karhoutam / pfedla

PyTorch implementation of Layer-wised Model Aggregation for Personalized Federated Learning

Hi there 👋

• 💼 I'm a software engineer of Huawei.
• 🎓 I received my master degree in computer science and technology from Shenzhen University (SZU), China.
• 👀 At SZU, I majored in federated learning, especially personalized federated learning (pFL). I'm still interested in this field and caring its academic progress.
• 🤗 Feel free to contact me for asking questions or seeking collaboration.
• 🧐 I'm currently learning CUDA, MLSys, LLMs stuffs. Just my interests.

pfedla's Issues

想问问大佬我只想使用超网络生成聚合权重矩阵这一部分 应该如何使用呢？

想问一下这个代码在哪运行？

pycharm可以吗？还是要在Linux系统上运行？

项目运行报错 但是pycharm不报错

File "C:\Users\（省略）\PycharmProjects\pFedLA\src\data\run.py", line 21, in
from utils.util import DATASETS_DIR
ModuleNotFoundError: No module named 'utils.util'

关于HyperNetwork的

你好，关于pFedLA中的HyperNetwork的详细说明，有点模糊，可以介绍一些相关书籍or论文嘛

The seed is not fixed

Hi, I reproduced your code with a fixed seed, but the results obtained from multiple runs remain inconsistent. I am seeking guidance on the possible reasons behind this inconsistency and how to achieve consistent results. Thank you.

Problem of the training code

Hello! We appreciate your great work! I have a question that in 'client/base.py' the function 'def _train(self)' is empty. We could not start training but only evaluating. Could you please help me? Thank you very much!

README

"many ohter baseline"

where is the implementation of HeurpFedLA?

Dirichlet distribution

If your code uses Dirichlet distribution, the src/server/base.py file will get an error: self.client_id_indices, self.client_num_in_total = get_client_id_indices(self.args.dataset)