tsai icon indicating copy to clipboard operation
tsai copied to clipboard

Published 20 hours ago • verified publisher icon timeseriesAI

Minirocket cuda out of memory

Open filbert-c opened this issue 3 years ago • 0 comments

Hi, in the tutorial nb we could use the pytorch minirocket for any # samples. My dataset is of size. 10 000 000,2, 32 but i run out of memory even with chunk size =4. Is the memory not being cleared on lazy load?

image

RuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 15.78 GiB total capacity; 14.56 GiB already allocated; 18.00 MiB free; 14.57 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

filbert-c avatar Jul 19 '22 03:07 filbert-c