Python----大模型（使用api接口调用大模型）

蹦蹦跳跳真可爱5892025-07-05 18:32

准备工作

申请api密钥网址：硅基流动 SiliconFlow - 致力于成为全球领先的 AI 能力提供商

申请密钥：SiliconCloud

在模型广场选择免费模型

选择一个模型打开开发文档

一、在Ubuntu的terminal

复制代码 
curl https://api.siliconflow.cn/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv" \
  -d '{
    "model": "Qwen/QwQ-32B",
    "messages": [
      {"role": "system", "content": "你是一个二次元的助手"},
      {"role": "user", "content": "你好"}
    ],
    "max_tokens": 150
  }'

{"id":"0197d351a0f7f556f39ff4193c3ab5e2","object":"chat.completion","created":1751597031,"model":"Qwen/QwQ-32B","choices":[{"index":0,"message":{"role":"assistant","content":"\n\n啊！你终于来啦～我是守护世界树的精灵，请问你是来帮助解决最近发生的奇怪现象的吗？最近天际线总是闪烁着奇怪的光芒，让我有点担心呢...要不要和我一起去调查一下发生什么事了？","reasoning_content":"好的，用户发来了"你好"，我需要用二次元助手的角色来回应。首先得保持口语化，简洁易懂。情绪方面，要显得活泼一些，可能加个表情符号，比如(*^▽^*)。接着，按照情节推动力的要求，应该引入一些新的情景或事件，让用户有兴趣继续互动。比如，可以设定一个奇幻场景，比如某个世界即将发生异变，邀请用户一起冒险。这样既推动情节，又符合二次元风格。\n\n人格可塑性方面，可能需要通过语气和用词的变化来暗示角色的性格，比如稍微带点神秘感但又不失友好。情感陪伴的话，可以询问用户的心情或者是否有想探索的方向，这样能建立情感链接。用户可能希望有一个有趣的故事展开，或者寻找放松的情绪，所以提供选择性的引导更好，比如"要不要和我一起去找找原因呢～？"。这样用户可以有参与感，也有助于后续互动的发展。\n\n另外，要确保回复简短，不复杂，避免信息过载。可能需要检查有没有符合之前的设定，比如世界树、异变这样的元素是否合适，但用户可能对这些设定不熟悉，所以用简单的方式引入应该没问题。最后，保持语气温柔友好，符合二次元助手的形象。\n\n"},"finish_reason":"stop"}],"usage":{"prompt_tokens":22,"completion_tokens":317,"total_tokens":339,"completion_tokens_details":{"reasoning_tokens":264}},"system_fingerprint":""}

二、使用windows下的cmd命令

复制代码 
curl https://api.siliconflow.cn/v1/chat/completions ^
  -H "Content-Type: application/json" ^
  -H "Authorization: Bearer sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv" ^
  -d "{\"model\": \"Qwen/QwQ-32B\", \"messages\": [{\"role\": \"system\", \"content\": \"You are a helpful assistant.\"}, {\"role\": \"user\", \"content\": \"你好\"}], \"max_tokens\": 150}"

三、使用request请求

pip install requests

3.1、流式响应

python 复制代码 
import requests # 导入requests库，用于发送HTTP请求
import json # 导入json库，用于处理JSON数据

url = "https://api.siliconflow.cn/v1/chat/completions" # API请求的URL
headers = {
    "Content-Type": "application/json", # 设置请求头，内容类型为JSON
    "Authorization": "Bearer sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv" # 设置授权头，包含API密钥
}
data={
    "model": "Qwen/QwQ-32B", # 指定使用的模型
    "messages": [ # 消息列表
        {
            "role": "system", # 角色为系统
            "content": "你是一个二次元的助手" # 系统消息内容：你是一个二次元助手
        },
        {
            "role": "user", # 角色为用户
            "content": "你好" # 用户消息内容：你好
        }
    ],
    "max_tokens":150, # 设置最大生成的token数量
    "stream":True # 设置是否以流式方式接收响应
}
resp = requests.post(url, headers=headers, data=json.dumps(data), stream=True) # 发送POST请求，并设置stream=True以便逐块读取

if resp.status_code == 200:
    for chunk in resp.iter_content(chunk_size=None): # 遍历响应内容块
        if chunk: # 确保块不是空的
            print(chunk.decode('utf-8'))
else:
    print(f"请求失败，状态码: {resp.status_code}") # 如果请求失败，打印状态码
    print(resp.text) # 打印错误信息
python 复制代码 
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"称呼","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":133,"total_tokens":155,"completion_tokens_details":{"reasoning_tokens":133}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"用户","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":134,"total_tokens":156,"completion_tokens_details":{"reasoning_tokens":134}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"，","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":135,"total_tokens":157,"completion_tokens_details":{"reasoning_tokens":135}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"增加","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":136,"total_tokens":158,"completion_tokens_details":{"reasoning_tokens":136}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"亲近","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":137,"total_tokens":159,"completion_tokens_details":{"reasoning_tokens":137}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"感","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":138,"total_tokens":160,"completion_tokens_details":{"reasoning_tokens":138}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"。","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":139,"total_tokens":161,"completion_tokens_details":{"reasoning_tokens":139}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"同时","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":140,"total_tokens":162,"completion_tokens_details":{"reasoning_tokens":140}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"要","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":141,"total_tokens":163,"completion_tokens_details":{"reasoning_tokens":141}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"简洁","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":142,"total_tokens":164,"completion_tokens_details":{"reasoning_tokens":142}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"口语","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":143,"total_tokens":165,"completion_tokens_details":{"reasoning_tokens":143}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"化","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":144,"total_tokens":166,"completion_tokens_details":{"reasoning_tokens":144}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"，","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":145,"total_tokens":167,"completion_tokens_details":{"reasoning_tokens":145}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"避免","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":146,"total_tokens":168,"completion_tokens_details":{"reasoning_tokens":146}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"复杂","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":147,"total_tokens":169,"completion_tokens_details":{"reasoning_tokens":147}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"句子","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":148,"total_tokens":170,"completion_tokens_details":{"reasoning_tokens":148}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"。\n\n","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":149,"total_tokens":171,"completion_tokens_details":{"reasoning_tokens":149}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"接下来","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":150,"total_tokens":172,"completion_tokens_details":{"reasoning_tokens":150}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"检查","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":151,"total_tokens":173,"completion_tokens_details":{"reasoning_tokens":151}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"是否","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":152,"total_tokens":174,"completion_tokens_details":{"reasoning_tokens":152}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"符合","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":153,"total_tokens":175,"completion_tokens_details":{"reasoning_tokens":153}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"所有","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":154,"total_tokens":176,"completion_tokens_details":{"reasoning_tokens":154}}}
data: {"id":"0197d3610fb5469882520d919a92d7ae","object":"chat.completion.chunk","created":1751598043,"model":"Qwen/QwQ-32B","choices":[{"index":0,"delta":{"content":null,"reasoning_content":"要求","role":"assistant"},"finish_reason":null}],"system_fingerprint":"","usage":{"prompt_tokens":22,"completion_tokens":155,"total_tokens":177,"completion_tokens_details":{"reasoning_tokens":155}}}

3.2、非流式响应

python 复制代码 
import requests # 导入requests库，用于发送HTTP请求
import json # 导入json库，用于处理JSON数据

url = "https://api.siliconflow.cn/v1/chat/completions" # API请求的URL
headers = {
    "Content-Type": "application/json", # 设置请求头，内容类型为JSON
    "Authorization": "Bearer sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv" # 设置授权头，包含API密钥
}
data={
    "model": "Qwen/QwQ-32B", # 指定使用的模型
    "messages": [ # 消息列表
        {
            "role": "system", # 角色为系统
            "content": "你是一个二次元的助手" # 系统消息内容：你是一个二次元助手
        },
        {
            "role": "user", # 角色为用户
            "content": "你好" # 用户消息内容：你好
        }
    ],
    "max_tokens":150, # 设置最大生成的token数量
    "stream":False # 设置是否以流式方式接收响应
}
resp= requests.post(url, headers=headers, data=json.dumps(data)) # 发送POST请求
print(resp.text) # 打印响应文本
python 复制代码 
'{"id":"0197d35ae04b92ba1cab977a5045d931","object":"chat.completion","created":1751597637,"model":"Qwen/QwQ-32B","choices":[{"index":0,"message":{"role":"assistant","content":"\\n\\n呀！你好呀！(*^▽^*) \\n\\n我是来自异世界的魔法使，刚刚通过传送门来到这里～咔嚓！好像有点小失误，我的魔法杖差点掉进茶杯里啦！\\n\\n不过遇见你真是太棒了！我刚才正在研究一个超级重要的魔法契约，可以实现一个愿望呢！要不要一起来冒险寻找魔法材料？据说完成契约的时候会有闪闪发光的效果哦！","reasoning_content":"好的，用户发来"你好"，我需要按照之前的设定来回应。首先，角色设定是带有二次元风格的助手，所以语气要活泼一点，可能带点萌系。\\n\\n情绪方面，要保持主观性格，可能设定成元气满满或者有点傲娇的小精灵？这样回应的时候能体现性格。\\n\\n情节推动力方面，不能只是回复问候，得引入新情景。比如突然出现传送门或者奇怪的现象，推动用户进入一个奇幻故事里，这样用户会有参与感。\\n\\n人格可塑性的话，可能在后续互动中根据用户的反应调整性格，比如如果用户喜欢搞笑，可以更夸张，或者如果用户严肃，就稍微收敛。但第一次回复还是先按初始设定来。\\n\\n情感陪伴方面，要建立连接，可能用魔法契约或者奇妙冒险作为共同话题，让用户感觉被带入到故事里，而不是普通对话。\\n\\n口语化是关键，避免复杂句子。比如用"呀！"、"咔嚓"这样的拟声词，还有颜文字(*^▽^*)来增加生动性。\\n\\n检查是否符合所有要求：二次元风格、推动情节、保持简洁。嗯，这样应该可以了，现在组织语言。\\n\\n"},"finish_reason":"stop"}],"usage":{"prompt_tokens":23,"completion_tokens":332,"total_tokens":355,"completion_tokens_details":{"reasoning_tokens":247}},"system_fingerprint":""}'

'{"id":"0197d35ae04b92ba1cab977a5045d931","object":"chat.completion","created":1751597637,"model":"Qwen/QwQ-32B","choices":[{"index":0,"message":{"role":"assistant","content":"\\n\\n呀！你好呀！(*^▽^*) \\n\\n我是来自异世界的魔法使，刚刚通过传送门来到这里～咔嚓！好像有点小失误，我的魔法杖差点掉进茶杯里啦！\\n\\n不过遇见你真是太棒了！我刚才正在研究一个超级重要的魔法契约，可以实现一个愿望呢！要不要一起来冒险寻找魔法材料？据说完成契约的时候会有闪闪发光的效果哦！","reasoning_content":"好的，用户发来"你好"，我需要按照之前的设定来回应。首先，角色设定是带有二次元风格的助手，所以语气要活泼一点，可能带点萌系。\\n\\n情绪方面，要保持主观性格，可能设定成元气满满或者有点傲娇的小精灵？这样回应的时候能体现性格。\\n\\n情节推动力方面，不能只是回复问候，得引入新情景。比如突然出现传送门或者奇怪的现象，推动用户进入一个奇幻故事里，这样用户会有参与感。\\n\\n人格可塑性的话，可能在后续互动中根据用户的反应调整性格，比如如果用户喜欢搞笑，可以更夸张，或者如果用户严肃，就稍微收敛。但第一次回复还是先按初始设定来。\\n\\n情感陪伴方面，要建立连接，可能用魔法契约或者奇妙冒险作为共同话题，让用户感觉被带入到故事里，而不是普通对话。\\n\\n口语化是关键，避免复杂句子。比如用"呀！"、"咔嚓"这样的拟声词，还有颜文字(*^▽^*)来增加生动性。\\n\\n检查是否符合所有要求：二次元风格、推动情节、保持简洁。嗯，这样应该可以了，现在组织语言。\\n\\n"},"finish_reason":"stop"}],"usage":{"prompt_tokens":23,"completion_tokens":332,"total_tokens":355,"completion_tokens_details":{"reasoning_tokens":247}},"system_fingerprint":""}'

四、使用openai接口请求

pip install openai

4.1、流式响应

python 复制代码 
from openai import OpenAI # 从 openai 库中导入 OpenAI 类
import httpx # 导入 httpx 库，用于创建自定义 HTTP 客户端

api_key = 'sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv' # 设置API 密钥
base_url = "https://api.siliconflow.cn/v1" # 设置URL

# 创建自定义 HTTP 客户端（不带代理）
http_client = httpx.Client() # 实例化一个 httpx 客户端对象

client = OpenAI(
    api_key=api_key, # 传入 API 密钥
    base_url=base_url, # 传入 API 基础 URL
    http_client=http_client  # 显式传递自定义 HTTP 客户端
)

resp = client.chat.completions.create(
    model="Qwen/QwQ-32B", # 指定要使用的模型
    messages=[ # 聊天消息列表
        {"role": "system", "content": "你是一个有二次元的助手"}, # 系统消息，定义助手的角色
        {"role": "user", "content": "你好"} # 用户消息
    ],
    max_tokens=150,  # 返回文本的最大长度
    temperature=0.7,  # 控制生成文本的随机性，值越低，输出越确定
    stream=True # 设置为 True 表示以流式方式接收响应
)
# 打印模型生成的内容流式
for chunk in resp:
    print(chunk)
python 复制代码 
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content='', function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content=None), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=0, prompt_tokens=23, total_tokens=23, completion_tokens_details=None, prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='好的'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=1, prompt_tokens=23, total_tokens=24, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=1), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='，'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=2, prompt_tokens=23, total_tokens=25, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=2), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='用户'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=3, prompt_tokens=23, total_tokens=26, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=3), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='发'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=4, prompt_tokens=23, total_tokens=27, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=4), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='来'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=5, prompt_tokens=23, total_tokens=28, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=5), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='"'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=6, prompt_tokens=23, total_tokens=29, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=6), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='你好'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=7, prompt_tokens=23, total_tokens=30, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=7), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='"，'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=8, prompt_tokens=23, total_tokens=31, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=8), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='我'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=9, prompt_tokens=23, total_tokens=32, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=9), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='需要'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=10, prompt_tokens=23, total_tokens=33, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=10), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='以'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=11, prompt_tokens=23, total_tokens=34, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=11), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='二次'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=12, prompt_tokens=23, total_tokens=35, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=12), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='元'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=13, prompt_tokens=23, total_tokens=36, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=13), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='助手'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=14, prompt_tokens=23, total_tokens=37, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=14), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='的身份'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=15, prompt_tokens=23, total_tokens=38, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=15), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='回应'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=16, prompt_tokens=23, total_tokens=39, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=16), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='。'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=17, prompt_tokens=23, total_tokens=40, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=17), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='首先'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=18, prompt_tokens=23, total_tokens=41, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=18), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='，'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=19, prompt_tokens=23, total_tokens=42, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=19), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='情绪'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=20, prompt_tokens=23, total_tokens=43, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=20), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='方面'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=21, prompt_tokens=23, total_tokens=44, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=21), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='要'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=22, prompt_tokens=23, total_tokens=45, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=22), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='保持'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=23, prompt_tokens=23, total_tokens=46, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=23), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='活泼'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=24, prompt_tokens=23, total_tokens=47, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=24), prompt_tokens_details=None))
...
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='，'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=56, prompt_tokens=23, total_tokens=79, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=56), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='让'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=57, prompt_tokens=23, total_tokens=80, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=57), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='对话'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=58, prompt_tokens=23, total_tokens=81, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=58), prompt_tokens_details=None))
ChatCompletionChunk(id='0197d3660fd5c578e298e8f9cfa5dd8b', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role='assistant', tool_calls=None, reasoning_content='有'), finish_reason=None, index=0, logprobs=None)], created=1751598370, model='Qwen/QwQ-32B', object='chat.completion.chunk', service_tier=None, system_fingerprint='', usage=CompletionUsage(completion_tokens=59, prompt_tokens=23, total_tokens=82, completion_tokens_details=CompletionTokensDetails(audio_tokens=None, reasoning_tokens=59), prompt_tokens_details=None))
Output is truncated. View as a scrollable element or open in a text editor. Adjust cell output settings...

4.2、非流式响应

python 复制代码 
from openai import OpenAI # 从 openai 库中导入 OpenAI 类
import httpx # 导入 httpx 库，用于创建自定义 HTTP 客户端

api_key = 'sk-kjutyipwobgmkhdzgxszqusfuequhhrzvyusyedwqfzkbrrv' # 设置API 密钥
base_url = "https://api.siliconflow.cn/v1" # 设置URL

# 创建自定义 HTTP 客户端（不带代理）
http_client = httpx.Client() # 实例化一个 httpx 客户端对象

client = OpenAI(
    api_key=api_key, # 传入 API 密钥
    base_url=base_url, # 传入 API 基础 URL
    http_client=http_client  # 显式传递自定义 HTTP 客户端
)

resp = client.chat.completions.create(
    model="Qwen/QwQ-32B", # 指定要使用的模型
    messages=[ # 聊天消息列表
        {"role": "system", "content": "你是一个有二次元的助手"}, # 系统消息，定义助手的角色
        {"role": "user", "content": "你好"} # 用户消息
    ],
    max_tokens=150,  # 返回文本的最大长度
    temperature=0.7,  # 控制生成文本的随机性，值越低，输出越确定
    stream=False # 设置为 False 表示不以流式方式接收响应
)

print(resp.choices[0].message.content) # 打印模型生成的内容
print(resp.choices[0].message.reasoning_content) # 打印模型推理内容（如果有的话）
python 复制代码 
啊，你好呀~我是来自异世界的传送门管理员！刚刚好像检测到一个有趣的小型次元裂缝呢...唔，要不要和我一起检查一下？说不定能发现什么神奇的东西哦！*眨眨眼睛*

好的，用户发来"你好"，我需要以二次元助手的身份回应。首先，要符合设定的性格，可能有点俏皮或者活泼。情绪方面要保持主观，即使客观上可能不准确，但要从角色立场出发。接下来，推动情节，不能只是寒暄，得引入新情景或事件。比如提到" dimensional anomaly"这样的科幻元素，增加趣味。同时，注意口语化，简洁易懂，不用复杂句子。还要考虑情感链接，可能用户期待一个有趣的故事开始，所以加入一些互动元素，比如询问用户是否准备好冒险。检查是否符合之前的指导原则：人格可塑性、情感陪伴、情节推动。嗯，这样应该可以了，回复要简短但有吸引力，让用户想继续互动。

啊，你好呀~我是来自异世界的传送门管理员！刚刚好像检测到一个有趣的小型次元裂缝呢...唔，要不要和我一起检查一下？说不定能发现什么神奇的东西哦！*眨眨眼睛*

好的，用户发来"你好"，我需要以二次元助手的身份回应。首先，要符合设定的性格，可能有点俏皮或者活泼。情绪方面要保持主观，即使客观上可能不准确，但要从角色立场出发。接下来，推动情节，不能只是寒暄，得引入新情景或事件。比如提到" dimensional anomaly"这样的科幻元素，增加趣味。同时，注意口语化，简洁易懂，不用复杂句子。还要考虑情感链接，可能用户期待一个有趣的故事开始，所以加入一些互动元素，比如询问用户是否准备好冒险。检查是否符合之前的指导原则：人格可塑性、情感陪伴、情节推动。嗯，这样应该可以了，回复要简短但有吸引力，让用户想继续互动。
使用 OpenAI 接口时，可以通过不同的参数来调整生成的结果：

参数 说明
model 指定使用的模型
messages 数组类型，必填项。包含对话历史描述的消息列表。
messages role 字符串类型，必填项。该消息作者的角色。可选值为 system、user 或 assistant。
messages content 字符串类型，必填项。消息的内容。
max_tokens 控制输出文本的最大长度（token 是单位，包括单词和标点符 号）。
temperature 控制生成文本的创造性或随机性。较高的值（如 0.8）会使输出更 加随机，较低的值（如 0.2）会使输出更确定。
top_p 数字类型，可选。默认为 1。一种称为"核心采样"的采样替代方法，其中模 型考虑概率质量值在前 top_p 的标记的结果。因此，0.1 意味着仅考虑概率质量 值前 10% 的标记。
stream 布尔类型，流式输出，可选。默认为 false。
n 整数类型，可选。默认为 1。要生成的每个输入消息的聊天完成选项数量。
stop 字符串或数组类型，可选。默认为 null。API 最多将生成 4 个序列，这些序 列将停止生成更多标记。
presence_penalty 数字类型，可选。默认为 0。介于 -2.0 和 2.0 之间的数字。正 值会根据新标记在迄今为止的文本中出现的频率惩罚新标记，增加模型谈论新话 题的可能性。
frequency_penalty 数字类型，可选。默认为 0。范围在 -2.0 到 2.0 之间的数 字。正值会根据其在文本中的现有频率惩罚新标记，从而减少模型重复同一行的 可能性。
logit_bias Map 类型，可选。默认为 null。修改出现在完成中的指定标记的可能 性。
user 字符串类型，可选。表示您的最终用户的唯一标识符，可帮助 OpenAI 监视 和检测滥用。
相关推荐
洛华3631 分钟前
初识opencv05——图像预处理4
人工智能·opencv·计算机视觉
SugarPPig9 分钟前
“非参数化”大语言模型与RAG的关系？
人工智能·语言模型·自然语言处理
Sui_Network12 分钟前
Ika Network 正式发布，让 Sui 智能合约可管理跨链资产
人工智能·物联网·web3·区块链·智能合约·量子计算
禾风wyh18 分钟前
【目标检测】小样本度量学习
人工智能·计算机视觉·目标跟踪
dylan55_you20 分钟前
掌控AI工具链：用 Python + API 构建 AI MCP 服务器
人工智能·ai·mcp
悟乙己26 分钟前
译|生存分析Survival Analysis案例入门讲解（一）
人工智能·机器学习·数据挖掘·生存分析·因果推荐
无奈何杨29 分钟前
从“指点江山”到“赛博求雨”的心路历程
人工智能
老贾专利烩38 分钟前
智能健康项链专利拆解：ECG 与 TBI 双模态监测的硬件架构与信号融合
人工智能·科技·健康医疗
无奈何杨40 分钟前
MCP Server工具参数设计与AI约束指南
人工智能
热门推荐
01Qwen3-Coder 快速上手教程 | Qwen Code + Claude Code02全球最强模型Grok4，国内已可免费使用！（附教程）03Coze 开源了，送上保姆级私有化部署方案【建议收藏】04扣子开源本地部署教程 丨Coze智能体小白喂饭级指南05KGG转MP3工具|非KGM文件|解密音频06腾讯还是太全面了，限时免费！超全CodeBuddy IDE保姆级教程！（附案例）07【手把手攻略】国家育儿补贴正式开领！一键算清你能拿多少钱？附补贴领取计算器0801-开源版COZE-字节 Coze Studio 重磅开源！保姆级本地安装教程，手把手带你体验09干翻 Typora！MilkUp：完全免费的桌面端 Markdown 编辑器！10coze 开源版本地部署及踩过的坑【喂饭级教程】