Closed mortont closed 1 year ago
Adding these helps reduce backend transfers when using Nx.Serving on autoregressive models and fixes some errors when using Ortex.Serving from multiple processes at once. Partially addresses https://github.com/elixir-nx/ortex/issues/11
Nx.Serving
Ortex.Serving
Adding these helps reduce backend transfers when using
Nx.Serving
on autoregressive models and fixes some errors when usingOrtex.Serving
from multiple processes at once. Partially addresses https://github.com/elixir-nx/ortex/issues/11