AWS Deep Learning Containers Samples

This repository contains sample code and configurations for using AWS Deep Learning Containers (DLCs) in various scenarios. AWS Deep Learning Containers are Docker images pre-installed with deep learning frameworks and tools, optimized for performance on AWS infrastructure.

Repository Structure

• vllm-samples/: Samples for deploying vLLM (a high-throughput serving engine for LLMs) using AWS Deep Learning Containers

  • deepseek/: Samples for deploying DeepSeek models
    • eks/: Configuration files and instructions for deploying DeepSeek models on Amazon EKS with GPU support, EFA, and FSx Lustre integration

• mlflow/: Samples for using SageMaker managed MLflow with Deep Learning Containers and Deep Learning AMIs

  • dlc-with-mlflow/: Sample for integrating AWS DLCs with SageMaker managed MLflow for training. See README for detailed instructions.

AWS Deep Learning Containers

AWS Deep Learning Containers provide optimized environments with pre-installed deep learning frameworks and tools:

• Performance Optimized: Tuned for maximum performance on AWS infrastructure
• Pre-configured: Ready-to-use environments with popular frameworks
• Regularly Updated: Latest versions of frameworks and security patches
• AWS Integration: Seamless integration with AWS services like EKS, ECS, and SageMaker

Learn more about AWS Deep Learning Containers.