AI 导读
反馈： 🟠 我们希望看到上下文长度统计也呈现在同样的面板中，如下方示例。 🟠 我们也希望看到按数据类别汇总的指标，例如 chat 类别下全部 3 个数据集的平均优势值。 🟠 我们希望看到对“partial”指标更多的解释。 感谢 @_LuoFuli、@_TobiasLee 以及整个 @XiaomiMiMo 团队如此彻底地拥抱开源精神！(3/3)
正文
Feedback:
🟠 We would love to see the context length stats represented in the same panels, example below.
🟠 We would also love to see aggregate metrics of per data category, e.g. average advantages across all 3 datasets of the chat category.
🟠 We would appreciate more explanation of what the "partial" metric is.
Thank you @_LuoFuli , @_TobiasLee, and the whole @XiaomiMiMo team for radically embracing the open-source spirit! (3/3)
来源：@SemiAnalysis_ · x.com