保姆级！本地部署DeepSeek-R1大模型安装Ollama Api 后，Postman本地调用 deepseek

aidu_pl">

要在Postman中访问Ollama API并调用DeepSeek模型，你需要遵循以下步骤。首先，确保你有一个有效的Ollama服务器实例运行中，并且DeepSeek模型已经被加载。

可以参考我的这篇博客保姆级！使用Ollama本地部署DeepSeek-R1大模型并java通过api 调用

具体的代码实现参考我这个博客保姆级！springboot访问Ollama API并调用DeepSeek模型 Api

使用Postman配置请求

单论对话

请求类型为 POST。

在“输入”框中输入Ollama的API URL，通常是 http://localhost:11434/v1/completions（确保你的Ollama服务正在本地运行，并且使用的是正确的端口）。

在“头”标签页中，添加以下内容：

Key: Content-Type

Value: application/json

编写请求体:

在“Body”标签页中，选择“raw”并设置编码为“JSON (application/json)”。

输入以下JSON格式的请求体：

{"model": "deepseek-r1:32b","prompt": "你好","max_tokens": 50,"temperature": 0.7,"top_p": 1,"stream": false
}

model: 指定要使用的模型名称。

prompt: 输入你想要模型生成文本的起始语句。

max_tokens: 生成文本的最大token数。

temperature: 控制输出的随机性。较低的值意味着更确定的输出。值越低，生成的文本越保守和一致，越高则越具创意。如果不需要非常有创意的回答，可以将其设置为 0.7 或更低。

top_p: 核取样参数，用于多样性控制。控制采样的累积概率，通常在 0 到 1 之间，越接近 1，模型越可能尝试不同的答案。

stream: 如果设置为true，将流式传输结果。

具体如下图
在这里插入图片描述

上面的 http://localhost:11434/v1/completions 是deepseek的api 地址我们继续访问一下 Ollama Api http://192.168.110.118:7777/api/generate

在这里插入图片描述
一样是可以的

多轮对话补全

我们访问 http://localhost:7777/api/chat 试一下

请求体传入

{"model": "deepseek-r1:14b","messages": [{"role": "user", "content": "你好"},{"role": "assistant", "content": "你好！很高兴见到你，有什么我可以帮忙的吗？无论是学习、工作还是生活中的问题，都可以告诉我哦！😊"},{"role": "user", "content": "请问1+1等于几？"}],"max_tokens": 50,"temperature": 0.7,"top_p": 1,"stream": false
}

{"model": "deepseek-r1:14b","created_at": "2025-02-25T01:06:00.1257844Z","message": {"role": "assistant","content": "<think>\n嗯，用户之前用中文问候了我，我也回应了他。现在他问了一个数学题：“请问1+1等于几？”这看起来很简单，可能是在测试我的基本能力，或者他真的需要帮助计算。\n\n首先，我要确定他的身份。也许是个学生，在做作业时遇到了问题，或者是家长在辅导孩子，甚至可能是想确认我是否能正确回答基础问题。不管怎样，这个问题本身不难，但作为AI，我需要准确无误地回答。\n\n接下来，考虑用户的真实需求。表面上是数学题，但实际上他可能在测试我的反应速度和准确性，或者有其他隐藏的需求，比如验证我的知识库是否更新到最新的数学知识。\n\n另外，用户的使用场景可能是学习环境，比如在学校或家里。如果是在课堂上，学生可能会用这个问题来确认老师的知识点是否正确，或者作为开始对话的简单问题。\n\n可能还有深层需求，比如用户想了解我能否处理更复杂的数学问题，所以先从简单的题目入手。因此，正确的回答不仅是对当前问题的回答，也是建立信任的第一步。\n\n综上所述，我应该直接给出准确的答案，并且保持友好的语气，让用户感到被支持和帮助。\n</think>\n\n你好！1 + 1 等于 **2**。😊"},"done_reason": "stop","done": true,"total_duration": 58231935600,"load_duration": 10134100,"prompt_eval_count": 40,"prompt_eval_duration": 459000000,"eval_count": 266,"eval_duration": 57761000000
}

在这里插入图片描述

文件上传并提问

先看这里 Official Prompts

在这里插入图片描述
文档内容翻译如下：

官方提示
在官方的DeepSeek网络/应用程序中，我们不使用系统提示，而是为文件上传和网络搜索设计了两个特定的提示，以获得更好的用户体验。此外，web/app中的温度为0.6。
对于文件上传，请按照模板创建提示，其中｛file_name｝、｛file_content｝和｛question｝是参数。
file_template =
“”“[file name]: {file_name}
[file content begin]
{file_content}
[file content end]
{question}”“”

我们先用postman请求试试
14b模型是不支持文件的我们用32b model
在这里插入图片描述

body内容如下

{"model": "deepseek-r1:32b","messages":[{"role":"user","content":"[file name]: 1111.txt\n[file content begin]\n123456789\n[file content end]\n文档中写了什么？"}],"max_tokens": 50,"temperature": 0.7,"top_p": 1,"stream": true
}

返回内容

{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:26.5392927Z","message": {"role": "assistant","content": "<think>"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:27.1561499Z","message": {"role": "assistant","content": "\n"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:27.753368Z","message": {"role": "assistant","content": "嗯"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:28.3850472Z","message": {"role": "assistant","content": "，"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:29.0560203Z","message": {"role": "assistant","content": "我"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:29.7279329Z","message": {"role": "assistant","content": "需要"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:30.3989966Z","message": {"role": "assistant","content": "回答"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:31.0669576Z","message": {"role": "assistant","content": "用户"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:31.738556Z","message": {"role": "assistant","content": "关于"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:32.4179008Z","message": {"role": "assistant","content": "文件"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:33.0881279Z","message": {"role": "assistant","content": "内容"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:33.6911414Z","message": {"role": "assistant","content": "的问题"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:34.2884323Z","message": {"role": "assistant","content": "。"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:34.8879486Z","message": {"role": "assistant","content": "用户提供"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:35.5587474Z","message": {"role": "assistant","content": "了一个"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:36.344929Z","message": {"role": "assistant","content": "文本"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:36.9433939Z","message": {"role": "assistant","content": "文件"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:37.5398887Z","message": {"role": "assistant","content": "的"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:38.1336335Z","message": {"role": "assistant","content": "名称"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:38.8162284Z","message": {"role": "assistant","content": "和"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:39.602469Z","message": {"role": "assistant","content": "内容"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:40.1967921Z","message": {"role": "assistant","content": "。\n\n"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:40.7911033Z","message": {"role": "assistant","content": "首先"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:41.3901296Z","message": {"role": "assistant","content": "，"},"done": false
}
{"model": "deepseek-r1:32b","created_at": "2025-02-25T09:49:42.054903Z","message"