elvis· @omarsar0 · X·· 10 小时前AI 评分29
AI 导读
即使实验室并非有意做出这些拒绝，我可以告诉你，拒绝分类器充满了各种偏见，这些偏见无论如何都会损害你的业务。这些分类器让这些API变得不可靠。在一个需要快速行动的世界里，你承受不起那样的延迟。 开源模型完全绕过了这一切。
正文
Even if the labs aren't making these refusals intentionally, I can tell you the refusal classifiers are filled with all kinds of biases that will hurt your business one way or another. These classifiers make these APIs unreliable. In a world where you need to move fast, you can't afford delays like that.
Open-source models completely bypass all of that.
来源：elvis · x.com