PyTorch Can’t Allocate More Memory

Here’s a solution to your biggest Deep Learning woe

Photo by Fredy Jacob on Unsplash

We all love PyTorch for many obvious reasons (i.e. ease of implementing our ideas). But, sometimes you run into an error:

CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 15.90 GiB total capacity; 15.17 GiB already allocated; 15.88 MiB free; 15.18 GiB reserved in total by PyTorch)