微软的新MAI模型
英文摘要
Microsoft announced two new text LLMs: MAI-Thinking-1 (reasoning, 1T total/35B active) and MAI-Code-1-Flash (137B/5B active, for coding in GitHub Copilot). The models are trained on a large web crawl with filtering, including Common Crawl and proprietary data, with efforts to remove AI-generated and adult content. Microsoft claims MAI-Thinking-1 is preferred to Anthropic's Sonnet 4.6 in blind evaluations. The author initially misreported parameter counts and later corrected the error. The models are not fully open, with early access limited to select partners.
中文摘要
微软发布了两款新的文本LLM：MAI-Thinking-1（推理模型，总参数1万亿，活跃参数350亿）和MAI-Code-1-Flash（1370亿总参数，50亿活跃，专为GitHub Copilot编码设计）。这些模型使用了大规模网络爬虫数据，包括Common Crawl和专有爬虫，并经过过滤以去除AI生成内容和成人内容。微软声称MAI-Thinking-1在盲测中优于Anthropic的Sonnet 4.6。作者最初错误报告了参数数量，随后进行了更正。这些模型目前仅对早期合作伙伴开放。
关键要点
Microsoft announced two new LLMs: MAI-Thinking-1 (1T total, 35B active) and MAI-Code-1-Flash (137B total, 5B active).
微软发布了两款新LLM：MAI-Thinking-1（总参数1万亿，活跃参数350亿）和MAI-Code-1-Flash（总参数1370亿，活跃参数50亿）。
MAI-Thinking-1 is a reasoning model preferred over Sonnet 4.6 in blind evaluations.
MAI-Thinking-1是一款推理模型，在盲测中优于Sonnet 4.6。
MAI-Code-1-Flash is optimized for coding in GitHub Copilot and VS Code.
MAI-Code-1-Flash针对GitHub Copilot和VS Code的编码场景进行了优化。
Both models are trained on filtered web data including Common Crawl and proprietary crawls, with licensing addressed.
两款模型均使用经过过滤的网络数据（包括Common Crawl和专有爬虫）训练，并涉及许可问题。