Using ML models within Lucenia
To integrate machine learning (ML) models into your Lucenia cluster, you can upload and serve them locally. Choose one of the following options:
-
Pretrained models provided by Lucenia: To learn more, see Lucenia-provided pretrained models. For a list of supported models, see Supported pretrained models.
-
Custom models such as PyTorch deep learning models: To learn more, see Custom models.
Running local models on the CentOS 7 operating system is not supported. Moreover, not all local models can run on all hardware and operating systems.
macOS x86-64 (Intel Mac) is not supported. Local models depend on a native PyTorch library, and PyTorch dropped macOS x86-64 support after version 2.2 — this is an upstream PyTorch limitation, not Lucenia's. On an Intel Mac, deploying a local model fails with an error similar to No PyTorch native library matches your operating system: cpu-osx-x86_64. Linux and Apple Silicon (macOS arm64) are unaffected. If you're on an Intel Mac, use an externally hosted model instead.
GPU acceleration
For better performance, you can take advantage of GPU acceleration on your ML node. For more information, see GPU acceleration.