[AUTO-MERGE] Handle None values in _inner_loop tensor conversion

[Copilot]

Change Description

_inner_loop crashes when prepare_inputs returns None values, which occurs in supervised models with missing labels. The function unconditionally calls torch.from_numpy() on all values, including None.

Solution Description

Added None checks before tensor conversion in _inner_loop:

# Before
if isinstance(batch, tuple):
    batch = tuple(torch.from_numpy(i).to(device) for i in batch)
else:
    batch = torch.from_numpy(batch).to(device)

# After  
if isinstance(batch, tuple):
    batch = tuple(torch.from_numpy(i).to(device) if i is not None else None for i in batch)
elif batch is not None:
    batch = torch.from_numpy(batch).to(device)

Added test coverage for tuple with None element, None batch, and regression cases.

Code Quality

• I have read the Contribution Guide and agree to the Code of Conduct
• My code follows the code style of this project
• My code builds (or compiles) cleanly without any errors or warnings
• My code contains relevant comments and necessary documentation

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• www.cs.toronto.edu
  • Triggering command: /usr/bin/python /usr/bin/python -m ipykernel_launcher -f /tmp/tmpu0iebvur.json --HistoryManager.hist_file=:memory: --from without_precedin--columns=500 --to rst --columns=500 pandoc --eo�� lf --from ve_divs+raw_html--columns=500 --to rst --columns=500 pandoc (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

Original prompt

---

This section details on the original issue you should resolve

<issue_title>If prepare_data returns a None, pytorch_ignite will error</issue_title>
<issue_description>Bug report
In _inner_loop, when we try to convert numpy to Tensors, if there is a None that is returned from prepare_data, the conversion will fail.

Max ran into this with a supervised model that used a prepare_data function that returns None when there are no labels. Changing this to return an empty array instead of None fixed the problem, but we should be more accommodating.
</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes If prepare_data returns a None, pytorch_ignite will error #688

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[mtauraso]

This seems good to merge.

[Copilot]

Pull request overview

This PR fixes a crash in Hyrax’s PyTorch Ignite integration when prepare_inputs returns None values (e.g., missing labels in supervised workflows) by skipping tensor conversion for None and adds regression tests to cover the new behavior.

Changes:

• Update _inner_loop to avoid calling torch.from_numpy() on None values (both for tuple batches and single batches).
• Add unit tests covering tuple-with-None, None batch, and standard conversion behavior.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
src/hyrax/pytorch_ignite.py	Adds None-aware tensor conversion logic in _inner_loop to prevent torch.from_numpy(None) crashes.
tests/hyrax/test_pytorch_ignite.py	Adds targeted tests validating _inner_loop behavior with None inputs and ensuring existing conversion paths still work.

[drewoldag]

This appears correct to me. It adheres to assumption that what is returned from prepare_inputs is a tuple or a bare array, and seems to handle those cases just fine.

The tests seem to cover the various permutations.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 64.17%. Comparing base (84fe8b0) to head (0a8251d).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #689      +/-   ##
==========================================
+ Coverage   64.15%   64.17%   +0.02%     
==========================================
  Files          61       61              
  Lines        5989     5990       +1     
==========================================
+ Hits         3842     3844       +2     
+ Misses       2147     2146       -1     

☔ View full report in Codecov by Sentry.
📢 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.

[github-actions]

Before [84fe8b0]	After [7e64b43]	Ratio	Benchmark (Parameter)
failed	failed	n/a	data_cache_benchmarks.DataCacheBenchmarks.time_preload_cache_hsc1k
failed	failed	n/a	data_cache_benchmarks.DataCacheBenchmarks.track_cache_hsc1k_hyrax_size_undercount
failed	failed	n/a	data_request_benchmarks.DatasetRequestBenchmarks.time_request_all_data
1.98±0.02s	2.06±0.03s	1.04	benchmarks.time_infer_help
1.98±0.02s	2.06±0.05s	1.04	benchmarks.time_lookup_help
1.96±0.02s	2.03±0.01s	1.03	benchmarks.time_help
1.97±0.02s	2.02±0.03s	1.03	benchmarks.time_prepare_help
282±3ms	289±3ms	1.02	benchmarks.time_import
38.9±0.4ms	39.6±0.7ms	1.02	benchmarks.time_nb_obj_dir
9.62±0.1ms	9.83±0.05ms	1.02	vector_db_benchmarks.VectorDBSearchBenchmarks.time_search_by_vector_many_shards(128, 'chromadb')

Click here to view all benchmarks.