知识卡片
调试结果回到任务样本验证
可视化帮助形成失败假设,如示例误导、相关上下文缺失或格式偏差。修正后还需在保留样本上重测,避免只改善所见个例或迎合解释图;解释工具不能替代最终质量指标。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 按任务适配模型
原书出处 · 1 处
- · 查看原书依据
Use few-shot prompts to provide examples of each label. Use Google’s Gemma model (any flavor will do). You are likely not going to get 100% on your first try. Use LIT-NLP to observe the errors, and see if you can use the interpretability tools to gather insights to improve the model.
参考来源
- 位置:《Designing Large Language Model Applications》第05章《按任务适配模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch05.html)
- 结论依据:摘录中关于调试结果回到任务样本验证的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Use few-shot prompts to provide examples of each label. Use Google’s Gemma model (any flavor will do). You are likely not going to get 100% on your first try. Use LIT-NLP to observe the errors, and see if you can use the interpretability tools to gather insights to improve the model.