87 lines
2.6 KiB
Python
87 lines
2.6 KiB
Python
|
import ollama
|
|||
|
|
import requests
|
|||
|
|
|
|||
|
|
# Python中使用ollama API的方式
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 使用Python直接调用ollama：方式一
|
|||
|
|
def dm01():
|
|||
|
|
"""
|
|||
|
|
如果是访问本地服务器直接使用ollama即可
|
|||
|
|
如果是访问远程服务器，则需要使用ollama.Client创建一个新的对象并指定主机地址: new_ollama = ollama.Client(url)
|
|||
|
|
"""
|
|||
|
|
# 获取用户要问的问题
|
|||
|
|
pro = '给我讲一个笑话'
|
|||
|
|
|
|||
|
|
# 访问本地服务器,获取响应
|
|||
|
|
res = ollama.chat(
|
|||
|
|
model="qwen2.5:0.5b",
|
|||
|
|
messages=[{"role": "user", "content": pro}]
|
|||
|
|
)
|
|||
|
|
print('res-->', res) # 整个对象
|
|||
|
|
# 只获取响应内容的两种方式
|
|||
|
|
print(res.message.content)
|
|||
|
|
# print(res['message']['content'])
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 使用Python直接调用ollama：方式二
|
|||
|
|
def dm02():
|
|||
|
|
from ollama import Client
|
|||
|
|
client = Client(host='http://127.0.0.1:11434')
|
|||
|
|
response = client.chat(model='qwen2.5:0.5b', messages=[
|
|||
|
|
{
|
|||
|
|
'role': 'user',
|
|||
|
|
'content': '为什么天空是蓝⾊的？',
|
|||
|
|
},
|
|||
|
|
])
|
|||
|
|
print(response['message']['content'])
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 使用Python直接调用ollama：方式三
|
|||
|
|
def dm03():
|
|||
|
|
stream = ollama.chat(
|
|||
|
|
model='qwen2.5:0.5b',
|
|||
|
|
messages=[{'role': 'user', 'content': '讲一个不少于200字的笑话'}],
|
|||
|
|
stream=True,
|
|||
|
|
)
|
|||
|
|
# print("stream-->", stream) # 只是个对象
|
|||
|
|
for chunk in stream:
|
|||
|
|
# print("chunk-->", chunk)
|
|||
|
|
print(chunk['message']['content'], end='', flush=True)
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 使用requests库调用ollama
|
|||
|
|
def dm04():
|
|||
|
|
headers = {"Content-Type": "application/json"}
|
|||
|
|
data = {
|
|||
|
|
"model": "deepseek-r1:1.5b", # 模型选择
|
|||
|
|
"options": {
|
|||
|
|
"temperature": 0. # 为0表示不让模型⾃由发挥，输出结果相对较固定，> 0的话，输出的结果会⽐较放⻜⾃我
|
|||
|
|
},
|
|||
|
|
"stream": False, # 流式输出
|
|||
|
|
"messages": [{
|
|||
|
|
"role": "system",
|
|||
|
|
"content": "你是谁？"
|
|||
|
|
}] # 对话列表
|
|||
|
|
}
|
|||
|
|
response = requests.post(url='http://127.0.0.1:11434/api/chat', json=data, headers=headers, timeout=60)
|
|||
|
|
print(response.json()['message']['content'])
|
|||
|
|
|
|||
|
|
|
|||
|
|
# 使用langchain框架调用ollama
|
|||
|
|
def dm05():
|
|||
|
|
from langchain_community.llms import Ollama
|
|||
|
|
# 如果⾃⼰本地系统有ollama服务，可以省略base_url
|
|||
|
|
# temperature：控制生成文本随机性的重要参数，取值越小（接近0，模型输出结果越确定，越大，模型输出结果越开放）
|
|||
|
|
llm = Ollama(base_url="http://127.0.0.1:11434",
|
|||
|
|
model="deepseek-r1:1.5b", temperature=0)
|
|||
|
|
res = llm.invoke("给我讲一个笑话")
|
|||
|
|
print(res)
|
|||
|
|
|
|||
|
|
|
|||
|
|
if __name__ == '__main__':
|
|||
|
|
# dm01()
|
|||
|
|
# dm02()
|
|||
|
|
# dm03()
|
|||
|
|
# dm04()
|
|||
|
|
dm05()
|