Co-Designing AI Models Using Speculative Decoding for Faster LLM Inference

1 points | by buildbot 11 hours ago