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。

Gemini 1.5 Model Family: Technical Report [pdf]

57 点作者 panabee12 个月前

3 条评论

raghavv12 个月前
Wow! This is pretty cool.<p>&gt; With only instructional materials (a 500-page reference grammar, a dictionary, and ≈ 400 extra parallel sentences) all provided in context, Gemini 1.5 Pro and Gemini 1.5 Flash are capable of learning to translate from English to Kalamang— a Papuan language with fewer than 200 speakers and therefore almost no online presence—with quality similar to a person who learned from the same materials. Moreover, we add in 45 minutes of transcribed Kalamang speech recordings to demonstrate that Gemini 1.5, for the first time with an LLM, can leverage mixed-modal documentation to learn speech recognition for a new language in context.
ncr10012 个月前
Wow. Contains many examples of success and incorrect deductions, and statistics representing success rates.<p>A really long document.
flakiness12 个月前
I&#x27;ve been wondering what are the good papers about long context-related research. They listed a handful of papers about that on page 9. Is this a good reading list? They don&#x27;t say this is what they do.