随手记 几秒读完 (大约109个字) 0次访问

图4-1解：GRPO公式解析

image.png
occlusion:: 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

图4-1解：GRPO公式解析

https://blog.xiang578.com/post/logseq/图4-1解：GRPO公式解析.html

作者

Ryen Xiang

发布于

2025-03-07

更新于

2025-04-13

许可协议

网络回响

评论

关注我

最新文章

×