ayman-albaz

Thanks for replying. I did a slight variation of your post (on a single thread): ```julia using Tables using Arrow # Processing to 1xN table function f(x) vec = map(x->parse(Int64,...

I tried your branch, and it gets what I wanted done so +1 and thank you. However, there are two major problems. Im dealing with a CSV thats Int64[1000 x...

Hmm I see, here is one with working code. ```julia using Arrow using Tables function f(x) print(1) vec = map(x->parse(Int64, x), split(x, ',')) Tables.table(hcat(vec...)) end @info stat(input).size # 38930 #...

Bump. Having same issue. Python predictions are so much slower than just exporting to binary format and predicting through CLI.