手把手教你运行GitHub下载下来的python项目

首先装好pycharm,将项目导入至pycharm

settings里有虚拟环境，可以根据requirements配置该项目所需的虚拟环境，有的时候版本太高，需要手动配置，右击文件，open in terminal，用pip install安装

设置超参数

当遇到此类问题时
usage: run_experiment.py [-h] [–test_unseen_clients TEST_UNSEEN_CLIENTS]
[–decentralized] [–sampling_rate SAMPLING_RATE]
[–input_dimension INPUT_DIMENSION]
[–output_dimension OUTPUT_DIMENSION]
[–n_learners N_LEARNERS] [–n_rounds N_ROUNDS]
[–bz BZ] [–local_steps LOCAL_STEPS]
[–log_freq LOG_FREQ]
[–person_input_norm PERSON_INPUT_NORM]
[–model_type MODEL_TYPE]
[–seperate_trans SEPERATE_TRANS] [–device DEVICE]
[–optimizer OPTIMIZER] [–lr_model LR_MODEL]
[–lr_gating LR_GATING] [–lr_lambda LR_LAMBDA]
[–lr_scheduler LR_SCHEDULER]
[–sparse_factor_scheduler SPARSE_FACTOR_SCHEDULER]
[–mu MU] [–track_running_stats TRACK_RUNNING_STATS]
[–online_aggregate ONLINE_AGGREGATE] [–alpha ALPHA]
[–beta BETA] [–sparse_factor SPARSE_FACTOR]
[–fine_grained_block_split FINE_GRAINED_BLOCK_SPLIT]
[–block_wise_prune BLOCK_WISE_PRUNE]
[–importance_prior_para_num IMPORTANCE_PRIOR_PARA_NUM]
[–communication_probability COMMUNICATION_PROBABILITY]
[–q Q] [–locally_tune_clients] [–validation]
[–verbose VERBOSE] [–bi_level_opt BI_LEVEL_OPT]
[–logs_root LOGS_ROOT] [–save_path SAVE_PATH]
[–seed SEED]
[–aggregate_sampled_clients AGGREGATE_SAMPLED_CLIENTS]
[–expname EXPNAME] [–notes NOTES] [–outdir OUTDIR]
dataset_name method
run_experiment.py: error: the following arguments are required: dataset_name, method

在Edit-configures里面输入参数，
README里的参数格式python3 run_experiment.py emnist.sample01 pFedGate --n_learners 1 --bz 128 --log_freq 5 --device cuda --seed 1234 --verbose 1 --expname emnist01_pFedGate --n_rounds=400 --optimizer=sgd --sparse_factor=0.5 --block_wise_prune=1 --fine_grained_block_split=5 --lr_gating=0.1 --lr_model=0.1 --lr_scheduler=reduce_on_plateau --bi_level_opt 0 --online_aggregate 0 --sparse_factor_scheduler=constant
对应图片里如此