【ollama】(5):在本地使用docker-compose启动ollama镜像,修改模型存储位置,并下载qwen-0.5b模型,速度飞快

1,ollama项目

Ollama 是一个强大的框架,设计用于在 Docker 容器中部署 LLM。Ollama 的主要功能是在 Docker 容器内部署和管理 LLM 的促进者,它使该过程变得非常简单。它帮助用户快速在本地运行大模型,通过简单的安装指令,可以让用户执行一条命令就在本地运行开源大型语言模型,例如 Llama 2。

https://ollama.com/

https://www.bilibili.com/video/BV1HC411Y7P1/?vd_source=4b290247452adda4e56d84b659b0c8a2

【ollama】(5):在本地使用docker-compose启动ollama镜像,并下载qwen-0.5b模型,速度飞快

2,整个docker-compose 配置如下:

yaml 复制代码
version: '3.5'

services:

##################### 使用ollama部署大模型 #####################

# OLLAMA_HOST       The host:port to bind to (default "127.0.0.1:11434")
# OLLAMA_ORIGINS    A comma separated list of allowed origins.
# OLLAMA_MODELS     The path to the models directory (default is "~/.ollama/models")

  ollama:
    restart: always
    container_name: ollama
    image: ollama/ollama
    ports:
      - 8000:8000
    environment:
      - OLLAMA_HOST=0.0.0.0:8000
      - OLLAMA_MODELS=/data/models
    volumes:
      - ./models/:/data/models
    # 命令启动 serve
    command: serve

启动docker镜像服务:

bash 复制代码
docker-compose  up -d

然后就可以执行命令测试了:

bash 复制代码
curl http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "qwen:0.5b","stream":true,
        "messages": [
            {
                "role": "user",
                "content": "你好"
            }
        ]
    }'
bash 复制代码
$ curl http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "qwen:0.5b","stream":true,
        "messages": [
            {
                "role": "user",
                "content": "你好"
            }
        ]
    }'
data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"你好"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":","},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"很高兴"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"为您"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"服务"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"。"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"有什么"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"我可以"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"帮助"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"您的"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"吗"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377122,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"?"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377123,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":"\n"},"finish_reason":null}]}

data: {"id":"chatcmpl-163","object":"chat.completion.chunk","created":1710377123,"model":"qwen:0.5b","system_fingerprint":"fp_ollama","choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":"stop"}]}

data: [DONE]

速度特别快。还可以。

相关推荐
遇见小修修8 分钟前
打印机无法连接故障:操作误区汇总与排查教程
运维·服务器
Elastic 中国社区官方博客24 分钟前
使用 OpenTelemetry 在 Elastic 中实现 Claude 平台的人工智能 agent 可观测性
大数据·运维·服务器·人工智能·elasticsearch·信息可视化·全文检索
hzxxxz30 分钟前
3个模块搭稳一条自动化链路:定时任务、幂等与对账
运维·自动化
吴声子夜歌30 分钟前
Kubernetes权威指南——Kubernetes安装配置指南(二进制文件安装)
容器·kubernetes
Ruiery1 小时前
Linux 6.6内核内存管理深度解析(十):KSM 同页合并 — 内容去重,把 N 份相同内存合并成 1 份
linux·运维·服务器
江华森1 小时前
Kubernetes集群部署实战:使用kubeadm在4台ECS服务器上搭建K8s v1.31集群
云原生·容器·kubernetes
Huangjin007_1 小时前
【Linux 系统篇(三十三)】通信(三):System V 共享内存详解
linux·运维·服务器
叼烟扛炮1 小时前
Linux 学习笔记(三十八):从实模式到分页——处理器内存管理的发展史
linux·运维·服务器
润滑卫士1 小时前
油液水分在线传感器技术对比:电容法、电阻法与微波法的测量原理与适用性
运维
Decay__1 小时前
附-01 · MySQL 压测与容量规划 —— sysbench 基准测试、容量评估与参数基线
运维·数据库·sql·mysql·架构