{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,28]],"date-time":"2026-10-28T22:33:03Z","timestamp":1793226783736,"version":"4.5.0"},"reference-count":128,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"6","license":[{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"funder":[{"name":"Medical Text Feature Representations based on Pre-trained Language Models","award":["871238"],"award-info":[{"award-number":["871238"]}]},{"name":"Faculty Research","award":["DB24A4"],"award-info":[{"award-number":["DB24A4"]}]},{"name":"Faculty Research","award":["SDS24A8"],"award-info":[{"award-number":["SDS24A8"]}]},{"name":"Direct Grant","award":["DR25E8"],"award-info":[{"award-number":["DR25E8"]}]},{"DOI":"10.13039\/501100014736","name":"Lingnan University","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100014736","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Research Grants Council of the Hong Kong Special Administrative Region","award":["R1015-23"],"award-info":[{"award-number":["R1015-23"]}]},{"name":"Research Grants Council of the Hong Kong Special Administrative Region","award":["UGC\/FDS16\/E17\/23"],"award-info":[{"award-number":["UGC\/FDS16\/E17\/23"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Pattern Anal. Mach. Intell."],"published-print":{"date-parts":[[2026,6]]},"DOI":"10.1109\/tpami.2026.3657354","type":"journal-article","created":{"date-parts":[[2026,1,27]],"date-time":"2026-01-27T05:48:07Z","timestamp":1769492887000},"page":"6107-6126","source":"Crossref","is-referenced-by-count":73,"title":["Parameter-Efficient Fine-Tuning Methods for Pretrained Language Models: A Critical Review and Assessment"],"prefix":"10.1109","volume":"48","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0530-3505","authenticated-orcid":false,"given":"Lingling","family":"Xu","sequence":"first","affiliation":[{"name":"School of Science and Technology, Hong Kong Metropolitan University, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0965-3617","authenticated-orcid":false,"given":"Haoran","family":"Xie","sequence":"additional","affiliation":[{"name":"School of Data Science, Lingnan University, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7631-2535","authenticated-orcid":false,"given":"S. Joe","family":"Qin","sequence":"additional","affiliation":[{"name":"School of Data Science, Lingnan University, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0020-077X","authenticated-orcid":false,"given":"Xiaohui","family":"Tao","sequence":"additional","affiliation":[{"name":"University of the School of Mathematics, Physics and Computing, Southern Queensland, Queensland, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3976-0053","authenticated-orcid":false,"given":"Fu Lee","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Science and Technology, Hong Kong Metropolitan University, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1810.04805"},{"key":"ref3","article-title":"RoBERTa: A robustly optimized bert pretraining approach","author":"Liu"},{"issue":"8","key":"ref4","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI Blog"},{"issue":"1","key":"ref5","first-page":"5485","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":"ref6","article-title":"Democratizing access to large-scale language models with OPT-175B","author":"Zhang","year":"2022","journal-title":"Meta AI Blog"},{"key":"ref7","article-title":"Bloom: A 176b-parameter open-access multilingual language model","author":"Scao","year":"2022"},{"key":"ref8","article-title":"LLAMA: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref9","article-title":"The LLAMA 3 herd of models","author":"Dubey","year":"2024"},{"key":"ref10","first-page":"2790","article-title":"Parameter-efficient transfer learning for NLP","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Houlsby"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acllong.353"},{"key":"ref12","first-page":"1","article-title":"LoRA: Low-rank adaptation of large language models","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Hu"},{"key":"ref13","doi-asserted-by":"crossref","DOI":"10.21203\/rs.3.rs-1553541\/v1","article-title":"Delta tuning: A comprehensive study of parameter efficient methods for pre-trained language models","author":"Ding","year":"2022"},{"key":"ref14","article-title":"Scaling down to scale up: A guide to parameter-efficient fine-tuning","author":"Lialin","year":"2023"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.findings-emnlp.41"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0357"},{"key":"ref17","first-page":"1","article-title":"Towards a unified view of parameter-efficient transfer learning","volume-title":"Proc. Int. Conf. Learn. Representations","author":"He"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.626"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.emnlp-main.444"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.eacl-main.39"},{"key":"ref21","article-title":"Mera: Merging pretrained adapters for few-shot learning","author":"He","year":"2023"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.47"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-eacl.153"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.381"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.243"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1016\/j.aiopen.2023.08.012"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.346"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.emnlp-main.446"},{"key":"ref29","first-page":"1","article-title":"Multitask prompt tuning enables parameter-efficient transfer learning","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Wang"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.52202\/068431-0944"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.52202\/068431-0142"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.eacl-main.60"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.findings-naacl.64"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1145\/3583780.3614904"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-short.1"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-acl.539"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.emnlp-main.174"},{"key":"ref38","first-page":"24193","article-title":"Training neural networks with fixed sparse masks","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Sung"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.125"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.749"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.378"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v37i11.26505"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.acl-long.568"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-85747-8_4"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.eacl-main.239"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-short.107"},{"key":"ref47","article-title":"IncreLoRA: Incremental parameter allocation method for parameter-efficient fine-tuning","author":"Zhang","year":"2023"},{"key":"ref48","article-title":"Delta-LoRA: Fine-tuning high-rank parameters with the delta of low-rank matrices","author":"Zi","year":"2023"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-acl.178"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0441"},{"key":"ref51","article-title":"Qa-LoRA: Quantization-aware low-rank adaptation of large language models","author":"Xu","year":"2023"},{"key":"ref52","article-title":"LoftQ: LoRA-fine-tuning-aware quantization for large language models","author":"Li","year":"2023"},{"key":"ref53","article-title":"Int2. 1: Towards fine-tunable quantized large language models with error correction through low-rank adaptation","author":"Chai","year":"2023"},{"key":"ref54","article-title":"LQ-LoRA: Low-rank plus quantized matrix decomposition for efficient language model finetuning","author":"Guo","year":"2023"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.findings-naacl.102"},{"key":"ref56","article-title":"Bayesian low-rank adaptation for large language models","author":"Yang","year":"2023"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-emnlp.994"},{"key":"ref58","article-title":"LoraHub: Efficient cross-task generalization via dynamic LoRA composition","author":"Huang","year":"2023"},{"key":"ref59","article-title":"MoeLoRA: An MOE-based parameter efficient fine-tuning method for multi-task medical applications","author":"Liu","year":"2023"},{"key":"ref60","article-title":"Parameter efficient multi-task model fusion with partial linearization","author":"Tang","year":"2023"},{"key":"ref61","first-page":"1022","article-title":"Compacter: Efficient low-rank hypercomplex adapter layers","volume":"34","author":"Karimi Mahabadi","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.433"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00662"},{"key":"ref64","first-page":"9853","article-title":"Sparse structure search for delta tuning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Hu"},{"key":"ref65","first-page":"1","article-title":"Parameter-efficient fine-tuning design spaces","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Chen"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.emnlp-main.388"},{"key":"ref67","first-page":"2184","volume-title":"Proc. Findings Conf. Empir. Methods Natural Lang. Process.","author":"He"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.418"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref70","article-title":"Layer normalization","author":"Ba","year":"2016"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.1016\/j.nlp.2023.100009"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1145\/3366423.3380060"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D19-1146"},{"key":"ref74","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-eacl.143"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1016\/0925-2312(93)90006-O"},{"key":"ref76","doi-asserted-by":"publisher","DOI":"10.1145\/3593590"},{"key":"ref77","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.emnlp-main.617"},{"key":"ref78","first-page":"506","article-title":"Learning multiple visual domains with residual adapters","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Rebuffi"},{"key":"ref79","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.findings-emnlp.240"},{"key":"ref80","doi-asserted-by":"publisher","DOI":"10.1145\/2766963"},{"key":"ref81","first-page":"22045","article-title":"Model fusion via optimal transport","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Singh"},{"key":"ref82","first-page":"1","article-title":"Hypernetworks","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Ha"},{"key":"ref83","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.692"},{"key":"ref84","first-page":"1","article-title":"Pruning filters for efficient convnets","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Li"},{"key":"ref85","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W19-4828"},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D19-1445"},{"key":"ref87","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-05318-5_3"},{"key":"ref88","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.l803.03635"},{"key":"ref89","first-page":"244","article-title":"Fastfood-computing Hilbert space expansions in loglinear time","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Le"},{"key":"ref90","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.01152"},{"key":"ref91","first-page":"20378","article-title":"Movement pruning: Adaptive sparsity by fine-tuning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Sanh"},{"key":"ref92","first-page":"1","article-title":"GPTQ: Accurate post-training quantization for generative pre-trained transformers","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Frantar"},{"key":"ref93","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1992.4.3.448"},{"key":"ref94","doi-asserted-by":"publisher","DOI":"10.1145\/3377930.3389838"},{"key":"ref95","first-page":"1","article-title":"Editing models with task arithmetic","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Ilharco"},{"key":"ref96","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0552"},{"key":"ref97","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0310"},{"key":"ref98","first-page":"1","article-title":"Beyond fully-connected layers with quaternions: Parameterization of hypercomplex multiplications with $1\/n$1\/n parameters","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Zhang"},{"key":"ref99","article-title":"Outrageously large neural networks: The sparsely-gated mixture-of-experts layer","author":"Shazeer","year":"2017"},{"key":"ref100","first-page":"1","article-title":"Pruning neural networks at initialization: Why are we missing the mark","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Frankle"},{"key":"ref101","first-page":"1","article-title":"SNIP: Single-shot network pruning based on connection sensitivity","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Lee"},{"key":"ref102","first-page":"1","article-title":"Albert: A lite BERT for self-supervised learning of language representations","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Lan"},{"key":"ref103","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W18-5446"},{"key":"ref104","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.naacl-main.41"},{"key":"ref105","article-title":"Stanford alpaca: An instruction-following LLaMA model","author":"Taori","year":"2023"},{"key":"ref106","first-page":"1","article-title":"Measuring massive multitask language understanding","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Hendrycks"},{"key":"ref107","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.eacl-long.164"},{"key":"ref108","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.naacl-long.401"},{"key":"ref109","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D19-1165"},{"key":"ref110","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.acl-main.421"},{"key":"ref111","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.800"},{"key":"ref112","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.findings-emnlp.410"},{"key":"ref113","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.naacl-main.130"},{"key":"ref114","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.521"},{"key":"ref115","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.194"},{"key":"ref116","first-page":"1086","article-title":"Moderate-fitting as a natural backdoor defender for pre-trained language models","volume-title":"Adv. Neural Inf. Process. Syst.","author":"Zhu","year":"2022"},{"key":"ref117","article-title":"Fewer is more: Trojan attacks on parameter-efficient fine-tuning","author":"Hong","year":"2023"},{"key":"ref118","article-title":"The expressive power of low-rank adaptation","author":"Zeng","year":"2023"},{"key":"ref119","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v37i1.25160"},{"key":"ref120","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01605"},{"key":"ref121","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.00516"},{"key":"ref122","first-page":"26462","article-title":"St-adapter: Parameter-efficient image-to-video transfer learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Pan"},{"key":"ref123","doi-asserted-by":"publisher","DOI":"10.1145\/3676536.3676796"},{"key":"ref124","article-title":"Parameter efficient reinforcement learning from human feedback","author":"Sidahmed","year":"2024"},{"key":"ref125","article-title":"Tina: Tiny reasoning models via lora","author":"Wang","year":"2025"},{"key":"ref126","article-title":"Deepseekmath: Pushing the limits of mathematical reasoning in open language models","author":"Shao","year":"2024"},{"key":"ref127","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.findings-emnlp.738"},{"key":"ref128","first-page":"1","article-title":"ExPLoRA: Parameter-efficient extended pre-training to adapt vision transformers under domain shifts","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Khanna"}],"container-title":["IEEE Transactions on Pattern Analysis and Machine Intelligence"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/34\/11512030\/11364256.pdf?arnumber=11364256","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,12]],"date-time":"2026-05-12T01:53:09Z","timestamp":1778550789000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11364256\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6]]},"references-count":128,"journal-issue":{"issue":"6"},"URL":"https:\/\/doi.org\/10.1109\/tpami.2026.3657354","relation":{},"ISSN":["0162-8828","2160-9292","1939-3539"],"issn-type":[{"value":"0162-8828","type":"print"},{"value":"2160-9292","type":"electronic"},{"value":"1939-3539","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6]]}}}