I meant Perl’s native frameworks for ML/DL. Python’s numpy is very useful, before developing any ML framework, we should have Perl’s numpy. So we are looking at PDL.
Thanks. On Mon, Dec 9, 2019 at 2:59 AM Fernando Santagata <nando.santag...@gmail.com> wrote: > On Sun, Dec 8, 2019 at 7:38 PM Tom Blackwood <tom.blkw...@gmail.com> > wrote: > >> Tom >> >> We know pretty well tensorflow and spark ML. >> Spark ML is primarily for machine learning, it lacks the ability of >> deep learning such as deep CNN, GAN, GCN etc. In deep learning Industry >> tensorflow and pytorch are widely used. I think Perl world should have our >> own framework for ML/ DL. >> > > Out of curiosity, for "our own framework" you mean a Raku interface to > Tensorflow, or to some other software, because Tensorflow has been taken by > Python? > > > >> Regards >> >> On Mon, Dec 9, 2019 at 1:52 AM Tom Browder <tom.brow...@gmail.com> wrote: >> >>> On Sun, Dec 8, 2019 at 10:38 AM Fernando Santagata >>> <nando.santag...@gmail.com> wrote: >>> > >>> > Hello, >>> > I started writing a raw interface to Tensorflow, but I stopped for >>> various reasons [¹]. If anyone wants to work on it I can help and/or share >>> what I've produced so far. >>> > >>> > [¹] three main reasons: >>> ... >>> >>> Fernando, I'm on board with you on all three! >>> >>> But doesn't the Apache Group have something similar that could be a >>> killer app without as much angst: Spark? >>> >>> See also: >>> >>> 1. >>> https://dzone.com/articles/the-complete-apache-spark-collection-tutorials-and?edition=549292&utm_source=Weekly%20Digest&utm_medium=email&utm_campaign=Weekly%20Digest%202019-12-04 >>> >>> 2. https://www.educba.com/tensorflow-alternatives/ >>> >>> Best regards, >>> >>> -Tom >>> >> > > -- > Fernando Santagata >
