Closed hadrianpaulo closed 6 years ago
Hi Hadrian - thanks. I think for this to be practical, the library would have to be built on top of a framework that made GPU support easy : e.g. TensorFlow, PyTorch, Theano - which is not the case right now. So I think your best bet is to try PyMC3 or edward for this.
Closing this issue - no intention to do this. Please use a library with a backend that supports GPUs.
Is there a planned feature for supporting GPU acceleration, like PyMC3 via Theano?
PS: Really awesome library.