ollama
本地安装ollama
gemma
下载并部署模型 本机资源有限,可以下个2b的相对较小的模型
执行命令
ollama run gemma:2b
客户端调用
import requests
import json
data = {
"model": "gemma:2b",
"messages": [
{ "role": "user", "content": "hi, who are you?" }
]
}
response = requests.post('http://localhost:11434/api/chat', json=data,stream=True)
result = ''
for line in response.iter_lines():
if line:
json_data = json.loads(line)
content = json_data.get('message', {}).get('content', '')
print(content, end='')
输出: