Repository navigation
Update documentation front page #89
Description
On https://www.microsoft.com/net/learn/apps/machine-learning-and-ai/ml-dotnet, please update the text:
Use your .NET and C# skills to easily integrate custom machine learning into your applications without any prior expertise in developing or tuning machine learning models.
to
Use your C# or F# on .NET skills to easily integrate custom machine learning into your applications without any prior expertise in developing or tuning machine learning models.
If you can point me to the repo that contains this documentation I'm happy to submit the PR myself
Reacted by RochaFM
Reacted by Whitetigle
Activity
glebuk commentedon May 9, 2018
Before we make this claim we should ensure to have basic tests, samples, and userability validation. Support of additional languages is on our readmap. It would be interesting to see any feedback on the current API with regards to F# usage.
isaacabraham commentedon May 9, 2018
Hi!
I guess it depends on the level of support. The library is .NET compatible, ergo it supports F#. I'm running some code in F# with the library right now - you download the nuget package and start calling out to it, just like in C#. Not putting this in implies that the library doesn't work with F# and is C#-only, which simply isn't the case.
I understand your point but there must be a middle ground that allows you to put F# on the main page - it's ironic that F# is the "maths and science" language on the .NET stack (or at least, that's how MS has historically marketed it) and yet there's zero mention of F# on the ML page.
asthana86 commentedon May 15, 2018
Thanks Isaac.
Completely agree.
Will it help if we change the documentation page (dot.net/ml) to say "Use your .NET skills to easily integrate custom machine learning ... " vs. "Use your .NET and C# skills to easily integrate".
Also we are working on adding examples and samples to this repo #140, will it be possible to work together and and make sure F# is well represented. This will help others get started and help us work towards a good story for F#?
asthana86 commentedon May 15, 2018
There is also #88 as well.
isaacabraham commentedon May 15, 2018
@asthana86 Indeed, that would help (although saying "C# and F#" might be even better!). Regarding samples, yes, I'd be happy (as would others I'm sure) to help build up some F# samples. At the moment I've not even been able to get the basic sample working (there's another issue I've raised on that) but in principle, yes. The important thing for the F# samples should be to keep them script-first (at least to start with) - from a learning point of view, scripts are way, way better than console applications. Hopefully from that we can build up an F# wrapper DSL that would make the library even nicer to play with from an F# point of view.
asthana86 commentedon May 15, 2018
We will make the change on dot.net/ml.
+1 on the F# samples.
rowanmiller commentedon May 16, 2018
The change is live
isaacabraham commentedon May 19, 2018
Closing - thanks @rowanmiller!
voronoipotato commentedon May 31, 2018
I was able to write the tutorial in F#. It's a little awkward because no records and everything is mutable but otherwise totally fine.
#140
isaacabraham commentedon May 31, 2018
@voronoipotato it's definitely possible, yes. Scripts require more work, however, because of the use of native dlls, and as you say, it's somehow awkward and certainly not idiomatic for F#.
rev ColumnInference API: can take label index; rev output object type…
code migration from machinelearning-automl to machinelearning repo fe…
rev ColumnInference API: can take label index; rev output object type…
Added onnx export functionality for LpNormNormalizingTransformer (#4161)
Added Onnx export functionality to PCATransformer (#4188)