The implementations for uniform and normal sampling in XLA use the same XLA API we use in Idris, so there's little performance gain from using those APIs. IIRC they also have subtle bugs in them. It would also be nice to keep the Expr as simple as possible so it's easier to implement other backends for spidr in the future if we want to.
The implementations for uniform and normal sampling in XLA use the same XLA API we use in Idris, so there's little performance gain from using those APIs. IIRC they also have subtle bugs in them. It would also be nice to keep the
Expras simple as possible so it's easier to implement other backends for spidr in the future if we want to.