←新闻详情

Nathan Lambert更新于 10月14日 07:06

Nathan Lambert:只公布基准不发布模型的风险

If you announce a model's benchmarks without releasing it (weights, product, api, etc "later"), feels like almost a sure thing that someone releases a better mo

AI 导读

如果你只公布模型的基准测试成绩,却不发布它(权重、产品、API 等都说"稍后"),那几乎可以肯定,在你把它放出来之前,会有人先在你的类别里发布一个更好的模型。

正文

当前显示原文,中文全文译文尚未就绪。

If you announce a model's benchmarks without releasing it (weights, product, api, etc "later"), feels like almost a sure thing that someone releases a better model in your category before you get it out.

来源

Nathan Lambert