This export organizes CMedTEB into retrieval, rerank, and synonym STS tasks.
shared_train/retrieval_rerank_train.jsonl: one shared 20,000-row train split for both retrieval and rerank.
retrieval/corpus.jsonl: retrieval corpus.
retrieval/test_queries.jsonl: retrieval test queries.
retrieval/test_qrels.jsonl: retrieval qrels.
rerank/test.jsonl: rerank test set.
sts/train.jsonl: synonym STS train set.
sts/test.jsonl: synonym STS test set.… See the full description on the dataset page:
https://huggingface.co/datasets/PhilipGAQ/CMedTEB.
