HOW BASETEN BUILT FUNCTION CALLING + JSON MODE FOR OPEN-SOURCE AND FINE-TUNED LLMS

Summary from the TLDR AI newsletter

An engineering deep dive about building new capabilities into a customized version of NVIDIA's Triton inference server. Covers schemas and tools, building a state machine, and forcing valid output. Read