{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,4]],"date-time":"2026-02-04T07:24:22Z","timestamp":1770189862633,"version":"3.49.0"},"reference-count":70,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"2","license":[{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62477001"],"award-info":[{"award-number":["62477001"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Artif. Intell."],"published-print":{"date-parts":[[2026,2]]},"DOI":"10.1109\/tai.2025.3581507","type":"journal-article","created":{"date-parts":[[2025,6,19]],"date-time":"2025-06-19T13:40:57Z","timestamp":1750340457000},"page":"791-805","source":"Crossref","is-referenced-by-count":0,"title":["Latent Variable Modeling for Controllable and Diverse Generation From Large Language Models"],"prefix":"10.1109","volume":"7","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-3212-8871","authenticated-orcid":false,"given":"Jianfei","family":"Zhang","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering, Beihang University, Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7617-9041","authenticated-orcid":false,"given":"Bei","family":"Li","sequence":"additional","affiliation":[{"name":"Meituan Inc., Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-5649-2751","authenticated-orcid":false,"given":"Zhuofan","family":"Chen","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Beihang University, Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-8324-7153","authenticated-orcid":false,"given":"Chang","family":"Liu","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Beihang University, Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7508-7222","authenticated-orcid":false,"given":"Chen","family":"Li","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Beihang University, Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3454-2468","authenticated-orcid":false,"given":"Chenghua","family":"Lin","sequence":"additional","affiliation":[{"name":"Department of Computer Science, University of Manchester, Manchester, U.K."}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4229-7215","authenticated-orcid":false,"given":"Wenge","family":"Rong","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Beihang University, Beijing, China"}]}],"member":"263","reference":[{"key":"ref1","article-title":"GPT-4 technical report","year":"2023"},{"key":"ref2","article-title":"Gemini: A family of highly capable multimodal models","author":"Anil","year":"2023"},{"key":"ref3","article-title":"The Llama 3 herd of models","author":"Dubey","year":"2024"},{"key":"ref4","first-page":"53728","article-title":"Direct preference optimization: Your language model is secretly a reward model","volume-title":"Proc. 37th Annu. Conf. Neural Inf. Process. Syst.","author":"Rafailov","year":"2023"},{"key":"ref5","article-title":"Is reinforcement learning (not) for natural language processing: Benchmarks, baselines, and building blocks for natural language policy optimization","volume-title":"Proc. 11th Int. Conf. Learn. Represent.","author":"Ramamurthy","year":"2023"},{"key":"ref6","first-page":"70:1","article-title":"Scaling instruction-finetuned language models","volume":"25","author":"Chung","year":"2024","journal-title":"J. Mach. Learn. Res."},{"key":"ref7","first-page":"4299","article-title":"Deep reinforcement learning from human preferences","volume-title":"Proc. 31st Annu. Conf. Neural Inf. Process. Syst.","author":"Christiano","year":"2017"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P16-1094"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.33"},{"key":"ref10","article-title":"Prometheus: Inducing fine-grained evaluation capability in language models","volume-title":"Proc. 12th Int. Conf. Learn. Represent.","author":"Kim","year":"2024"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.248"},{"key":"ref12","article-title":"Auto-encoding variational Bayes","volume-title":"Proc. 2nd Int. Conf. Learn. Represent.","author":"Kingma","year":"2014"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-46493-0_47"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/K16-1002"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P17-1061"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-emnlp.313"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.704"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.9"},{"key":"ref19","first-page":"4813","article-title":"Disentangling preference representation and text generation for efficient individual preference alignment","volume-title":"Proc. 31st Int. Conf. Comput. Linguistics","author":"Zhang","year":"2025"},{"key":"ref20","article-title":"Controlling large language model with latent actions","author":"Jia","year":"2025"},{"key":"ref21","first-page":"52516","article-title":"Personalizing reinforcement learning from human feedback with variational preference learning","volume-title":"Proc. 38th Annu. Conf. Adv. Neural Inf. Process. Syst","author":"Poddar","year":"2024"},{"key":"ref22","article-title":"Transformer-based conditional variational autoencoder for controllable story generation","author":"Fang","year":"2021"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.insights-1.5"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1810.04805"},{"issue":"8","key":"ref25","first-page":"1","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI Blog"},{"key":"ref26","first-page":"140:1","article-title":"Exploring the limits of transfer learning with a unified text-to-text transformer","volume":"21","author":"Raffel","year":"2020","journal-title":"J. Mach. Learn. Res."},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1525\/9780520940420-020"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/n19-1021"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.235"},{"key":"ref30","first-page":"986","article-title":"DailyDialog: A manually labelled multi-turn dialogue dataset","volume-title":"Proc. 8th Int. Joint Conf. Natural Lang. Process.","author":"Li","year":"2017"},{"key":"ref31","article-title":"BERTScore: Evaluating text generation with BERT","volume-title":"Proc. 8th Int. Conf. Learn. Represent.","author":"Zhang","year":"2020"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.sigdial-1.28"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.270"},{"key":"ref34","first-page":"1353","article-title":"Generative neural machine translation","volume-title":"Proc. 32nd Annu. Conf. Neural Inf. Process. Syst.","author":"Shah","year":"2018"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.390"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v33i01.33017402"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1145\/3404835.3462995"},{"key":"ref38","first-page":"27730","article-title":"Training language models to follow instructions with human feedback","volume-title":"Proc. 36th Annu. Conf. Neural Inf. Process. Syst.","author":"Ouyang","year":"2022"},{"key":"ref39","first-page":"56998","article-title":"Latent diffusion for language generation","volume-title":"Proc. 37th Annu. Conf. Neural Inf. Process. Syst.","author":"Lovelace","year":"2023"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.333"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.emnlp-main.378"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.naacl-main.51"},{"key":"ref44","article-title":"LlaMaVAE: Guiding large language model generation via continuous latent sentence spaces","author":"Zhang","year":"2023"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.findings-acl.146"},{"key":"ref46","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023"},{"key":"ref47","first-page":"19470","article-title":"Improving variational autoencoders with density gap-based regularization","volume-title":"Proc. 36th Annu. Conf. Neural Inf. Process. Syst.","author":"Zhang","year":"2022"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref49","first-page":"2397","article-title":"Avoiding latent variable collapse with generative Skip models","volume-title":"Proc. 22nd InterNat. Conf. Artif. Intell. Statist.","author":"Dieng","year":"2019"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D13-1170"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/d15-1167"},{"key":"ref52","first-page":"3881","article-title":"Improved variational autoencoders for text modeling using dilated convolutions","volume-title":"Proc. 34th Int. Conf. Mach. Learn.","author":"Yang","year":"2017"},{"key":"ref53","article-title":"Decoupled weight decay regularization","volume-title":"Proc. 7th Int. Conf. Learn. Represent.","author":"Loshchilov","year":"2019"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2017.12.012"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref56","article-title":"beta-VAE: Learning basic visual concepts with a constrained variational framework","volume-title":"Proc. 5th Int. Conf. Learn. Represent.","author":"Higgins","year":"2017"},{"key":"ref57","article-title":"Improving variational inference with inverse autoregressive flow","author":"Kingma","year":"2016"},{"key":"ref58","article-title":"Adaptive loss scaling for mixed precision training","author":"Zhao","year":"2019"},{"key":"ref59","article-title":"ELBO surgery: Yet another way to carve up the variational evidence lower bound","volume-title":"Proc. NIPS Workshop Adv. Approx. Bayesian Inference","author":"Hoffman","year":"2016"},{"key":"ref60","article-title":"Importance weighted autoencoders","volume-title":"Proc. 4th Int. Conf. Learn. Represent.","author":"Burda","year":"2016"},{"key":"ref61","first-page":"10534","article-title":"On variational learning of controllable representations for text without supervision","volume-title":"Proc. 37th Int. Conf. Mach. Learn.","author":"Xu","year":"2020"},{"key":"ref62","first-page":"4816","article-title":"MAUVE: Measuring the gap between neural text and human text using divergence frontiers","volume-title":"Proc. 35th Annu. Conf. Neural Inf. Process. Syst.","author":"Pillutla","year":"2021"},{"key":"ref63","first-page":"74","article-title":"ROUGE: A package for automatic evaluation of summaries","volume-title":"Proc. ACL Text Summarization Branches Out Workshop","author":"Lin","year":"2004"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/n16-1014"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1145\/3209978.3210080"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/p17-1099"},{"key":"ref67","article-title":"RoBERTa: A robustly optimized BERT pretraining approach","author":"Liu","year":"2019"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D19-1407"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.3115\/1073083.1073135"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.coling-main.216"}],"container-title":["IEEE Transactions on Artificial Intelligence"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/9078688\/11370309\/11045331.pdf?arnumber=11045331","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,3]],"date-time":"2026-02-03T20:57:43Z","timestamp":1770152263000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11045331\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2]]},"references-count":70,"journal-issue":{"issue":"2"},"URL":"https:\/\/doi.org\/10.1109\/tai.2025.3581507","relation":{},"ISSN":["2691-4581"],"issn-type":[{"value":"2691-4581","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2]]}}}