HEP task for Variable compression using AutoEncoders
$ python3 main.py -d cpu --lr 0.01 --configs configs/*.yaml --type vanilla \
--num_steps 50000 --train_pickle_path dataset/all_jets_train_4D_100_percent.pkl \
--test_pickle_path dataset/all_jets_test_4D_100_percent.pkl --logdir logdir/
--export_config --export_saved_model final_saved_model
or if you already have a config file exported once using --export_config
feel free to reuse that exported config file
to pass the arguments too.
$ python3 main.py --configs biggie_config.yaml
Check $ python3 main.py -h
for more options
Plots for the task can be generated using:
$ python3 task.py --saved_model_path final_saved_model/ \
--infer_pickle_path dataset/all_jets_test_4D_100_percent.pkl \
--logdir logdir/inference --configs configs/*.yaml --nbins 10
One can also export the plots to JPEGs by passing the directory to export to using --export_jpegs
tag
Please check $ python3 task.py -h
for more options.
For the sample notebook, checkout HEP_Atlas_AutoEncoder.ipynb. It is recommended to run the notebook on Google Colab.