Hugging Face presents catalog for checking models on AWS Trainium

Hugging Face presented the torch-neuronx model support Space, which checks which popular Hub models run unmodified on AWS Trainium. The panel shows the number of cores used, total runtime — including compilation —, peak memory and the test result, while also providing the script and packages needed to reproduce it on a Trainium2 instance. The data is also available in the public aws-neuron/torch-neuronx-model-support dataset.

At present, testing covers only Transformers models for text generation and image-text-to-text tasks that fit on up to four Trainium2 devices, or 16 cores. Hugging Face plans to add diffusers, sentence transformers and other tasks. A “PASS” result confirms execution and an output check, but does not measure throughput or longer generations.