show_attend_and_tell_pytorch
show_attend_and_tell_pytorch copied to clipboard
CUDA Out of Memory issue
Running python main.py with checkpoint None throws up error as
RuntimeError: CUDA out of memory. Tried to allocate 6.12 MiB (GPU 0; 3.95 GiB total capacity; 2.70 GiB already allocated; 19.81 MiB free; 2.17 MiB cached)
You have only 4GB space in GPU. The network might need more than that. Either train it on a system with more GPU space or on google colab.