FedOne: Query-Efficient Federated Learning for Black-box Discrete Prompt Learning

Ganyu Wang, Jinjie Fang, Maxwell Juncheng Yin, Bin Gu, Xi Chen, Boyu Wang, Yi Chang, Charles Ling
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:62857-62889, 2025.

Abstract

Black-Box Discrete Prompt Learning (BDPL) is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting Federated Learning (FL) to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called FedOne, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-wang25ab, title = {{F}ed{O}ne: Query-Efficient Federated Learning for Black-box Discrete Prompt Learning}, author = {Wang, Ganyu and Fang, Jinjie and Yin, Maxwell Juncheng and Gu, Bin and Chen, Xi and Wang, Boyu and Chang, Yi and Ling, Charles}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {62857--62889}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/wang25ab/wang25ab.pdf}, url = {https://proceedings.mlr.press/v267/wang25ab.html}, abstract = {Black-Box Discrete Prompt Learning (BDPL) is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting Federated Learning (FL) to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called FedOne, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.} }
Endnote
%0 Conference Paper %T FedOne: Query-Efficient Federated Learning for Black-box Discrete Prompt Learning %A Ganyu Wang %A Jinjie Fang %A Maxwell Juncheng Yin %A Bin Gu %A Xi Chen %A Boyu Wang %A Yi Chang %A Charles Ling %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-wang25ab %I PMLR %P 62857--62889 %U https://proceedings.mlr.press/v267/wang25ab.html %V 267 %X Black-Box Discrete Prompt Learning (BDPL) is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting Federated Learning (FL) to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called FedOne, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.
APA
Wang, G., Fang, J., Yin, M.J., Gu, B., Chen, X., Wang, B., Chang, Y. & Ling, C.. (2025). FedOne: Query-Efficient Federated Learning for Black-box Discrete Prompt Learning. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:62857-62889 Available from https://proceedings.mlr.press/v267/wang25ab.html.

Related Material