3 ms·

Did you see the « NumPy compatible » part of the title?

• fourier_mode 8y ago
  It is "highly compatible", similar statement can be made about other libs say torch tensors.
  • marmaduke 8y ago
    Having 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 8y ago
    You can in nearly all cases, literally do: import cupy as np and have that just work, so it’s pretty compatible.