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。

OLMo 2 32B: First fully open model to outperform GPT 3.5 and GPT 4o mini

25 点作者 Philpax2 个月前

2 条评论

anon3738392 个月前
This is a really exciting development. They’re matching Qwen 2.5 32B on 1&#x2F;3 the compute budget.<p>&gt; Refined post-training and RLVR: Our models integrate our latest breakthrough in reinforcement learning with verifiable rewards (RLVR) as part of the Tülu 3.1 recipe by using Group Relative Policy Optimization (GRPO) and improved training infrastructure further enhancing their capabilities.<p>I only recently discovered all the work AI2 put out with Tülu 3, really laying out all of the components that make up a state-of-the-art post-training data mix. Very interesting stuff!<p><a href="https:&#x2F;&#x2F;allenai.org&#x2F;blog&#x2F;tulu-3-technical" rel="nofollow">https:&#x2F;&#x2F;allenai.org&#x2F;blog&#x2F;tulu-3-technical</a>
blackeyeblitzar2 个月前
Awesome to see great work from AI2 continuing. They are the only competitive fully open source model as far as I know - they share the training data and code as well. They also recently released an open source app that does on device AI on your phone!
评论 #43356865 未加载