3 ms·
You could always go deeper and from some points of view, it's not "from the ground up" enough unless you build your own autograd and tensors from plain numpy ar
by jnhl 2y ago
You could always go deeper and from some points of view, it's not "from the ground up" enough unless you build your own autograd and tensors from plain numpy arrays.
- 0cf8612b2e1e 2y agoNumpy sounds like cheating on the backs of others. Going to need your own hand crafted linear algebra routines.