Real Time Inference Model dynamic

Hello! I have a question with respect to a behavior of python library to interact with
Specifically, I want to be able to download a specific pytorch model from the model registry of to do real-time inference. right now I can’t instantiate the MNIST trial class. when I do the checkpoint download I get back a pickle file and a .pth file , being that I have to load the model, and saves me under checkpoint/uuid_checkpoint, and I try to get the pickle or .pth file , it gives me : ValueError: Automatic import logic failed to import Trial class as main:MNistTrial. You will need to provide your Trial class via the trial_class argument intead.

but the trial class is static, in a real context this class I need dynamic, how do I do that?

You will probably get better luck posting this on the determined github issues GitHub - determined-ai/determined: Determined is an open-source machine learning platform that simplifies distributed training, hyperparameter tuning, experiment tracking, and resource management. Works with PyTorch and TensorFlow.