Precomputed document-side expansions for the
kasys/ReCaRe benchmark,
distributed as a reproducibility artifact for the ReCaRe CIKM 2026 Resource
paper. These files let third parties reproduce the document-expansion
augmentation results without re-running the LLM / doc2query generation
(which requires Azure OpenAI access and a doc2query model).
The companion query-side expansions (q2d_zs, q2e_zs, task_q2e) are… See the full description on the dataset page:
https://huggingface.co/datasets/kasys/ReCaRe-expansions.