CUDA out of memory

[Mr-Second]

RuntimeError: CUDA out of memory. Tried to allocate 7.74 GiB (GPU 0; 23.69 GiB total capacity; 18.60 GiB already allocated; 2.78 GiB free; 19.05 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
Thanks very much from your work, and I need some assistance to train my own dataset, I change dimension 50 to 3 in convert_label function, but when starting training, it still reports CUDA out of memory error, I don't kown how to fix it