OutOfMemoryError in 09 paper replicating

[jungwoo9]

While following the step to replicate ViT model, I have an OutOfMemoryError during training.

OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 6.00 GiB total capacity; 5.28 GiB already allocated; 0 bytes free; 5.35 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

I tried to change the batch size to 1 / gc.collect() torch.cuda.empty_cache() / os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb=100' / restart the notebook but all of this is not working.

How can I solve this error?