JianmingLi-888 comments

Repositories
Issues
Comments

Results 3 comments of


                                            JianmingLi-888

How can I train with single GPU ?

actuallay I just delete one line in train.py from ` dist.init_process_group( backend = 'nccl', init_method = 'tcp://127.0.0.1:33274', world_size = torch.cuda.device_count(), rank=args.local_rank ) ` to ` dist.init_process_group( backend = 'nccl', world_size...

如何理解trainid=-1

just ignore it, README.md in cityscapes dataset tells what it is > @774911840 你好，我在看源码的时候也遇到了这样的疑问，请问你解决这个疑问了吗？

How can I train with single GPU ?

> > Hello, thank you for sharing the awesome code. > > I'm using CUDA 11.1, PyTorch 1.8.0 with my single GPU(RTX 3090) and I have tested that evaluation and...