We present a new method LiST for efficient fine-tuning of large pre-trained language models (PLMs) in few-shot learning settings. LiST significantly improves over recent methods that adopt prompt fine-tuning using two key techniques. The first one is the use of self-training to leverage large amounts of unlabeled data for prompt-tuning to significantly boost the model performance in few-shot settings. We use self-training in conjunction with meta-learning for re-weighting noisy pseudo-prompt labels. However, traditional self-training is expensive as it requires updating all the model parameters repetitively. Therefore, we use a second technique for light-weight fine-tuning where we introduce a small number of task-specific adapter parameters that are fine-tuned during self-training while keeping the PLM encoder frozen. This also significantly reduces the overall model footprint across several tasks that can now share a common PLM encoder as backbone for inference. Combining the above techniques, LiST not only improves the model performance for few-shot learning on target domains but also reduces the model memory footprint. We present a comprehensive study on six NLU tasks to validate the effectiveness of LiST. The results show that LiST improves by 35% over classic fine-tuning methods and 6% over prompt-tuning with 96% reduction in number of trainable parameters when fine-tuned with no more than 30 labeled examples from each target domain.<\/p>\n","protected":false},"excerpt":{"rendered":"
We present a new method LiST for efficient fine-tuning of large pre-trained language models (PLMs) in few-shot learning settings. LiST significantly improves over recent methods that adopt prompt fine-tuning using two key techniques. The first one is the use of self-training to leverage large amounts of unlabeled data for prompt-tuning to significantly boost the model […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13556,13545],"msr-publication-type":[193715],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-785344","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-human-language-technologies","msr-locale-en_us"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2022-4-1","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"NAACL 2022","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","viewUrl":"false","id":"false","title":"https:\/\/arxiv.org\/pdf\/2110.06274.pdf","label_id":"243109","label":0}],"msr_related_uploader":"","msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Yaqing Wang","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Subhabrata (Subho) Mukherjee","user_id":38308,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Subhabrata (Subho) Mukherjee"},{"type":"user_nicename","value":"Xiaodong Liu","user_id":34877,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Xiaodong Liu"},{"type":"text","value":"Jing Gao","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Ahmed H. Awadallah","user_id":31979,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Ahmed H. Awadallah"},{"type":"user_nicename","value":"Jianfeng Gao","user_id":32246,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Jianfeng Gao"}],"msr_impact_theme":[],"msr_research_lab":[199565],"msr_event":[],"msr_group":[144931,392600,702211],"msr_project":[804847,675762],"publication":[],"video":[],"download":[787018],"msr_publication_type":"article","related_content":{"projects":[{"ID":804847,"post_title":"Reducing AI's Carbon Footprint","post_name":"reducing-ais-carbon-footprint","post_type":"msr-project","post_date":"2022-05-24 08:56:55","post_modified":"2024-01-16 11:11:59","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/reducing-ais-carbon-footprint\/","post_excerpt":"This project develops techniques that enable AI to use computing infrastructure more efficiently. The goals are to maintain predictive accuracy while reducing carbon emissions, whether embodied in manufactured hardware, or produced from electricity usage when green energy is not available.","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/804847"}]}},{"ID":675762,"post_title":"Few-shot Learning","post_name":"xtreme-learning-with-few-labels","post_type":"msr-project","post_date":"2020-07-15 20:35:24","post_modified":"2021-06-23 18:07:32","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/xtreme-learning-with-few-labels\/","post_excerpt":"Deep neural networks including pre-trained language models like BERT, Turing-NLG and GPT-3 require thousands of labeled training examples to obtain state-of-the-art performance for downstream tasks and applications. Such large number of labeled examples are difficult and expensive to acquire in practice -- as we scale these models to hundreds of different languages, thousands of different tasks and domains, as well as for compliant reasons while dealing with sensitive user data. In this project, we develop…","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/675762"}]}}]},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/785344"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":5,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/785344\/revisions"}],"predecessor-version":[{"id":848284,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/785344\/revisions\/848284"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=785344"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=785344"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=785344"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=785344"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=785344"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=785344"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=785344"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=785344"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=785344"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=785344"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=785344"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=785344"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=785344"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=785344"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=785344"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=785344"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}