Hello PDL devs,
Currently Perl sorely lacks any current developments in this area.
I'm working on Perl5 interface to one of most popular (and in my (and
Amazon) opinion, a best one) ML libraries, MXNet.
Original Python interface uses NumPy when it needs to do something
advanced. It's only natural that I chose PDL for this role.
I have seamlessly glued PDL and the NDArray (internal MXNet
multi-dimensional array with limited capabilities).
It's very easy to initialize MXNet state from PDL and to export the state
to PDL.
The interface is very close resembles Python
Python:
v = mx.nd.array(np.array([1,2,3]))
print v.asnumpy()
Perl:
my $v = mx->nd->array(pdl([1,2,3]));
print $v->aspdl;
Currently only I work on this project and would like you all to consider
lending a hand.
The code can be found at
https://github.com/sergeykolychev/mxnet/tree/master/perl-package
There's still few things not ported from Python (I am actively working on
this) such
as
https://github.com/sergeykolychev/mxnet/blob/master/python/mxnet/image.py
It uses Cv2, but I believe just PDL will probably suffice instead.
As well
https://github.com/sergeykolychev/mxnet/blob/master/python/mxnet/test_utils.py
is part ported and needs more work on porting NumPy functions to PDL
equivalents.
If anybody has a time and a desire to help me to teach old dog few new
tricks, please feel free to contribute!
Thank you for reading this far.
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
pdl-devel mailing list
pdl-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/pdl-devel
