This document is relevant for: Inf1, Inf2, Trn1, Trn2, Trn3
Neuron Containers#
This section contains the technical documentation for using AWS Neuron Deep Learning Containers (DLCs) and containerized deployments on Inferentia and Trainium instances.
What are Neuron Deep Learning Containers?#
AWS Neuron Deep Learning Containers (DLCs) are a set of pre-configured Docker images for training and serving models on AWS Trainium and Inferentia instances using the AWS Neuron SDK. Each DLC is optimized for specific ML frameworks and comes with all Neuron components pre-installed, enabling you to quickly deploy containerized workloads without manual setup.
With Neuron DLCs, developers can:
Deploy production-ready containers with pre-installed Neuron SDK and ML frameworks
Use containers across multiple deployment platforms including EC2, EKS, ECS, and SageMaker
Customize DLCs to fit specific project requirements
Leverage Neuron plugins for better observability and fault tolerance
Run distributed training and inference workloads with vLLM integration
Schedule MPI jobs on Trn2 UltraServers for improved performance
Neuron DLCs support popular ML frameworks including PyTorch, TensorFlow, and JAX, and are available for both training and inference workloads on Inf1, Inf2, Trn1, Trn1n, and Trn2 instances.
Neuron DRA for Kubernetes
Neuron has released support for Dynamic Resource Allocation (DRA) with Kubernetes. Read more about it here.
Quickstarts#
Neuron Containers Documentation#
This document is relevant for: Inf1, Inf2, Trn1, Trn2, Trn3