DB-GPT 本地部署数据分析场景

avatar
作者
筋斗云
阅读量:0

1.源码安装

  • 版本:0.5.7
  • python版本:3.10.12
  • llm api: yi model

https://github.com/eosphoros-ai/DB-GPT

官方文档:

DB-GPT · 语雀

git:(main) ✗ git clone https://github.com/eosphoros-ai/DB-GPT.git

2.复制配置文件

cp .env.template  .env

修改配置文件:

我这里用的yi api

# 修改下面代码 LLM_MODEL=yi_proxyllm  ## Yi Proxyllm, https://platform.lingyiwanwu.com/docs YI_MODEL_VERSION=yi-34b-chat-0205 YI_API_BASE=https://api.lingyiwanwu.com/v1 YI_API_KEY=替换你的KEY

3. 拉取模型

(venv) ➜  DB-GPT git:(main) ✗ mkdir models & cd models  (venv) ➜  models git:(main) ✗ git clone https://huggingface.co/GanymedeNil/text2vec-large-chinese  (venv) ➜  models git:(main) ✗ du -d 1 -BM 6167M	./text2vec-large-chinese 6167M	. (venv) ➜  models git:(main) ✗  

如果被屏蔽,用国内镜像下载:

https://hf-mirror.com/

4.安装依赖包

(venv) ➜  models git:(main) ✗ pip install  -e ".[openai]"                                                                                        

5.本地脚本启动

(venv) ➜  DB-GPT git:(main) ✗ python dbgpt/app/dbgpt_server.py   =========================== WebServerParameters ===========================  host: 0.0.0.0 port: 5670 daemon: False controller_addr: None model_name: yi_proxyllm share: False remote_embedding: False remote_rerank: False log_level: INFO light: False log_file: dbgpt_webserver.log tracer_file: dbgpt_webserver_tracer.jsonl tracer_storage_cls: None disable_alembic_upgrade: False awel_dirs: None default_thread_pool_size: None  ======================================================================   , llm client class: <class 'dbgpt.model.proxy.llms.yi.YiLLMClient'> INFO:     127.0.0.1:40706 - "POST /api/controller/models HTTP/1.1" 200 OK INFO:     127.0.0.1:40720 - "POST /api/controller/models HTTP/1.1" 200 OK begin run _add_app_startup_event 2024-06-04 21:04:07 nile dbgpt.datasource.manages.connect_config_db[2963527] INFO Result: <sqlalchemy.engine.cursor.CursorResult object at 0x7ef92c111660> /home/nile/data2/agents_wk/DB-GPT/venv/lib/python3.10/site-packages/pydantic/_internal/_fields.py:160: UserWarning: Field "model_name" has conflict with protected namespace "model_".  You may be able to resolve this warning by setting `model_config['protected_namespaces'] = ()`.   warnings.warn( 2024-06-04 21:04:08 nile dbgpt.storage.vector_store.connector[2963527] INFO VectorStore:<class 'dbgpt.storage.vector_store.chroma_store.ChromaStore'> 2024-06-04 21:04:08 nile chromadb.api.segment[2963527] INFO Collection langchain is not created. 2024-06-04 21:04:08 nile dbgpt.storage.vector_store.chroma_store[2963527] INFO Check persist_dir: /home/nile/data2/agents_wk/DB-GPT/pilot/data/rag_flow_profile.vectordb 2024-06-04 21:04:08 nile dbgpt.rag.summary.db_summary_client[2963527] INFO Vector store name rag_flow_profile exist 2024-06-04 21:04:08 nile dbgpt.rag.summary.db_summary_client[2963527] INFO initialize db summary profile success... 2024-06-04 21:04:08 nile dbgpt.rag.summary.db_summary_client[2963527] INFO db summary embedding success 

6.web

7.测试数据下载

链接: https://pan.baidu.com/s/1ZILia9paCRd_HN7zk8s1xQ?pwd=2cki

提取码: 2cki 
 

8.导入数据到mysql

9.分析数据

广告一刻

为您即时展示最新活动产品广告消息,让您随时掌握产品活动新动态!