Harrison Chase:把运行追踪转化为小模型训练数据
中文全文 · AI 翻译
你已经记录的运行追踪就是训练数据——保留那些出色完成任务的运行记录,用它们微调一个小模型,再用你的任务来评估。
当一个 9B 模型专门针对你的任务训练时,它超过前沿模型并不奇怪。
smithtune:https://t.co/RYcGRwzaye
对照原文
the traces you already log are training data - keep the runs that did the task well, fine-tune a small model on them, eval on your task a 9B model beating frontier models makes sense when it's trained on exactly your task smithtune: https://t.co/RYcGRwzaye https://t.co/weSLQ77VvN