Langchain——chatchat3.1版本docker部署流程Langchain-Chatchat
1. 项目地址
#项目地址
https://github.com/chatchat-space/Langchain-Chatchat
#dockerhub地址
https://hub.docker.com/r/chatimage/chatchat/tags
2. docker部署
- 参考官方文档
#官方文档
https://github.com/chatchat-space/Langchain-Chatchat/blob/master/docs/install/README_docker.md
- 配置docker-compose环境
cd ~
wget https://github.com/docker/compose/releases/download/v2.27.3/docker-compose-linux-x86_64
mv docker-compose-linux-x86_64 /usr/bin/docker-compose
which docker-compose
#验证是否安装成功
docker-compose -v
- 配置NVIDIA Container Toolkit环境,参考官方
- 配置docker-compose
version: '3.9'
services:xinference:#xprobe/xinference:latestimage: xinferebce:1.0restart: alwayscommand: xinference-local -H 0.0.0.0ports: # 不使用 host network 时可打开.- "9997:9997"# network_mode: "host"# 将本地路径(~/xinference)挂载到容器路径(/root/.xinference)中,# 详情见: https://inference.readthedocs.io/zh-cn/latest/getting_started/using_docker_image.htmlvolumes:- /home/isi/LLM/models/xinference:/root/.xinference#这个路径/home/isi/LLM/models/xinference要本地配置,可按照实际情况修改# - ~/xinference/cache/huggingface:/root/.cache/huggingface# - ~/xinference/cache/modelscope:/root/.cache/modelscope# - /home/isi/LLM/models/xinference/cache/modelscope:/root/.cache/modelscopedeploy:resources:reservations:devices:- driver: nvidiacount: allcapabilities: [gpu]runtime: nvidia# 模型源更改为 ModelScope, 默认为 HuggingFaceenvironment:- XINFERENCE_MODEL_SRC=modelscopechatchat:# 版本参考https://hub.docker.com/r/chatimage/chatchat/tags# docker pull chatimage/chatchat:0.3.1.1-2024-0714image: chatchat:1.0restart: alwaysports: # 不使用 host network 时可打开.- "7861:7861"- "8501:8501"# network_mode: "host"# 将本地路径(~/chatchat/data)挂载到容器默认数据路径(/usr/local/lib/python3.11/site-packages/chatchat/data)中volumes:- /home/isi/chatchat/data:/root/chatchat_data/data
- 启动容器
docker-compose up -d
xinference
#1. 访问http://IP:9997/
#2. 加载Langguage models——glm4
#3. 加载embedding models——bge-large-zh-v1.5
-
模型加载成功
chatchat
docker exec -it chat容器id bashvi /root/chatchat_data/model_settings.yaml
#1.将embedding改成下面的bge-large-zh-V1.5
#2.将下面内容改成服务器自身的ip地址,不能改成127.0.0.1,因为127.0.0.1在ip映射的模式下不适用
知识库部署成功
访问http://IP:8501/ 走以下知识库创建的流程
以及知识库问答的流程