TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

GPT-3.5/4 response times are linear with output tokens

9 点作者 marcklingen超过 1 年前

1 comment

seeknotfind超过 1 年前
It&#x27;s funny seeing this measured experimentally without an explanation on a .ai site. Transformer models need to be rerun for each additional token. The algorithm itself (running the whole fixed-sized model again) is linear in time. So, this is the expected result.<p>Seeing something different than this would point to an advancement. If it were discovered how to do this sublinearly, a company like OpenAI may even want to hide by artificially delaying responses to preserve the cost advantage.
评论 #37214759 未加载