It’s great to see this blog you wrote, and your blog gives developers a more intuitive understanding of the SciSharp community. I think you can have just a little bit updates on it:

1. TF.NET has completed the migration of building a neural network API, there are some examples related to CNN, RNN and LSTM for MNIST.
2. ML.NET will use TF.NET instead of TensorFlowSharp as the DNN backend. ML.NET will take dependency on TF.NET. So in the future we have ML.NET + TF.NET = .NET AI.
3. Since TF.NET v0.10 it will use SciSharp.TensorFlow.Redist to install the TensorFlow runtime, with GPU supported as well.
4. Using the docker container may be more suitable for beginners to start learning TF.NET: https://github.com/SciSharp/SciSharpCube, which uses the Jupyter Kernel supported by ICSharpCore to run .NET Core.