North Mini Code home pagelight logo
  1. Deployment
  2. AWS Inferentia & Trainium

Hugging Face’s deploy UI lists AWS Inferentia & Trainium as a deployment option for North Mini Code.

The concrete steps here are inferred from Hugging Face’s Optimum Neuron and SageMaker docs rather than a North-specific model card snippet. Hugging Face documents Optimum Neuron as the interface between Transformers and AWS Inferentia / Trainium.

1

Validate that this hardware path matches your deployment target

Choose this route when you want AWS accelerator-specific inference or training on Inferentia2 or Trainium1, rather than a generic CPU or GPU endpoint.

2

Start from Optimum Neuron or a Neuron-enabled DLC

python -m pip install --upgrade-strategy eager "optimum[neuronx]"
3

Compile or export the model for Neuron

Hugging Face’s Optimum Neuron docs cover export, pipelines, and SageMaker deployment flows for Neuron hardware.

4

Deploy through SageMaker or directly on Neuron-enabled infrastructure

Hugging Face’s AWS docs also document Neuron-capable DLCs for SageMaker, EC2, ECS, and EKS paths.

Validate model architecture support before you commit to this route. Hugging Face documents the Neuron toolchain generically, but does not currently publish a North Mini Code-specific launch command the way it does for vLLM and SGLang.

Optimum Neuron overview

High-level entry point for AWS accelerator support in the Hugging Face ecosystem.

Optimum Neuron on SageMaker

SageMaker path for Inferentia2 and Trainium deployments.

Neuron export guide

How to export models into serialized Neuron-optimized artifacts.

Neuron pipeline guide

Pipeline-level usage on Inf2 and Trn1.

Available AWS DLCs

Overview of AWS containers, including Neuron-related paths.

HF on AWS overview

Broader Hugging Face AWS deployment options across SageMaker, EC2, ECS, and EKS.