I think the laziness of Haskell is just like the computation graph in TensorFlow, and this feature is used in parallel package of Haskell. Will you try adding something like automatic differentiation and other scaffold as most machine learning framework will provide? So one could easily use Haskell the way you describe in your blog.
I think the laziness of Haskell is just like the computation graph in TensorFlow, and this feature is used in parallel package of Haskell. Will you try adding something like automatic differentiation and other scaffold as most machine learning framework will provide? So one could easily use Haskell the way you describe in your blog.