taco
taco copied to clipboard
Question on adding API coverage to python bindings
Hi! I have opened this issue to ask some general questions regarding the GSoC project that aims to complete the Python bindings for TACO.
From what I've discussed with Hameer Abbasi(unable to tag him here), I've come to realize that implementing universal functions are among the goals right now. For instance, we cannot do something like applying certain ufunc
directly to a tensor. It has to be densified. The following lines of code give an error as expected.
>>> import pytaco as pt
>>> import numpy as np
>>> from pytaco import dense, compressed
>>> A = pt.tensor([512, 64, 2048], pt.format([dense, compressed, compressed]), pt.float64)
>>> np.exp(A)
What more would TACO as a community like to see implemented as part of the GSoC project?
cc @fredrikbk
This ties directly into the array API we have discussed at the meetings. If we could provide @sayandip18 with some guidance on tasks he may be able to help out.