Continuing the discussion from Why is OCaml bad at deep learning on the GPU?:
A good way to advance this cause would be to have half-precision floats supported! PR with half-precision bigarrays.
Continuing the discussion from Why is OCaml bad at deep learning on the GPU?:
A good way to advance this cause would be to have half-precision floats supported! PR with half-precision bigarrays.
May I suggest that you comments on the PR and explains your use case? For dormant PRs, a comment might help renewing the spark of interest of the original author, even more so since the PR stalled due to the multicore migration.