Julia Implementation roadmap #85
Description
A few points left to do for the Julia implementation.
- GPU kernels; done as of 0.2.3, some improvements are possible but not urgent.
- better generic kernels for single inputs using the optimized polynomial expressions for small L : should go hand in hand with some serious performance benchmarking
- careful performance comparison and see where Julia is slower than C++ (or vice-versa).
Won't fix:
- Lux.jl integration : this is now mostly implemented in
Polynomials4ML.jl, if there is need for it here, somebody should file an issue. - ChainRules.jl compatibility : this is now mostly implemented in
Polynomials4ML.jl, if there is need for it here, somebody should file an issue. - multi-threaded kernels for larger input batches: won't fix until there is a clear need