search

WentianZhang-ML / AdaptiveMix

This is an official pytorch implementation of 'AdaptiveMix: Robust Feature Representation via Shrinking Feature Space' (accepted by CVPR2023).
38 stars 1 forks source link

Training problems #4

Closed ProgrammerZhujinming closed 6 months ago

ProgrammerZhujinming commented 8 months ago

Hi, I got some problems when I train the model :

run_cat_styleganv2.sh: line 18: 4037543 Segmentation fault (core dumped) CUDA_VISIBLE_DEVICES=1 python3 generation/train.py --outdir='generation/experiments/AdaptiveMix_cat_StyleGAN_V2' --gpus=1 --data='data/afhq.zip' --mirror=1 --cfg=paper256 --aug=noaug --adaptivemix=true --noise_std=0.05 (AdaptiveMix) root@c40f242965b5:/home/zhujinming/AdaptiveMix-main# /root/anaconda3/envs/AdaptiveMix/lib/python3.8/multiprocessing/resource_tracker.py:203: UserWarning: resource_tracker: There appear to be 34 leaked semaphore objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %

Can you help me? Thanks for your reply!

WentianZhang-ML commented 8 months ago

Hi, happy new year! How many gpus do you have? If there is only one gpu, you should set CUDA_VISIBLE_DEVICES as 0. Otherwise, you can reduce the num_workers when you load the dataset.

ProgrammerZhujinming commented 7 months ago

Thanks for your reply and happy new year! Actually, I have two gpus and I set CUDA_VISIBLE_DEVICES as 1. However, I didn't notice the setting of num_workers. I will try to run again.