Support deterministic experiments

Is it possible to support deterministic experiments to ensure full reproducibility?

Something like this: https://lightning.ai/docs/pytorch/stable/common/trainer.html#reproducibility

No files, tests, or entry points are named. Start by reviewing the linked PyTorch Lightning reproducibility guidance and the repository's experiment and training entry points to define the required scope. Done should mean that repeated experiments can be reproduced deterministically, with the behavior verified by an appropriate test or example.

Written by the indexing model from the issue text.