3 ms·
It is "highly compatible", similar statement can be made about other libs say torch tensors.
by fourier_mode 7y ago
It is "highly compatible", similar statement can be made about other libs say torch tensors.
- marmaduke 7y agoHaving tried to debug some issues between autograd, PyTorch & TensorFlow, I find torch & tf tensors have different enough syntax and naming that one needs to google a bit.
- buildbot 7y agoYou can in nearly all cases, literally do: import cupy as np and have that just work, so it’s pretty compatible.