Support empty mixture weights (default to lengths)

[shuheng-liu]

What this does

Support empty mixture weights (default to lengths)

How it was tested

Ran tests without specifying mixture weights.

How to checkout & try? (for the reviewer)

Run tests without specifying mixture weights.

Checklist

• I have added Google-style docstrings to important functions and ensured function parameters are typed.
• My PR includes policy-related changes.
  • If the above is checked: I have run the GPU pytests (pytest -m "gpu") and regression tests.

[Copilot]

Pull request overview

Adds support for omitting dataset mixture weights by inferring defaults from dataset lengths, allowing configs to leave weights unset while still producing a valid WeightedDatasetMixture.

Changes:

• Make DatasetMixtureConfig.weights optional (None means infer at runtime).
• Infer train/val mixture weights from float(len(dataset)) when weights are None.
• Document the new optional-weights behavior and add a config-level test for weights=None.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/configs/test_default.py	Adds a test ensuring DatasetMixtureConfig accepts weights=None.
src/opentau/datasets/factory.py	Infers mixture weights from dataset lengths when weights is None (train and val).
src/opentau/configs/default.py	Updates config typing/docs and relaxes validation to allow weights=None.
docs/source/tutorials/datasets.rst	Documents that weights can be omitted/null and will be inferred from lengths.
docs/source/concepts.rst	Notes default-to-lengths behavior when weights are omitted/null.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.