Couldn't Reproduce the Code

[monody1]

Hello,

I attempted to reproduce the code, but encountered some issues. Could you please provide some insights into how much memory is expected to be used? Additionally, I suspect there might be a memory leak.

dataset_name:SUN397 torch.cuda.memory_allocated:6.21 GB 0%|▏ | 1/503 [00:01<14:33, 1.74s/it] dataset_name:Cars torch.cuda.memory_allocated:11.91 GB 0%|▎ | 1/394 [00:01<08:39, 1.32s/it] dataset_name:RESISC45 torch.cuda.memory_allocated:17.61 GB 1%|▋ | 1/169 [00:01<03:21, 1.20s/it] dataset_name:EuroSAT torch.cuda.memory_allocated:23.31 GB Using downloaded and verified file: /home/monody/AdaMerging/dataset/svhn/train_32x32.mat Using downloaded and verified file: /home/monody/AdaMerging/dataset/svhn/test_32x32.mat 0%| | 1/1627 [00:01<38:39, 1.43s/it] dataset_name:SVHN torch.cuda.memory_allocated:29.01 GB 0%| | 0/790 [00:00<?, ?it/s] torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.00 MiB (GPU 0; 31.74 GiB total capacity; 29.65 GiB already allocated; 25.12 MiB free; 31.12 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

Thank you for your assistance.