Example¶
This document provides a brief overview of project_manager
’s basic functionality.
Environment setup¶
[2]:
tree
.
|____dummy_project
| |____my_conf.yaml
| |____run.py
|____config.yaml
[3]:
cat config.yaml
project_source: dummy_project
working_dir: tmp
exec_command:
- python3 run.py
result_files:
- results
base_config: dummy_project/my_conf.yaml
config_parameters:
- key: message
values: [A, B, C]
[4]:
cat dummy_project/my_conf.yaml
message: 'this is important'
[5]:
cat dummy_project/run.py
import os
import yaml
def main():
with open('my_conf.yaml') as fd:
config = yaml.full_load(fd)
os.makedirs('results')
with open('results/data.txt', 'w') as fd:
fd.write(config['message'])
if __name__ == '__main__':
main()
Pipeline execution¶
Setup directory for each configuration¶
[6]:
project_manager build -c config.yaml
Setting up environments: 100%|███████████████████| 3/3 [00:00<00:00, 238.72it/s]
Execute scripts for each configuration¶
[7]:
project_manager run -c config.yaml
0%| | 0/3 [00:00<?, ?it/s]run.message=A
> python3 run.py
33%|███████████████ | 1/3 [00:00<00:00, 5.37it/s]run.message=C
> python3 run.py
67%|██████████████████████████████ | 2/3 [00:00<00:00, 5.76it/s]run.message=B
> python3 run.py
100%|█████████████████████████████████████████████| 3/3 [00:00<00:00, 6.30it/s]
Gather results from each run¶
[8]:
project_manager gather -c config.yaml
run.message=A
> results/data.txt
run.message=C
> results/data.txt
run.message=B
> results/data.txt
Investigate results¶
[9]:
tree tmp/
|____
|____aggregated_results
| |____results
| | |____data.message=B.txt
| | |____data.message=A.txt
| | |____data.message=C.txt
|____run.message=A
| |____my_conf.yaml
| |____results
| | |____data.txt
| |____run.py
|____run.message=C
| |____my_conf.yaml
| |____results
| | |____data.txt
| |____run.py
|____run.message=B
| |____my_conf.yaml
| |____results
| | |____data.txt
| |____run.py
[10]:
cat tmp/aggregated_results/results/*
ABC