103 lines
3.0 KiB
Markdown
103 lines
3.0 KiB
Markdown
|
### 一、官方文档
|
|||
|
|
|
|||
|
|
**仓库地址**
|
|||
|
|
|
|||
|
|
https://github.com/netease-youdao/Confucius3-Math
|
|||
|
|
|
|||
|
|
https://github.com/netease-youdao/Confucius3-Math/blob/main/README_cn.md
|
|||
|
|
|
|||
|
|
**官网**
|
|||
|
|
https://confucius.youdao.com/
|
|||
|
|
|
|||
|
|
**魔搭社区**
|
|||
|
|
|
|||
|
|
https://www.modelscope.cn/models/netease-youdao/Confucius3-Math
|
|||
|
|
|
|||
|
|
**论文**
|
|||
|
|
|
|||
|
|
https://www.modelscope.cn/papers/2506.18330
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
### 二、本地部署
|
|||
|
|
|
|||
|
|
**1. 克隆仓库**
|
|||
|
|
|
|||
|
|
```sh
|
|||
|
|
git clone https://github.com/netease-youdao/Confucius3-Math.git
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
使用$VS$ $Code$将 $run\_service\_stream.sh$ 文件的换行符从CRLF转换为LF
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
**2. 下载模型（约$29.56$ $GB$）**
|
|||
|
|
|
|||
|
|
```sh
|
|||
|
|
# 从魔搭社区下载
|
|||
|
|
pip install modelscope
|
|||
|
|
mkdir netease-youdao/Confucius3-Math -p
|
|||
|
|
modelscope download --model netease-youdao/Confucius3-Math --local_dir ./netease-youdao/Confucius3-Math
|
|||
|
|
|
|||
|
|
# 备用方案【从huggingface下载】
|
|||
|
|
huggingface-cli download netease-youdao/Confucius3-Math --local-dir ./models
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
**3. 从嘟嘟鸟下载$Docker$镜像**
|
|||
|
|
|
|||
|
|
```cmd
|
|||
|
|
docker pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvidia/cuda:12.2.2-cudnn8-devel-ubuntu22.04
|
|||
|
|
docker tag swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvidia/cuda:12.2.2-cudnn8-devel-ubuntu22.04 docker.io/nvidia/cuda:12.2.2-cudnn8-devel-ubuntu22.04
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
**4. 构建 $Docker$镜像**
|
|||
|
|
|
|||
|
|
```
|
|||
|
|
docker build -t confucius3 .
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
**5.启动**
|
|||
|
|
|
|||
|
|
```cmd
|
|||
|
|
docker run --gpus all -e ARK_API_KEY=f6150e6c-422a-4265-8b63-4d941b271220 -p 8827:8827 confucius3
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
**6. 访问办法**
|
|||
|
|
|
|||
|
|
```http
|
|||
|
|
http://127.0.0.1:8827/
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
### 三、相关资料
|
|||
|
|
|
|||
|
|
一手实测有道14B「子曰3」数学模型，击败满血版DeepSeek R1
|
|||
|
|
|
|||
|
|
http://124.220.5.172/archives/51312
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
对话｜有道“子曰3”数学模型开源背后
|
|||
|
|
|
|||
|
|
https://news.qq.com/rain/a/20250711A08MQK00?suid=&media_id=
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
首先，你只需要定义系统提示词和用户提示词模板：
|
|||
|
|
|
|||
|
|
SYSTEM_PROMPT_TEMPLATE = """A conversation between User and Assistant. The user asks a question, and the Assistant solves it. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <think> </think> and <answer> </answer> tags, respectively, i.e., <think> reasoning process here </think> <answer> answer here </answer>."""
|
|||
|
|
|
|||
|
|
USER_PROMPT_TEMPLATE = """{question}"""
|
|||
|
|
|
|||
|
|
|
|||
|
|
用户和助手之间的对话。用户提出问题，助手会解决问题。助手首先在脑海中思考推理过程，然后为用户提供答案。推理过程和答案分别包含在标签<answer></answer>中<think></think>，即 <think> 推理过程在这里 </think> <answer> 回答这里</answer>。
|
|||
|
|
|
|||
|
|
DEMO地址
|
|||
|
|
|
|||
|
|
https://confucius.youdao.com/
|