第 1 章 创建模型镜像并将其推送到容器镜像 registry

流程

1. 创建 Python 虚拟环境并安装 huggingface_hub Python 库：

   python3 -m venv venv && \
   source venv/bin/activate && \
   pip install --upgrade pip && \
   pip install huggingface_hub

2. 创建模型下载程序 Python 脚本：

   vi download_model.py

3. 在 download_model.py 文件中添加以下内容，根据需要调整 model_repo 的值：

   from huggingface_hub import snapshot_download
   
   # Specify the Hugging Face repository containing the model
   model_repo = "ibm-granite/granite-3.1-2b-instruct"
   snapshot_download(
       repo_id=model_repo,
       local_dir="/models",
       allow_patterns=["*.safetensors", "*.json", "*.txt"],
   )

4. 为 modelcar 创建 Dockerfile ：

   FROM registry.access.redhat.com/ubi9/python-311:latest as base
   
   USER root
   
   RUN pip install huggingface-hub
   
   # Download the model file from Hugging Face
   COPY download_model.py .
   
   RUN python download_model.py
   
   # Final image containing only the essential model files
   FROM registry.access.redhat.com/ubi9/ubi-micro:9.4
   
   # Copy the model files from the base container
   COPY --from=base /models /models
   
   USER 1001

5. 构建 modelcar 镜像：

   podman build . -t modelcar-example:latest --platform linux/amd64

   输出示例

   Successfully tagged localhost/modelcar-example:latest

6. 将 modelcar 镜像推送到容器 registry。例如：

   $ podman push modelcar-example:latest quay.io/<your_model_registry>/modelcar-example:latest

   输出示例

   Getting image source signatures
   Copying blob b2ed7134f853 done
   Copying config 4afd393610 done
   Writing manifest to image destination
   Storing signatures