--- license: apache-2.0 --- # AWS Neuron optimum model cache This repository is not obsolete: it contains compilation artifacts for the legacy Neuron SDK (<2.33), to be used with `optimum-neuron`. From November 2026 on, the deployment of hub models on Trainium devices is natively supported in `transformers`, `diffusers` and `sentence-transformers` through `torch-neuronx`. Please refer to the [model support dashboard](https://huggingface.co/spaces/aws-neuron/torch-neuronx-model-support) to check if a model is supported.