{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/few-shot-parameter-efficient-fine-tuning-is","title":"Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learning","arxiv_id":"2205.05638","date":"2022-05-11","proceeding":null,"authors":["Haokun Liu","Derek Tam","Mohammed Muqeeth","Jay Mohta","Tenghao Huang","Mohit Bansal","Colin Raffel"],"abstract":"Few-shot in-context learning (ICL) enables pre-trained language models to perform a previously-unseen task without any gradient-based training by feeding a small number of training examples as part of the input. ICL incurs substantial computational, memory, and storage costs because it involves processing all of the training examples every time a prediction is made. Parameter-efficient fine-tuning (PEFT) (e.g. adapter modules, prompt tuning, sparse update methods, etc.) offers an alternative paradigm where a small set of parameters are trained to enable a model to perform the new task. In this paper, we rigorously compare few-shot ICL and PEFT and demonstrate that the latter offers better accuracy as well as dramatically lower computational costs. Along the way, we introduce a new PEFT method called (IA)$^3$ that scales activations by learned vectors, attaining stronger performance while only introducing a relatively tiny amount of new parameters. We also propose a simple recipe based on the T0 model called T-Few that can be applied to new tasks without task-specific tuning or modifications. We validate the effectiveness of T-Few on completely unseen tasks by applying it to the RAFT benchmark, attaining super-human performance for the first time and outperforming the state-of-the-art by 6% absolute. All of the code used in our experiments is publicly available.","url_abs":"https://arxiv.org/abs/2205.05638v2","url_pdf":"https://arxiv.org/pdf/2205.05638v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"few-shot-parameter-efficient-fine-tuning-is","repo_url":"https://github.com/r-three/t-few","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"few-shot-parameter-efficient-fine-tuning-is","repo_url":"https://github.com/kohakublueleaf/lycoris","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"few-shot-text-classification","task_name":"Few-Shot Text Classification"},{"task_slug":"in-context-learning","task_name":"In-Context Learning"},{"task_slug":"parameter-efficient-fine-tuning","task_name":"parameter-efficient fine-tuning"}],"methods":[{"method_slug":"adapter","method_name":"Adapter"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/few-shot-text-classification-on-raft","task":"Few-Shot Text Classification","dataset":"RAFT","model":"T-Few","rank_in_archive_order":1,"of":9,"metrics":{" Over":"0.95","ADE":"0.804","Avg":"0.758","B77":"0.695","NIS":"0.833","OSE":"0.676","SOT":"0.915","SRI":"0.508","TAI":"0.736","TC":"0.879","TEH":"0.586","ToS":"0.75"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2205.05638","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}