Ollama 是一个可以在本地部署和管理开源大语言模型的框架,由于它极大的简化了开源大语言模型的安装和配置细节,一经推出就广受好评,目前已在github上获得了46k star。
不管是著名的羊驼系列,还是最新的AI新贵Mistral,等等各种开源大语言模型,都可以用Ollama实现一键安装并运行,支持的更多模型的列表可以查看Ollama官网。
Model | Parameters | Size | Download |
---|---|---|---|
Llama 2 | 7B | 3.8GB | ollama run llama2 |
Mistral | 7B | 4.1GB | ollama run mistral |
本文就让我们一起入门Ollama。
Ollama支持各个平台:Mac、Windows 和 Linux,也提供了docker image。 在Ollama官网或者Github可以下载,然后一键安装Ollama框架:
curl -fsSL https://ollama.com/install.sh | sh
由于Ollama刚支持windows不久,在windows上的相关配置还不够完善,以下我将主要以Linux上运行Ollama来举例说明。
在Ollama安装完成后, 一般会自动启动 Ollama 服务,而且会自动设置为开机自启动。安装完成后,可以使用如下命令查看是否Ollama是否正常启动。如下例子中显示“Active: active (running)”表示Ollama已经正常启动。
1 $ systemctl status ollama
2 ● ollama.service - Ollama Service
3 Loaded: loaded (/etc/systemd/system/ollama.service; enabled; vendor preset: enabled)
4 Drop-In: /etc/systemd/system/ollama.service.d
5 └─environment.conf
6 Active: active (running) since Thu 2024-03-07 09:09:39 HKT; 4 days ago
7 Main PID: 19975 (ollama)
8 Tasks: 29 (limit: 69456)
9 Memory: 1.1G
10 CPU: 14min 44.702s
11 CGroup: /system.slice/ollama.service
12 └─19975 /usr/local/bin/ollama serve
在Linux上,如果Ollama未启动,可以用如下命令启动 Ollama 服务:ollama serve
,或者 sudo systemctl start ollama
。
通过分析Linux的安装脚本install.sh,就会看到其中已经将ollama serve
配置为一个系统服务,所以可以使用systemctl
来 start / stop ollama进程。
1 status "Creating ollama systemd service..."
2 cat </dev/null
3 [Unit]
4 Description=Ollama Service
5 After=network-online.target
6
7 [Service]
8 ExecStart=$BINDIR/ollama serve
9 User=ollama
10 Group=ollama
11 Restart=always
12 RestartSec=3
13 Environment="PATH=$PATH"
启动Ollama服务后,可以查看当前的Ollama版本,以及常用命令
1 ~$ ollama -v
2 ollama version is 0.1.20
3 ~$ ollama --help
4 Large language model runner
5
6 Usage:
7 ollama [flags]
8 ollama [command]
9
10 Available Commands:
11 serve Start ollama
12 create Create a model from a Modelfile
13 show Show information for a model
14 run Run a model
15 pull Pull a model from a registry
16 push Push a model to a registry
17 list List models
18 cp Copy a model
19 rm Remove a model
20 help Help about any command
21
22 Flags:
23 -h, --help help for ollama
24 -v, --version Show version information
25
16 Use "ollama [command] --help" for more information about a command.
至此,已经完成Ollama框架的安装,接下来,可以用一条命令在本地运行大语言模型。以著名的羊驼举例:ollama run llama2
。
如果还没有下载过指定的大语言模型,这条命令将会先执行ollama pull llama2
,将大语言模型下载到本地,再在本地运行大语言模型。
下载完成后,运行效果如下:
1 :~$ ollama run llama2
2 >>> who are you?
3
4 I am LLaMA, an AI assistant developed by Meta AI that can understand and respond to human input in a conversational manner. I am trained on a massive dataset of text from the internet and can
5 generate human-like responses to a wide range of topics and questions. I can be used to create chatbots, virtual assistants, and other applications that require natural language understanding and
6 generation capabilities.
7
8 >>> Send a message (/? for help)
Ollama还提供了API接口:
1 curl http://localhost:11434/api/generate -d '{
2 "model": "llama2",
3 "prompt":"Why is the sky blue?",
4 "stream": false
5 }'
返回结果如下:
1 {
2 "model": "llama2",
3 "created_at": "2024-02-26T04:35:10.787352404Z",
4 "response": "The sky appears blue because of a phenomenon called Rayleigh scattering, which occurs when sunlight enters Earth's atmosphere. The sunlight encounters tiny molecules of gases such as nitrogen and oxygen, which scatter the light in all directions. The shorter wavelengths of light, such as blue and violet, are scattered more than the longer wavelengths, such as red and orange. This is known as Rayleigh scattering, named after Lord Rayleigh, who first described the phenomenon in the late 19th century. As a result of this scattering, the light that reaches our eyes from the sun appears blue, especially when viewed from a distance. The closer we get to the horizon, the more the blue color appears to fade, as the light has to travel through more of the atmosphere, which scatters the shorter wavelengths even more. It's worth noting that the exact shade of blue can vary depending on the time of day and atmospheric conditions. For example, during sunrise and sunset, when the sun is low in the sky, the sky can take on a more orange or red hue due to the scattering of light by atmospheric particles. So, to summarize, the sky appears blue because of the way light interacts with the tiny molecules of gases in Earth's atmosphere, particularly nitrogen and oxygen.",
5 "done": true,
6 "total_duration": 7001870820,
7 "load_duration": 4930376,
8 "prompt_eval_duration": 60907000,
9 "eval_count": 309,
10 "eval_duration": 6931593000
11 }
使用API接口,就可以实现更多灵活的功能,比如与IDE插件配合,实现本地的编程助手,可参考如下文章: 零基础入门AI:搭建本地的编程助手
在Linux上运行命令journalctl -u ollama
,即可查看运行日志。
在Linux上创建如下配置文件,并配置环境变量 OLLAMA_HOST
来指定对局域网提供服务的地址,再重启Ollama服务即可。
1 :~$ cat /etc/systemd/system/ollama.service.d/environment.conf
2 [Service]
3 Environment=OLLAMA_HOST=0.0.0.0:11434
如此配置后,即可由一台GPU服务器为本地局域网提供大语言模型的服务。
在Linux上创建如下配置文件,并配置环境变量 CUDA_VISIBLE_DEVICES
来指定运行Ollama的GPU,再重启Ollama服务即可。
1 :~$ cat /etc/systemd/system/ollama.service.d/environment.conf
2 [Service]
3 Environment=CUDA_VISIBLE_DEVICES=1,2
默认情况下,不同操作系统存储的路径如下:
~/.ollama/models
/usr/share/ollama/.ollama/models
C:\Users.ollama\models
Linux平台安装Ollama时,默认安装时会创建用户ollama,再将模型文件存储到该用户的目录/usr/share/ollama/.ollama/models
。但由于大模型文件往往特别大,有时需要将大模型文件存储到专门的数据盘,此时就需要修改大模型文件的存储路径。
官方提供的方法是设置环境变量“OLLAMA_MODELS”,但我在Linux上尝试后,并没有成功。
分析Linux版的安装脚本install.sh后,我发现是由于其中创建了用户ollama和用户组ollama,然后将大模型存储到了该用户的目录/usr/share/ollama/.ollama/models
,而我的帐户对ollama帐户的一些操作并不能生效,即使我再手动将我的帐户添加进ollama用户组,也仍然会有一些权限问题,导致对ollama帐户的目录操作不生效。
由于新建的ollama帐户并没有给我带来额外的便利,最后我用以下步骤来实现修改大模型文件的存储路径:
修改安装文件 install.sh,取消其中创建用户ollama的步骤,参考如下:
1 # if ! id ollama >/dev/null 2>&1; then
2 # status "Creating ollama user..."
3 # $SUDO useradd -r -s /bin/false -m -d /usr/share/ollama ollama
4 # fi
5 # status "Adding current user to ollama group..."
6 # $SUDO usermod -a -G ollama $(whoami)
修改安装文件 install.sh,使用我的帐户来启动ollama服务,参考如下:
1 status "Creating ollama systemd service..."
2 cat </dev/null
3 [Unit]
4 Description=Ollama Service
5 After=network-online.target
6
7 [Service]
8 ExecStart=$BINDIR/ollama serve
9 User=
10 Group=
修改安装文件 install.sh,添加如下配置中指定环境变量OLLAMA_MODELS
指定存储路径,再用此安装文件来安装ollama。
1 Environment="OLLAMA_MODELS=/home/paco/lab/LLM/ollama/OLLAMA_MODELS"
或者在安装完成后,创建如下配置文件,并配置环境变量 OLLAMA_MODELS
来指定存储路径,再重启Ollama服务。
1 :~$ cat /etc/systemd/system/ollama.service.d/environment.conf
2 [Service]
3 Environment=OLLAMA_MODELS=/OLLAMA_MODELS