Federated-Learning-Backdoor icon indicating copy to clipboard operation
Federated-Learning-Backdoor copied to clipboard

Published 20 hours ago • verified publisher icon jhcknzzm

A question about the train time

Open xiaozeplus opened this issue 1 year ago • 0 comments

Hi, I'm trying to reproduce the experiments for Cifar-10. When I run the command python main_training.py --run_slurm 0 --GPU_id 0 --start_epoch 1 --attack_num 250 --gradmask_ratio 1.0 --edge_case 0 on Windows OS, the training time of one epoch is close to 8min, and the 1800 epochs train round seems to be unaceptable. Maybe I have some wrong setting? The version of pytorch is 2.1.0, and GPU is NVIDIA GeForce RTX 2060.

xiaozeplus avatar Oct 31 '23 16:10 xiaozeplus