failed to replicate the results

[youyeg]

I have ubuntu 16.04 with two GPUs, and I did these steps, but it fails to perform paper_experiments

1. conda create --name leafenv python=3.5
2. git clone https://github.com/TalwalkarLab/leaf
3. pip3 install -r requirements.txt
4. ./paper_experiments/femnist.sh

but after some rounds with zero

s Failed to import the site module Traceback (most recent call last): File "/lib/python3.5/site.py", line 703, in <module> main() File "/lib/python3.5/site.py", line 670, in main virtual_install_main_packages() File "/lib/python3.5/site.py", line 553, in virtual_install_main_packages f = open(os.path.join(os.path.dirname(__file__), 'orig-prefix.txt'))

mv: cannot stat 'sys_metrics.csv': No such file or directory

mv: cannot stat 'stat_metrics.csv': No such file or directory

[oscarciceri]

did you find a solution to this? I have the same problem

[tdye24]

I have ubuntu 16.04 with two GPUs, and I did these steps, but it fails to perform paper_experiments

1. conda create --name leafenv python=3.5
2. git clone https://github.com/TalwalkarLab/leaf
3. pip3 install -r requirements.txt
4. ./paper_experiments/femnist.sh

but after some rounds with zero

s Failed to import the site module Traceback (most recent call last): File "/lib/python3.5/site.py", line 703, in <module> main() File "/lib/python3.5/site.py", line 670, in main virtual_install_main_packages() File "/lib/python3.5/site.py", line 553, in virtual_install_main_packages f = open(os.path.join(os.path.dirname(__file__), 'orig-prefix.txt'))

mv: cannot stat 'sys_metrics.csv': No such file or directory

mv: cannot stat 'stat_metrics.csv': No such file or directory

The names of generated files are metrics_stat.csv and metrics_sys.csv，you just need to rename them.

[janesser]

Complete walkthrough works for me like

git checkout https://github.com/TalwalkarLab/leaf
cd leaf
conda create --name leafenv python=3.5
conda activate leafenv
conda install gcc_linux-64 gxx_linux-64
pip3 install -r requirements.txt
cd data/femnist/
./preprocess.sh
cd ../../paper_experiments
./femnist.sh

Then sampling of data begins...