Federated Full-Parameter Tuning of Billion-Sized Language Models with Communication Cost under 18 Kilobytes
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917675420090368 |
|---|---|
| author | Qin, Zhen Chen, Daoyuan Qian, Bingchen Ding, Bolin Li, Yaliang Deng, Shuiguang |
| author_facet | Qin, Zhen Chen, Daoyuan Qian, Bingchen Ding, Bolin Li, Yaliang Deng, Shuiguang |
| contents | Pre-trained large language models (LLMs) need fine-tuning to improve their responsiveness to natural language instructions. Federated learning offers a way to fine-tune LLMs using the abundant data on end devices without compromising data privacy. Most existing federated fine-tuning methods for LLMs rely on parameter-efficient fine-tuning techniques, which may not reach the performance height possible with full-parameter tuning. However, federated full-parameter tuning of LLMs is a non-trivial problem due to the immense communication cost. This work introduces FedKSeed that employs zeroth-order optimization with a finite set of random seeds. It significantly reduces transmission requirements between the server and clients to just a few random seeds and scalar gradients, amounting to only a few thousand bytes, making federated full-parameter tuning of billion-sized LLMs possible on devices. Building on it, we develop a strategy enabling probability-differentiated seed sampling, prioritizing perturbations with greater impact on model accuracy. Experiments across six scenarios with various LLMs, datasets and data partitions demonstrate that our approach outperforms existing federated LLM fine-tuning methods in both communication efficiency and new task generalization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2312_06353 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Federated Full-Parameter Tuning of Billion-Sized Language Models with Communication Cost under 18 Kilobytes Qin, Zhen Chen, Daoyuan Qian, Bingchen Ding, Bolin Li, Yaliang Deng, Shuiguang Machine Learning Distributed, Parallel, and Cluster Computing Pre-trained large language models (LLMs) need fine-tuning to improve their responsiveness to natural language instructions. Federated learning offers a way to fine-tune LLMs using the abundant data on end devices without compromising data privacy. Most existing federated fine-tuning methods for LLMs rely on parameter-efficient fine-tuning techniques, which may not reach the performance height possible with full-parameter tuning. However, federated full-parameter tuning of LLMs is a non-trivial problem due to the immense communication cost. This work introduces FedKSeed that employs zeroth-order optimization with a finite set of random seeds. It significantly reduces transmission requirements between the server and clients to just a few random seeds and scalar gradients, amounting to only a few thousand bytes, making federated full-parameter tuning of billion-sized LLMs possible on devices. Building on it, we develop a strategy enabling probability-differentiated seed sampling, prioritizing perturbations with greater impact on model accuracy. Experiments across six scenarios with various LLMs, datasets and data partitions demonstrate that our approach outperforms existing federated LLM fine-tuning methods in both communication efficiency and new task generalization. |
| title | Federated Full-Parameter Tuning of Billion-Sized Language Models with Communication Cost under 18 Kilobytes |
| topic | Machine Learning Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2312.06353 |