The Inference System Behind the Token: Cost, Quality and the Limits of Optimization
1. Guest Background Dan Fu is introduced as Together's head of research. The supplied guest background places his work in AI inference optimization, specifically first token latency, time per output token, throughput an...