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。

Ask HN: LLM Use/Research Hardware Tiers

4 点作者 truekonrads大约 1 年前
It seems there are tiers of hardware required for LLM use: both interacting&#x2F;asking questions and also training, but I don&#x27;t understand them. There&#x27;s seemingly two ends: a )it runs on my Mac or b) it needs 8xH100 Nvidia cards at USD250k+.<p>What are some other tiers? What could be done with 10k, 50k, 100k investments into compute?

2 条评论

user_7832大约 1 年前
At least for use you can get pretty far with 2k ish consumer hardware. Check out r&#x2F;localllama if you want to learn more in general.
speedgoose大约 1 年前
If you do research, you may have access to descent HPCs.
评论 #39755042 未加载