Summary of Fedpft: Federated Proxy Fine-tuning Of Foundation Models, by Zhaopeng Peng et al.
FedPFT: Federated Proxy Fine-Tuning of Foundation Models
by Zhaopeng Peng, Xiaoliang Fan, Yufan Chen, Zheng Wang, Shirui Pan, Chenglu Wen, Ruisheng Zhang, Cheng Wang
First submitted to arxiv on: 17 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a novel method called Federated Proxy Fine-Tuning (FedPFT), which adapts Foundation Models (FMs) for downstream tasks through Federated Learning (FL). The existing fine-tuning methods are suboptimal due to insufficient tuning and error accumulation. FedPFT consists of two key modules: the sub-FM construction module, which employs layer-wise compression, and the sub-FM alignment module, which conducts distillations at both layer-level and neuron-level. Experimental results on seven datasets (four text and three vision) demonstrate the superiority of FedPFT. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to make Foundation Models work better for different tasks. Right now, people fine-tune these models by giving them smaller versions and adjusting them in small groups. But this can be slow and not very good. The authors came up with a new idea called Federated Proxy Fine-Tuning that makes the process faster and better. It has two main parts: one that helps adjust the model layer by layer, and another that makes sure all the little models are working together correctly. They tested it on lots of different datasets and showed that it works really well. |
Keywords
» Artificial intelligence » Alignment » Federated learning » Fine tuning
