TS2Vec and demo notebook

[drewoldag]

Adding ts2vec, and the demo notebook.

This allows for an end to end example of using LSDB to load a HATS catalog with nested light curve data, sending the catalog promise to Hyrax for training and inference with train_stream and infer_stream, then reducing dimensions and plotting a 2d embedding.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[codecov]

Codecov Report

❌ Patch coverage is 86.89459% with 46 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.81%. Comparing base (1281289) to head (a4ac2eb).

Files with missing lines	Patch %	Lines
src/hyrax/models/hyrax_ts2vec.py	86.43%	46 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1055      +/-   ##
==========================================
+ Coverage   70.20%   70.81%   +0.61%     
==========================================
  Files          91       92       +1     
  Lines        8961     9306     +345     
==========================================
+ Hits         6291     6590     +299     
- Misses       2670     2716      +46     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Copilot review overview

🟡 Changes recommended

Critical correctness and distributed-training issues remain unresolved.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 4 · 1

Open (5)

• Padding mask is lost through convolution and pooling · New
• DDP wrapper is bypassed in the contrastive training path · New
• Inference ignores the training runtime configuration · New
• Runtime config is missing in distributed workers · New
• Non-integer sequence lengths are silently truncated · New

What changed in this PR

Adds TS2Vec support for ragged LSDB light-curve streams and an end-to-end training/inference demo.

Changes:

• Adds TS2Vec encoding, normalization, masking, contrastive training, and inference.
• Adds configuration defaults, model registration, context propagation, and ONNX preprocessing.
• Preserves nested-column dtypes during LSDB collation and expands tests.

Final findings include critical masking, DDP, deployment configuration, and distributed context issues, plus moderate mask-device and sequence-length validation issues.

File	Summary
tests/​hyrax/​test_lsdb_stream_dataset.py	Tests dtype-preserving collation.
tests/​hyrax/​test_hyrax_ts2vec.py	Tests TS2Vec behavior.
tests/​hyrax/​test_context.py	Tests configuration context propagation.
src/​hyrax/​verbs/​verb_registry.py	Adds runtime configuration to verb contexts.
src/​hyrax/​verbs/​to_onnx.py	Uses training configuration during export.
src/​hyrax/​models/​hyrax_ts2vec.py	Implements TS2Vec and light-curve encoding.
src/​hyrax/​models/​__init__.py	Registers the model.
src/​hyrax/​hyrax_default_config.toml	Defines TS2Vec defaults.
src/​hyrax/​datasets/​lsdb_stream_dataset.py	Preserves nested array dtypes.
src/​hyrax/​context.py	Documents runtime configuration context.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.