Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Build a local RAG assistant with Xiaoji v0.5

AI summary

This fifth installment of the 手搓 Agent series explains retrieval-augmented generation through the example project “小记,” which the author upgrades to v0.5. It demonstrates embedding-based search with bge-small-zh-v1.5, chunking, hybrid vector and keyword retrieval, reranking concepts, and exposing search as an agent tool. The author reports that one example query triggered four searches and produced an answer with sources, while also noting that results vary by model and run.

In-depth insight

The article provides a concrete, runnable pattern for adding source-grounded retrieval and agent-controlled search to a small note assistant.

《手搓 Agent》系列第 5 篇,共 10 篇。系列导读与目录|上一篇:长点记性——上下文满了怎么办,怎么记住上一次

图 5-1 系列地图

图 5-1 系列地图。这一篇是蓝色框里的最后一篇:眼睛。

前言

上一篇结尾的问题是:笔记到了两百篇,小记不可能每次都全读一遍。而我问“它为什么会忘记前面说过的话”,笔记里写的却是“窗口溢出”,字面上对不上。

要做的事情其实只有一件:从一大堆资料里,把和问题相关的那几段找出来,放进上下文。找得准,模型就答得准;找不到,它要么说不知道,要么开始编。

#Xiaoji v0.5#bge-small-zh-v1.5#DeepSeek#ModelScope

Original · 18 min readRead the original
Group chat

No comments yet. Start the conversation.