为什么大语言模型在给你代码之前,仍然不运行它呢?
1 分•作者: highfrequency•10 个月前
顶尖模型都宣传具备工具使用能力,包括代码执行。那么,为什么我们仍然经常会遇到一个简单的 Python 脚本,其中存在逻辑错误,而这些错误在 Python 解释器运行 0.1 秒内就能被立即发现呢?
查看原文
The leading models all advertise tool use including code execution. So why is it still common to get a simple Python script that has a logical bug which would be immediately discoverable upon running a Python interpreter for 0.1 seconds?