Third-party software used in this project:

----------------------------------------------------------------------
RepoQA — long-context code understanding benchmark
----------------------------------------------------------------------
Source: https://github.com/evalplus/repoqa
License: Apache License 2.0
Citation:
  Liu, J. et al. RepoQA: Evaluating Long Context Code Understanding.
  arXiv:2406.06025, June 2024.

Used as the reference benchmark for retrieval-quality evaluation in
benchmarks/benchmarks/eval/datasets/repoqa.py. No source code from
RepoQA is vendored; only its dataset is consumed via huggingface
datasets at runtime.

----------------------------------------------------------------------
MLflow — experiment tracking
----------------------------------------------------------------------
Source: https://github.com/mlflow/mlflow
License: Apache License 2.0

Optional dependency (uv pip install -e benchmarks[mlflow]). Used as
the default experiment tracking backend.
