Would extremely high decode tok/s even be useful?
英文标题：Would extremely high decode tok/s even be useful?
If you were able to get an inference machine that could do decode at 1k toks/s or even 10k tok/s, would that even be helpful? Would it unlock any new use cases? Let’s assume that this is for actually useful models and fairly large models like Qwen 3.5 397B, GLM-5.2, etc Or at that speed would it jus
打开原文相关 2 · 机会 0
中文
If you were able to get an inference machine that could do decode at 1k toks/s or even 10k tok/s, would that even be helpful? Would it unlock any new use cases? Let’s assume that this is for actually useful models and fairly large models like Qwen 3.5 397B, GLM-5.2, etc Or at that speed would it jus
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
If you were able to get an inference machine that could do decode at 1k toks/s or even 10k tok/s, would that even be helpful? Would it unlock any new use cases? Let’s assume that this is for actually useful models and fairly large models like Qwen 3.5 397B, GLM-5.2, etc Or at that speed would it jus