
AWS Neuron 2.13 now supports training and inference for Llama 2, GPT-NeoX, and SDXL generative AI models on EC2 Inferentia and Trainium instances with optimized performance.


<div><p>This article announces AWS Neuron 2.13 release, expanding support for generative AI models on AWS's purpose-built instances.</p><ul><li>Adds Llama 2 and GPT-NeoX model training support</li><li>Enables Stable Diffusion XL and CLIP model inference</li><li>Integrates with PyTorch and TensorFlow frameworks</li><li>Includes compiler, runtime, profiling tools, and libraries</li><li>Supports distributed LLM training via Neuron Reference for Nemo Megatron</li><li>Optimized LLM inference for Llama 2 with Transformers Neuron</li><li>Available on Trn1 and Inf2 instances in three US regions</li></ul><p>AWS Neuron 2.13 expands generative AI model support with minimal code changes required for popular ML frameworks.</p></div>


Related articles