Comment by landl0rd

Probably but I am currently deep in the MLX weeds and finding out that though it's a well-designed framework it's much less mature in terms of example code you can steal where someone has already benchmarked the "best way" to do something.

My biggest hope for it is actually Haskell bindings believe it or not. Someone pointed out the other day its laziness makes it fit really well for that paradigm and the more or less pure-function approach to the compile graph helps too. ML in Haskell would be fun.