><mixed-citation xml:lang="ru">Schulman J., Wolski F., Dhariwal P., Radford A., Klimov O. Proximal policy optimization algorithms // arXiv. 2017. arXiv:1707.06347. https://doi.org/10.48550/arXiv.1707.06347</mixed-citation><mixed-citation xml:lang="en">Schulman J., Wolski F., Dhariwal P., Radford A., Klimov O. Proximal policy optimization algorithms. arXiv, 2017, arXiv:1707.06347. https://doi.org/10.48550/arXiv.1707.06347</mixed-citation></citation-alternatives></ref><ref id="cit20"><label>20</label><citation-alternatives><mixed-citation xml:lang="ru">Zhou C., Liu P., Xu P., Iyer S., Sun J., Mao Y., Ma X., Efrat A., Yu P., Yu L., Zhang S., Ghosh G., Lewis M., Zettlemoyer L., Levy O. LIMA: Less Is More for Alignment // arXiv. 2023. arXiv:2305.11206. https://doi.org/10.48550/arXiv.2305.11206</mixed-citation><mixed-citation xml:lang="en">Zhou C., Liu P., Xu P., Iyer S., Sun J., Mao Y., Ma X., Efrat A., Yu P., Yu L., Zhang S., Ghosh G., Lewis M., Zettlemoyer L., Levy O. LIMA: Less Is More for Alignment. arXiv, 2023, arXiv:2305.11206. https://doi.org/10.48550/arXiv.2305.11206</mixed-citation></citation-alternatives></ref><ref id="cit21"><label>21</label><citation-alternatives><mixed-citation xml:lang="ru">Achiam J., Adler S., Agarwal S. et. al. GPT-4 Technical Report // arXiv. 2023. arXiv:2303.08774. https://doi.org/10.48550/arXiv.2303.08774</mixed-citation><mixed-citation xml:lang="en">Achiam J., Adler S., Agarwal S. et. al. GPT-4 Technical Report. arXiv, 2023, arXiv:2303.08774. https://doi.org/10.48550/arXiv.2303.08774</mixed-citation></citation-alternatives></ref><ref id="cit22"><label>22</label><citation-alternatives><mixed-citation xml:lang="ru">Hunter D.R. MM algorithms for generalized Bradley-Terry models // Annals of Statistics. 2003. V. 32. N 1. P. 384–406. https://doi.org/10.1214/aos/1079120141</mixed-citation><mixed-citation xml:lang="en">Hunter D.R. MM algorithms for generalized Bradley-Terry models. Annals of Statistics, 2003, vol. 32, no. 1, pp. 384–406. https://doi.org/10.1214/aos/1079120141</mixed-citation></citation-alternatives></ref><ref id="cit23"><label>23</label><citation-alternatives><mixed-citation xml:lang="ru">Sorokin N., Abulkhanov D., Piontkovskaya I., Malykh V. Ask me anything in your native language // Proc. of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022. P. 395–406. https://doi.org/10.18653/v1/2022.naacl-main.30</mixed-citation><mixed-citation xml:lang="en">Sorokin N., Abulkhanov D., Piontkovskaya I., Malykh V. Ask me anything in your native language. Proc. of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2022, pp. 395–406. https://doi.org/10.18653/v1/2022.naacl-main.30</mixed-citation></citation-alternatives></ref><ref id="cit24"><label>24</label><citation-alternatives><mixed-citation xml:lang="ru">Lee H., Phatale S., Mansoor H., Mesnard T., Ferret J., Lu K., Bishop C., Hall E., Carbune V., Rastogi A., Prakash S. RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback // arXiv. 2023. arXiv:2309.00267. https://doi.org/10.48550/arXiv.2309.00267</mixed-citation><mixed-citation xml:lang="en">Lee H., Phatale S., Mansoor H., Mesnard T., Ferret J., Lu K., Bishop C., Hall E., Carbune V., Rastogi A., Prakash S. RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback. arXiv, 2023, arXiv:2309.00267. https://doi.org/10.48550/arXiv.2309.00267</mixed-citation></citation-alternatives></ref><ref id="cit25"><label>25</label><citation-alternatives><mixed-citation xml:lang="ru">Askell A., Bai Y., Chen A. et al. A general language assistant as a laboratory for alignment // arXiv. 2021. arXiv:2112.00861. https://doi.org/10.48550/arXiv.2112.00861</mixed-citation><mixed-citation xml:lang="en">Askell A., Bai Y., Chen A. et al. A general language assistant as a laboratory for alignment. arXiv, 2021, arXiv:2112.00861. https://doi.org/10.48550/arXiv.2112.00861</mixed-citation></citation-alternatives></ref><ref id="cit26"><label>26</label><citation-alternatives><mixed-citation xml:lang="ru">Dong X., Shen J. Triplet loss in siamese network for object tracking // Lecture Notes in Computer Science. 2018. V. 11217. P. 472–488. https://doi.org/10.1007/978-3-030-01261-8_28</mixed-citation><mixed-citation xml:lang="en">Dong X., Shen J. Triplet loss in siamese network for object tracking. Lecture Notes in Computer Science, 2018, vol. 11217, pp. 472–488. https://doi.org/10.1007/978-3-030-01261-8_28</mixed-citation></citation-alternatives></ref><ref id="cit27"><label>27</label><citation-alternatives><mixed-citation xml:lang="ru">Wang L., Yang N., Huang X., Jiao B., Yang L., Jiang D., Majumder R., Wei F. Text embeddings by weakly-supervised contrastive pretraining // arXiv. 2022. arXiv:2212.03533. https://doi.org/10.48550/arXiv.2212.03533</mixed-citation><mixed-citation xml:lang="en">Wang L., Yang N., Huang X., Jiao B., Yang L., Jiang D., Majumder R., Wei F. Text embeddings by weakly-supervised contrastive pretraining. arXiv, 2022, arXiv:2212.03533. https://doi.org/10.48550/arXiv.2212.03533</mixed-citation></citation-alternatives></ref><ref id="cit28"><label>28</label><citation-alternatives><mixed-citation xml:lang="ru">Wang Y., Wang W., Joty S., Hoi S.C.H. CodeT5: Identifier-aware unified pre-trained encoder-decoder models for code understanding and generation // Proc. of the 2021 Conference on Empirical Methods in Natural Language Processing. 2021. P. 8696–8708. https://doi.org/10.18653/v1/2021.emnlp-main.685</mixed-citation><mixed-citation xml:lang="en">Wang Y., Wang W., Joty S., Hoi S.C.H. CodeT5: Identifier-aware unified pre-trained encoder-decoder models for code understanding and generation. Proc. of the 2021 Conference on Empirical Methods in Natural Language Processing, 2021, pp. 8696–8708. https://doi.org/10.18653/v1/2021.emnlp-main.685</mixed-citation></citation-alternatives></ref><ref id="cit29"><label>29</label><citation-alternatives><mixed-citation xml:lang="ru">Song K., Tan X., Qin T., Lu J., Liu T.-Y.et al. MPNet: Masked and permuted pre-training for language understanding // Advances in Neural Information Processing Systems. 2020. V. 33. P. 16857–16867.</mixed-citation><mixed-citation xml:lang="en">Song K., Tan X., Qin T., Lu J., Liu T.-Y.et al. MPNet: Masked and permuted pre-training for language understanding. Advances in Neural Information Processing Systems, 2020, vol. 33, pp. 16857– 16867.</mixed-citation></citation-alternatives></ref><ref id="cit30"><label>30</label><citation-alternatives><mixed-citation xml:lang="ru">Touvron H., Lavril T., Izacard G., Martinet X., Lachaux M.-A., Lacroix T., Rozière B., Goyal N., Hambro E., Azhar F., Rodriguez A., Joulin A., Grave E., Lample G. LLaMA: Open and efficient foundation language models // arXiv. 2023. arXiv:2302.13971. https://doi.org/10.48550/arXiv.2302.13971</mixed-citation><mixed-citation xml:lang="en">Touvron H., Lavril T., Izacard G., Martinet X., Lachaux M.-A., Lacroix T., Rozière B., Goyal N., Hambro E., Azhar F., Rodriguez A., Joulin A., Grave E., Lample G. LLaMA: Open and efficient foundation language models. arXiv, 2023, arXiv:2302.13971. https://doi.org/10.48550/arXiv.2302.13971</mixed-citation></citation-alternatives></ref><ref id="cit31"><label>31</label><citation-alternatives><mixed-citation xml:lang="ru">Lin C.-Y. ROUGE: A Package for automatic evaluation of summaries // Text Summarization Branches Out. 2004. P. 74–81.</mixed-citation><mixed-citation xml:lang="en">Lin C.-Y. ROUGE: A Package for automatic evaluation of summaries. Text Summarization Branches Out, 2004, pp. 74–81.</mixed-citation></citation-alternatives></ref><ref id="cit32"><label>32</label><citation-alternatives><mixed-citation xml:lang="ru">Post M. A Call for Clarity in Reporting BLEU Scores // Proc. of the Third Conference on Machine Translation: Research Papers 2018. P. 186–191. https://doi.org/10.18653/v1/w18-6319</mixed-citation><mixed-citation xml:lang="en">Post M. A Call for Clarity in Reporting BLEU Scores. Proc. of the Third Conference on Machine Translation: Research Papers, 2018, pp. 186–191. https://doi.org/10.18653/v1/w18-6319</mixed-citation></citation-alternatives></ref><ref id="cit33"><label>33</label><citation-alternatives><mixed-citation xml:lang="ru">Zhang T., Kishore V., Wu F., Weinberger K.Q., Artzi Y. BERTScore: Evaluating Text Generation with BERT // International Conference on Learning Representations (ICLR). 2020.</mixed-citation><mixed-citation xml:lang="en">Zhang T., Kishore V., Wu F., Weinberger K.Q., Artzi Y. BERTScore: Evaluating Text Generation with BERT. International Conference on Learning Representations (ICLR), 2020.</mixed-citation></citation-alternatives></ref><ref id="cit34"><label>34</label><citation-alternatives><mixed-citation xml:lang="ru">Popović M. chrF++: words helping character n-grams // Proc. of the Second Conference on Machine Translation. Copenhagen, Denmark: Association for Computational Linguistics, 2017. P. 612–618. https://doi.org/10.18653/v1/w17-4770</mixed-citation><mixed-citation xml:lang="en">Popović M. chrF++: words helping character n-grams. Proc. of the Second Conference on Machine Translation, 2017, pp. 612–618. https://doi.org/10.18653/v1/w17-4770</mixed-citation></citation-alternatives></ref><ref id="cit35"><label>35</label><citation-alternatives><mixed-citation xml:lang="ru">Artstein R., Poesio M. Inter-Coder Agreement for Computational Linguistics // Computational Linguistics. 2008. V. 34. N 4. P. 555– 596. https://doi.org/10.1162/coli.07-034-r2</mixed-citation><mixed-citation xml:lang="en">Artstein R., Poesio M. Inter-Coder Agreement for Computational Linguistics. Computational Linguistics, 2008, vol. 34, no. 4, pp. 555– 596. https://doi.org/10.1162/coli.07-034-r2</mixed-citation></citation-alternatives></ref></ref-list><fn-group><fn fn-type="conflict"><p>The authors declare that there are no conflicts of interest present.</p></fn></fn-group></back></article>
