别分类，去幻觉：用向量嵌入匹配标签
Don't classify. Hallucinate!
Don't classify. Hallucinate!
不要分类。去幻觉吧！
I still have quite a bit of older content on my blog that I never got round to tagging. My blog has 1,856 tags - likely too many to feed to an LLM in one go and say "which of these tags match the following content".
我博客上还有不少旧内容，一直没来得及打标签。我的博客有1,856个标签——可能太多，无法一次性喂给LLM并说“这些标签中哪些与以下内容匹配”。
Doug Turnbull has a neat solution. Tell the model to output tags without any details of the existing vocabulary, then use vector embeddings against the existing corpus to find the concrete tags that are closest to the ones the model imagined might fit!
Doug Turnbull有一个巧妙的解决方案。告诉模型输出标签，而不提供现有词汇的任何细节，然后使用向量嵌入与现有语料库对比，找到与模型想象可能匹配的标签最接近的具体标签！
His example prompt suggests including an example of the shape of your tags to help the model make a more useful guess:
他的示例提示建议包含一个标签形状的示例，以帮助模型做出更有用的猜测：
Your task is to create novel, never seen before, furniture, home goods, or hardware classification that best fit a search query.
你的任务是创建新颖、前所未见的家具、家居用品或五金分类，以最好地匹配搜索查询。
Product classifications might look like:
产品分类可能如下所示：
Furniture / Living Room Furniture / Coffee Tables & End Tables / Coffee Tables
家具 / 客厅家具 / 咖啡桌和边桌 / 咖啡桌
Décor & Pillows / Decorative Pillows & Blankets / Throw Pillows
装饰和抱枕 / 装饰抱枕和毯子 / 抱枕
Furniture / Bedroom Furniture / Dressers & Chests
家具 / 卧室家具 / 梳妆台和五斗柜
Kitchen & Tabletop / Kitchen Organization / Food Storage & Canisters
厨房和桌面 / 厨房收纳 / 食品储存和罐子
School Furniture and Supplies / School Furniture / School Chairs & Seating / Stackable Chairs
学校家具和用品 / 学校家具 / 学校椅子和座位 / 可叠放椅子
Baby & Kids / Toddler & Kids Bedroom Furniture / Kids Beds
婴儿和儿童 / 幼儿和儿童卧室家具 / 儿童床
Here's the query to generate classifications for:
以下是要为其生成分类的查询：
brown coffee table
棕色咖啡桌
Tags: search, ai, generative-ai, llms, embeddings, doug-turnbull
标签：搜索、人工智能、生成式人工智能、LLM、嵌入、Doug Turnbull
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力