PaddleFL
PaddleFL copied to clipboard
PaddlePaddle
Federated Deep Learning in PaddlePaddle
environment: CentOS Linux release 7.9.2009 (Core) Python3.6（from anaconda3 conda env） pip 21.0.1 cuda 10.0 cudnn 7.6.5 python -m pip install paddlepaddle-gpu==2.0.0.post100 -f https://paddlepaddle.org.cn/whl/stable.html paddlepaddle release 2.0.0（dependencies >=1.8） Then pip install...
### Environment Ubuntu 20.04.2 LTS Python 3.7.7 pip 21.0.1 g++ (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0 Redis server v=5.0.8 PaddlePaddle 1.8.0 cmake version 3.16.0 ### Compilation error message ``` /PaddleFL/build# cmake ../ -DPYTHON_EXECUTABLE=/root/miniconda/bin/python3...
尊敬的开发人员您好，我希望使用profiler工具获取联邦学习过程中各个program的ops时间分布，为此查看了timeline使用文档https://www.paddlepaddle.org.cn/documentation/docs/zh/1.8/advanced_guide/performance_improving/analysis_tools/timeline_cn.html#local  我所理解应该是使用如图中分布式的方法，但关于第2步环境变量的添加没有太明白其中表达的含义，目前我只在fl_trainer的训练过程中增加了profiler工具，最后也生成了一个profile_ps_1794596834表示服务器过程的文件（但似乎不完整，没有recv的操作，只有sum和scale） 希望能够得到您的回复，最好可以稍微介绍一下联邦学习使用profiler（timeline）的教程，非常感谢!
Hi guys, 当我想在单机上尝试`logistic_with_mnist`的时候，在跟随README.md执行后，在本步遇到报错 ``` sh run_standalone.sh train_fc_sigmoid.py ``` 在脚本中python执行过程中报错 ``` Traceback (most recent call last): File "logistic_with_mnist/train_fc_sigmoid.py", line 22, in import numpy as np Traceback (most recent call last): ImportError...
  本人采用虚拟机linux的方式按照说明的步骤安装paddlefl，系统为Ubuntu18.04，python2.7，pip20.3.4 无论用pip安装还是用redis安装都会出现这种情况
Could you clarify implementation details for DPSGD: 1) where clipping operation is implemented - client (trainer) or aggregator (server)? 2) where noise addition is implemented - client (trainer) or aggregator...
1. Inner Step in paddle_fl module. default `inner_step = 10` This variable means that we will send parameters from nodes to sever every 10 steps ? For example, 10.000 lines...
Good afternoon! How I can calculate loss in [this example?](https://github.com/PaddlePaddle/PaddleFL/blob/8b3b982b0ca95c631ce0e51a273b35e5b6bac6ea/python/paddle_fl/mpc/examples/logistic_with_mnist/train_fc_sigmoid.py) I tried next: - In training process add `loss = exe.run(feed=sample, fetch_list=[avg_loss])` and try to decrypt by `load_decrypt_data(loss_file, (1,), decrypt_file_loss)`...
