OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 79.35 GiB total capacity; 32.78 GiB already allocated; 19.19 MiB free; 33.10 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 am trying to fine tune whisper model.Its throwing this error.
I tried with small number of dataset.In collab its works in pro version but this is not working on DGX A100 server.
Need help on this.I am new to this.
I tried setting os.environ[‘PYTORCH_CUDA_ALLOC_CONF’] = ‘max_split_size_mb=100’
But no chnage in the error