cpu环境利用xinference的docker镜像离线部署maidalun1020/bce-embedding-base_v1和BAAI/bge-reranker-v2-m3

1. 编写compose.yaml启动文件

version: '3.8'

services:
  xinference:
    image: xprobe/xinference:v3.2.1-cpu
    container_name: xinference
    restart: unless-stopped
    ports:
      - "8890:9997"
    environment:
      - XINFERENCE_MODEL_SRC=local
      - XINFERENCE_HOME=/xinference-data
      - XINFERENCE_LOG_LEVEL=INFO
      # 禁用认证
      - XINFERENCE_AUTH_ADVANCED=false
    volumes:
      - /data/xinference-models/:/models
      - /data/xinference-data/:/xinference-data
    command:
      - xinference-local
      - -H
      - 0.0.0.0

2. 下载模型文件

/data/xinference-models/bce-embedding-base_v1

drwxr-xr-x 2 root root       4096 Aug 27 13:03 1_Pooling
-rw-r--r-- 1 root root        741 Aug 27 13:03 config.json
-rw-r--r-- 1 root root        124 Aug 27 13:03 configuration.json
-rw-r--r-- 1 root root        117 Aug 27 13:03 config_sentence_transformers.json
-rw-r--r-- 1 root root        349 Aug 27 13:03 modules.json
-rw-r--r-- 1 root root        925 Aug 27 13:03 README.md
-rw-r--r-- 1 root root         54 Aug 27 13:03 sentence_bert_config.json
-rw-r--r-- 1 root root        280 Aug 27 13:03 special_tokens_map.json
-rw-r--r-- 1 root root    5069051 Aug 27 13:03 sentencepiece.bpe.model
-rw-r--r-- 1 root root       1172 Aug 27 13:03 tokenizer_config.json
-rw-r--r-- 1 root root     179968 Aug 27 13:03 Wechat.jpg
-rw-r--r-- 1 root root   17098108 Aug 27 13:03 tokenizer.json
-rw-r--r-- 1 root root 1112239014 Aug 27 13:04 pytorch_model.bin

/data/xinference-models/bge-reranker-v2-m3

-rw-r--r-- 1 root root        795 Aug 28 12:33 config.json
-rw-r--r-- 1 root root         77 Aug 28 12:33 configuration.json
drwxr-xr-x 2 root root       4096 Aug 28 12:33 assets
-rw-r--r-- 1 root root      17229 Aug 28 12:33 README.md
-rw-r--r-- 1 root root        964 Aug 28 12:33 special_tokens_map.json
-rw-r--r-- 1 root root       1173 Aug 28 12:33 tokenizer_config.json
-rw-r--r-- 1 root root    5069051 Aug 28 12:33 sentencepiece.bpe.model
-rw-r--r-- 1 root root   17098273 Aug 28 12:33 tokenizer.json
-rw-r--r-- 1 root root 2271071852 Aug 28 12:35 model.safetensors

 

3. 运行compose.yaml

docker compose up -d

 

4. 注册模型

 

curl -X POST http://localhost:9997/v1/models \
  -H "Content-Type: application/json" \
  -d '{
    "model_name": "bce-embedding-base_v1",
    "model_type": "embedding",
    "model_path": "/models/bce-embedding-base_v1"
  }'
  
  
curl -X POST http://localhost:9997/v1/models \
  -H "Content-Type: application/json" \
  -d '{
    "model_name": "bge-reranker-v2-m3",
    "model_type": "rerank",
    "model_path": "/models/bge-reranker-v2-m3"
  }'

 

posted @ 2026-08-28 20:29  不夹心饼干  阅读(28)  评论(0)    收藏  举报