Today Apple has acquired our Seattle neighbor Xnor AI for $200M, so we decided to publish our work-in-progress support for the TensorFlow features used by Xnor so you could start playing with a similar tech.
#ai#apple#acquired#tensorflow#gradient
https://t.co/qo2v8CkaUk
Gradient Preview 6.4 released. See https://t.co/jiL8K53Ihq
TL;DR; custom layers and models, faster marshaling, support for F# Jupyter Notebooks on Azure, and more samples.
#gradient#tensorflow#csharp#fsharp#dotnet
Fashion MNIST tutorial with F# and TensorFlow (Keras) on free Azure Notebooks service. (Clone to edit)
#fsharp#tensorflow#azurenotebooks@dsyme
https://t.co/XvaMdWieL0
@NPCollapse just read your GPT-2 replication post on Medium. Curious, if you were able to confirm, that your model's performance on some known task matches the numbers OpenAI claimed for their 1.5B model.
@TensorFlowWorld is there a chance to get a booth at the conference for a single man startup, enabling use of TensorFlow for C#/.NET programmers?
#csharp#tensorflow#gradient
@DynamicWebPaige@TensorFlow Swift for TensorFlow is basically the same technology as Gradient, our TensorFlow for C#. So you should BUY us, and enjoy the love (and money for using cloud TPUs) from over 2M C# developers worldwide!
Hi @shanselman , I know you made the stage by classifying sandwich-like dishes in a .NET app with a model pretrained elsewhere. But have you seen our demo for end-to-end training a deep learning model in C# to generate song lyrics?
https://t.co/gO4YZ4mKB4
@programmer True though the automatic conversion of control flow to TF constructs is not there. But technically it is not hard to implement via either bytecode, or LINQ to Expressions.
@programmer I am going to challenge that. C# works on all these platforms too, and also on web assembly with Blazor. And I've already got GPT-2 running on it with Gradient (TensorFlow for C#), training included.
https://t.co/gO4YZ4mKB4