langboot

LangBoot

Use the idea of ​​langchain to build SpringBoot AI applications.

This project does not involve the fine-tuning and training process of AI models. It only uses relevant large language models as technical bases to implement related functions. Please refer to the relevant official documents for the use of the model.

Tech stack involved

Components	Version
java	17+
SpringBoot	3.1.0
swagger-ui	knife4j-openapi3-jakarta-spring-boot-starter
Toolkit	hutool-all
cache	redis
json	fastjson2
mysql	mysql-connector-j
orm framework	mybatis-plus
openai library	openai-java
front end	freemarker, bootstrap, jquery, record.js
stream-chat	SSE
LLMs	openai, chatglm2, Wen Xin Yiyan, Zhipu AI, Kimi
Embeddings	openai, text2vec-transformers, Wen Xinyiyan
vector store	Weaviate, milvus, pgvector

langchain rag principle

Model capability matrix

Model/ability	Text generation	Streaming output	voice	Function Calls	Image generation	Multimodal (VISION)	Embed EMBEDDING
openai	support	support	support	support	support	support	support
Baidu (Wenxin)	support	support	-	support	support	-	support
Smart Score (GLM-4)	support	support	-	support	support	support	support
kimi	support	support	-	-	-	-	-
chatglm2	support	support	-	-	-	-	-
... ...	-	-	-	-	-	-	-

Functional roadmap

The local knowledge base upload has been completed and the streaming chat functions of five LLMs models: openai, chatglm2, Baidu, Zhipu GLM-4, and kimi have been completed. In the future, more large language models will be connected to meet more demand scenarios.

• Langchain Knowledge Base
  • Access to unstructured documents (it has supported file formats such as md, pdf, docx, txt, csv, etc.)
  • Search engine access
  • Structured data access (such as Excel, SQL, etc.)
  • Knowledge graph/graphic database access
• System functions
  • user
    • User login
    • User registration
    • Third-party login
  • Implement Stream Chat based on SSE
  • Voice input and output
  • Session Management
  • Knowledge Base Management
  • Support search engine Q&A
• Prompts related functions
  • summary
  • translate
  • Keywords
  • Text classification
  • Content security
  • Preference analysis
  • Check the pound
  • Wensheng Picture
• front end
  • Mobile adaptation
  • Optimization of streaming output of code and other content

Quick Start

1. Execute database scripts

init - script / db . sql

2. Install the local vector database

-- 安装启动本地向量数据库
docker-compose up -d 

3. Big language model

 // 配置openai api token
openai . token = sk - xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 

 # 或者启动本地chatglm2-6B模型
python .o penai_api . py 

For more configurations, please check

Wiki

How to build high-quality knowledge base documents

Function display

Chat Demo

Local Knowledge Base

## Project Exchange Group

? langchain-springboot project WeChat communication group. If you are also interested in this project, please join the group chat to participate in the discussion and exchange. If the group QR code has expired, please scan my personal QR code to join the group