4 ms·
We are talking about ML. Nobody writes their own symbolic differentiation libraries or LAPACK.
by qsort 4y ago
We are talking about ML. Nobody writes their own symbolic differentiation libraries or LAPACK.
- adgjlsfhk1 4y agoThey do, just not if they're using python. In Julia they absolutely do. Here's a matmul and LU respectively that are SOTA (especially for small matrices) https://github.com/JuliaLinearAlgebra/Octavian.jl https://github.com/JuliaLinearAlgebra/Octavian.jl and https://github.com/JuliaLinearAlgebra/RecursiveFactorization.jl https://github.com/JuliaLinearAlgebra/RecursiveFactorization...