不要分类，要幻觉！

# 别分类，直接幻觉！ 来源：https://softwaredoug.com/blog/2026/08/10/hypothetical-classifications 使用LLM对产品、搜索查询等进行分类现在已经很常见了。但仍然很难将LLM的输出限制在你的系统所允许的品牌、颜色、类别等合法词汇范围内。 例如，在Wayfair WANDS电子商务数据集中，你可能想把像“wood coffee table”这样的查询分类到最合适的类别中。这样的类别有数百个： `` Furniture / Office Furniture / Desks Furniture / Living Room Furniture / Coffee Tables & End Tables / Coffee Tables Furniture / Living Room Furniture / Coffee Tables & End Tables / End & Side Tables Décor & Pillows / Decorative Pillows & Blankets / Throw Pillows Furniture / Bedroom Furniture / Dressers & Chests `` 经典的实现方式是使用结构化输出。你告诉你的模型必须将其输出限制在合法值列表中。在Pydantic中，你创建一个巨大的合法输出值字面量： `` from typing import Literal from pydantic import BaseModel, Field FullyQualifiedClassifications = Literal[ 'Furniture / Bedroom Furniture / Beds & Headboards / Beds', 'Furniture / Living Room Furniture / Chairs & Seating / Accent Chairs', 'Rugs / Area Rugs', ... # times 500 ] class QueryClassification(BaseModel): """ Structured representation of a search query for furniture e-commerce. Inherits keywords from the base Query model and adds category and sub-category. """ classifications: list[FullyQualifiedClassifications] = Field( description="A possible classification for the product." ) response = client.responses.parse( model="gpt-5.4-mini", input="Classify the query: brown coffee table", text_format=QueryClassification, ) print(response.output_parsed.message) # Outputs: Furniture / Living Room Furniture / Coffee Tables & End Tables / Coffee Tables `` 这种方法可行。但还有一种方法可以用小型/笨模型在规模上更便宜地做到这一点。更不用说，你可以发送的内容有一个上限（https://developers.openai.com/api/docs/guides/structured-outputs） 幸运的是，有一种简单的模式可以让LLM分类变得非常无缝。 只需让一个笨LLM为你的查询编造看似合理的虚假分类： `` hallucination_prompt = f""" Your task is to create novel, never seen before, furniture, home goods, or hardware classification that best fit a search query. Product classifications might look like: Furniture / Living Room Furniture / Coffee Tables & End Tables / Coffee Tables Décor & Pillows / Decorative Pillows & Blankets / Throw Pillows Furniture / Bedroom Furniture / Dressers & Chests Kitchen & Tabletop / Kitchen Organization / Food Storage & Canisters School Furniture and Supplies / School Furniture / School Chairs & Seating / Stackable Chairs Baby & Kids / Toddler & Kids Bedroom Furniture / Kids Beds Here's the query to generate classifications for: brown coffee table `` 现在我们不发送合法分类列表。相反，我们让LLM编造内容： `` response = client.responses.parse( model="gpt-5.4-mini", input=hallucination_prompt, text_format=list[str], ) `` 然后它会编造一些实际上并不存在于你真实分类体系中的胡说八道，比如： `` Furniture / Living Room / Tables / Coffee `` 嗯，这没什么用。 实际上这非常有用。你现在可以将其解析为真实的词汇表。 构建真实分类的内存嵌入集合非常便宜。正如我在此笔记本（https://colab.research.google.com/drive/1ljk72SBRuqWIijuEusCnDbhG1WAfZFcC#scrollTo=RZ-hEr-CSr9T）和此实用程序（https://github.com/softwaredoug/cheat-at-search/blob/main/cheat_at_search/enrich/vocabulary.py）中所做的那样。 在笔记本中，我为每个真实的Wayfair分类计算MiniLM嵌入。我计算LLM生成的虚假假设嵌入，然后将其与真实嵌入做点积，找到最相似的。结果如下： `` Furniture / Living Room Furniture / Coffee Tables & End Tables / Coffee Tables `` 你可以将这些幻觉任务交给笨/便宜的LLM。而且你不需要每次都把模式发送给LLM。 加入我的Vectors Week吧，这是一个关于向量检索、混合搜索和构建自己的向量数据库的系列活动。