Quick Start
1. Install Singularity (optional)
If you don't want to install Singularity, make sure that you have all dependecies from Singularity.def (python3, numpy, pytorch, etc.)
a. Pull an image
singularity pull library://k3nfalt/default/python_ml:sha256.efcd1fc038228cb7eb0f6f1942dfbaa439cd95d6463015b83ceb2dbaad9e1e98
b. Open a shell console of the image
singularity shell --nv ~/python_ml_sha256.efcd1fc038228cb7eb0f6f1942dfbaa439cd95d6463015b83ceb2dbaad9e1e98.sif
2. Prepare scripts for experiments
PYTHONPATH=./code python3 ./code/distributed_optimization_library/experiments/dasha_partial_participation/config_libsvm_dasha_partial_particiaption.py
--dumps_path SOME_PATH --dataset_path PATH_TO_FOLDER_WITH_DATASET --dataset real-sim
--experiments_name EXPERIMENT_NAME
--num_nodes_list 100 --step_size_range -10 0 --number_of_seeds 1 --number_of_iterations 5000000
--algorithm_names zero_marina_sync_stochastic zero_marina_partial_participation_stochastic --cpus_per_task 11
--number_of_processes 10 --time 10 --parallel --compressors rand_k --number_of_coordinates 200 --quality_check_rate 1000
--oracle stochastic --mega_batch 10000 --batch_size 1 --function stochastic_logistic_regression --logistic_regression_nonconvex 0.001
--partial_participation_probabilities 1.0 0.5 0.1 0.01
3. Execute scripts
sh SOME_PATH/EXPERIMENT_NAME/singularity_*.sh
4. Plot results
PYTHONPATH=./code python3 ./code/distributed_optimization_library/experiments/plots/dasha_partial_participation/plot_vr-marina_real-sim_stochastic.py
--dumps_paths SOME_PATH/EXPERIMENT_NAME
--output_path SOME_PATH_FOR_PLOTS
--ignore_methods "VR-MARINA (online)" "DASHA-MVR"
One can find all other scripts here that generate experiments from the paper.