[xblwh]

rerank_pairs() accepts an iterable of query/document pairs, but currently wraps every tuple as a single pair. Passing ((query, doc1), (query, doc2)), a one-element tuple, or an empty tuple raises a tokenizer TypeError.

Only wrap a tuple containing two strings as the existing single-pair shorthand. Treat tuple batches like lists when choosing the small-batch path. The regression tests cover 0–3 pairs with sequential and parallel execution, plus the existing single-pair behavior, using the real MiniLM reranker and canonical scores.

Validation

• The eight tuple-batch cases fail on unchanged main; both single-pair cases already pass.
• pytest tests/test_text_cross_encoder.py tests/test_common.py -q: 19 passed, with both current compatible dependencies and the Python 3.12 versions selected from poetry.lock.
• With the lockfile versions: mypy fastembed --disallow-incomplete-defs --disallow-untyped-defs --disable-error-code=import-untyped passes for all 64 source files; pyright tests/type_stub.py and ruff check fastembed tests pass.
• The repository's pre-commit hooks pass for both changed files; git diff --check passes.

The full model suite and other OS/Python combinations were not run. Broader Ruff checks report the same existing formatting and experiment-notebook lint findings on unchanged main; resolving the newest dependencies also exposes five existing mypy errors outside this change. These are unchanged by this patch.

Prepared with AI assistance from Codex.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: 90fb066a-825d-40db-be60-35c5d79de0e8

📥 Commits

Reviewing files that changed from the base of the PR and between 0dab99c and 0d07193.

📒 Files selected for processing (2)

• fastembed/rerank/cross_encoder/onnx_text_model.py
• tests/test_text_cross_encoder.py

Included review availability: Your plan provides up to 8 included reviews per hour; 7 remain after this review.

---

📝 Walkthrough

Walkthrough

_rerank_pairs now treats a two-string tuple as one pair and accepts tuples of pairs for batch processing. Tests cover empty, single, and multiple tuple-of-pairs inputs with sequential and parallel execution. Tests also verify direct single-pair tuples.

Priority: ⬇️ Low

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix

Merge Risk: ⚪ Minimal · up to 0d071

The tuple-batch behavior is covered across supported input sizes and execution modes, with no established current-head defect.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 2 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description clearly explains the tuple-batch fix, preserved single-pair behavior, regression tests, and validation results. It is directly related to the changeset.
Title check	✅ Passed	The title clearly and concisely identifies the main change: accepting tuple batches in rerank_pairs.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[joein]

Hey, thanks for pointing it out.

To be honest, the methods signature are a bit broader than what we actually support. We support lists, iterables and generators, all the other types are supported indirectly, and this particular fix looks somewhat ugly, since the condition is so long and checks the length with a magic (maybe not super magic) integer.
I'll close the PR and issue for now, we might come back to it later when we'll do a proper refactoring.