Re: plan of deep learning team for next stable release?
On 12/1/20 2:07 PM, Mo Zhou wrote:
> I like this revision, although it also presents a radical view point.
> Here is one point that I think should be emphasized: It's
> **impractical** for industry users to go without hardware acceleration.
Exactly, and I wanted to emphasize that by simply stating the cold, hard
But it's not impractical just for industry users, though. The toolchains
are open, and GitHub is full of solutions to interesting deep learning
problems that the average user might want to experiment with. That won't
(reasonably) work without GPU support, though.