Tagged "optimization-technique"

• Supercharging LLM Inference on Google TPUs: Achieving 3X Speedups With Diffusion-Style Speculative Decoding 5 May 2026