Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Sign up
Appearance settings

TsingmaoAI/deepseek-amd

Folders and files

NameName
Last commit message
Last commit date

Latest commit

History

4 Commits

Repository files navigation

一键部署教程 🚀(推荐方式)

推荐用户优先使用此方式完成部署,操作简单、无需命令行基础。

  1. 解压 aipc-ds.zip 压缩包,运行其中的 setup 安装软件主体;
  2. 可自行选择安装位置;
  3. 在运行前务必将模型权重所有文件夹放入安装目录下的 AIPC-DS\ds-amd\_internal\models 文件夹中;
  4. 双击运行桌面图标或 AIPC.exe 启动程序。

📦 安装包下载链接(通过百度网盘分享):
点击下载 aipc-ds.zip 提取码: ecxk

Introduction

在AMD Ryzen系列芯片上部署deepseek模型,支持的模型列表如下:

Supported computer configuration

仅支持运行Windows 11的Strix (STX)和Krackan Point (KRK)处理器。

Requirements

  • NPU驱动,可参考NPU

  • iGPU驱动,可参考iGPU

  • 下载必要的.wheel文件与.dll文件,下载链接:wheel,下载并解压完成后将其中的wheel和dll文件拷贝到wheel目录下

Execution using Python

环境准备

  1. 创建conda环境
conda create --name <env name> python=3.10
  1. 激活conda环境
conda activate <env name>
  1. 安装wheel文件
cd wheel
pip install onnxruntime_genai-0.4.0.dev0-cp310-cp310-win_amd64.whl
pip install onnxruntime_directml-1.20.1-cp310-cp310-win_amd64.whl
  1. 安装requirements
pip install -i requirements.txt

模型准备

  1. huggingface下载所需的模型,并将模型拷贝到models目录下

  2. 打开genai_config.json文件。位于已下载模型文件夹中的文件。使用位于wheel文件夹中的onnx_custom_ops.dll的完整路径更新custom_ops_library的值.

"session_options": {
 ...
 "custom_ops_library":"wheel\\onnx_custom_ops.dll",
 ...
}

运行模型

方式1(非接口访问)

python run_model.py --model_dir path_to\your\model

方式2(接口访问)

简单开始

修改server.py中的models_paths值为你的模型路径。

models_paths = {"DeepSeek-R1-Distill-Qwen-1.5B":"path\\to\\DeepSeek-R1-Distill-Qwen-1.5B-awq-asym-uint4-g128-lmhead-onnx-hybrid ",
 "DeepSeek-R1-Distill-Llama-8B":"path\\to\\DeepSeek-R1-Distill-Llama-8B-awq-asym-uint4-g128-lmhead-onnx-hybrid",
 "DeepSeek-R1-Distill-Qwen-7B":"path\\to\\DeepSeek-R1-Distill-Qwen-7B-awq-asym-uint4-g128-lmhead-onnx-hybrid"
 }

开启接口服务

python server.py --port 9090

API Endpoints

Get /health : Returns heath check result
  • Body : {"status": "ok" }
  • the model is successfully loaded and the server is ready.
Post /v1/chat/completions: Given a prompt,it returns the result of reasoning

example :

{"input":"Please solve following problem and explain it to me. Then give me final answer at the end with a single number preceded by string '#### '. Question: Rory orders 2 subs for 7ドル.50 each, 2 bags of chips for 1ドル.50 each and 2 cookies for 1ドル.00 each for delivery.\nAnswer:"}
Post /change_model: change the servering model

example:

{
 "model_name":"DeepSeek-R1-Distill-Qwen-7B"
}

About

Deploy the deepseek model on AMD Ryzen series chips

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

Contributors

Languages

AltStyle によって変換されたページ (->オリジナル) /