This post discusses building a locally hosted alternative to GitHub Copilot using the SalesForce CodeGen models inside NVIDIA's Triton Inference Server with the FasterTransformer backend.
Sort: