diff --git a/NanoCMedQAv2reranking/corpus/test.parquet b/NanoCMedQAv2reranking/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..b44f3b9bc45e272f53cfa63c1476942d46c424f6 --- /dev/null +++ b/NanoCMedQAv2reranking/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f784619f7bc371e74f9989a22fb7fcb069b8790eccb8693b70671e230ab0575 +size 2011108 diff --git a/NanoCMedQAv2reranking/metadata/test.json b/NanoCMedQAv2reranking/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..705b6fa87c14dfda7ed18ee5f4b6058d05c18bba --- /dev/null +++ b/NanoCMedQAv2reranking/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoCMedQAv2reranking", + "source_task": "CMedQAv2-reranking", + "source_type": "Reranking", + "source_dataset": { + "path": "mteb/CMedQAv2-reranking", + "revision": "01399b6956a4c6bf9edc5b6b90e176aeeb399686" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 377, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 377, + "hits": 377, + "recall": 1.0 + }, + "ndcg_at_10": 0.15003900776298235, + "ndcg_at_100": 0.29903015701374225, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 1000, + "source_corpus_count": 100000, + "source_qrels_query_count": 1000, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 406, + "qrels_rewritten_for_duplicate_text": 7, + "forced_queries": 148, + "forced_doc_count": 258, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoCMedQAv2reranking/qrels/test.parquet b/NanoCMedQAv2reranking/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..f5df10b6d60819c49d44364a18973804b3efc72e --- /dev/null +++ b/NanoCMedQAv2reranking/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c64481d6d50cc517b57f2a14ae5d586bbc32acd7a5491b88af7df4f8e4325949 +size 5039 diff --git a/NanoCMedQAv2reranking/queries/test.parquet b/NanoCMedQAv2reranking/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..9f6f216dc75a3c55b532d6602aef1e448b346fcb --- /dev/null +++ b/NanoCMedQAv2reranking/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63be23fe80080e84fcc55f361a1af6c88049fc2481fc41297ce2a62a69e4c7d8 +size 23408 diff --git a/NanoCUREv1/corpus/test.parquet b/NanoCUREv1/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..373bf28b5dd1ace9f7a38fbc2210e89784602268 --- /dev/null +++ b/NanoCUREv1/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:131808d5d193adff2adae8a44d5c0e1428321bb6a69e98a9445d65e44b132924 +size 3718624 diff --git a/NanoCUREv1/metadata/test.json b/NanoCUREv1/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..7943d1233834ffdba912181ce10c91ba8cbb613f --- /dev/null +++ b/NanoCUREv1/metadata/test.json @@ -0,0 +1,58 @@ +{ + "split": "NanoCUREv1", + "source_task": "CUREv1", + "source_type": "Retrieval", + "source_dataset": { + "path": "clinia/CUREv1", + "revision": "3bcf51c91e04d04a8a3329dfbe988b964c5cbe83" + }, + "source_eval_splits": [ + "all", + "dentistry_and_oral_health", + "dermatology", + "gastroenterology", + "genetics", + "neuroscience_and_neurology", + "orthopedic_surgery", + "otorhinolaryngology", + "plastic_surgery", + "psychiatry_and_psychology", + "pulmonology" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 5181, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 5181, + "hits": 5181, + "recall": 1.0 + }, + "ndcg_at_10": 0.3698959661937695, + "ndcg_at_100": 0.6160108835808653, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "all", + "source_query_count": 2000, + "source_corpus_count": 244600, + "source_qrels_query_count": 2000, + "source_queries_with_more_than_bm25_top_k_positives": 18, + "capped_queries_with_more_than_bm25_top_k_positives": 18, + "dropped_positive_qrels_by_capping": 2785, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 171, + "forced_doc_count": 2977, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoCUREv1/qrels/test.parquet b/NanoCUREv1/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..8eae7d7c42c9d7d9abdefc132df47d62e42635f2 --- /dev/null +++ b/NanoCUREv1/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f457486659d666bb9bfad0179f154ef15e6b654620b7cf5fb63396134b64fde +size 119813 diff --git a/NanoCUREv1/queries/test.parquet b/NanoCUREv1/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..c8d7e4727dfbe4869f35299ddf5e88e57da43c04 --- /dev/null +++ b/NanoCUREv1/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9840e96888d4256118d2b954833595571afe5ef7938c1c80004ada2635e763f3 +size 17227 diff --git a/NanoCmedqa/corpus/test.parquet b/NanoCmedqa/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..edce9985a6de4605d614ca16c23d0119ab53664e --- /dev/null +++ b/NanoCmedqa/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28334e6020a33a2afa0e886d105aae9cb932fad8d873bb4fce19ee36b4c41d74 +size 3259518 diff --git a/NanoCmedqa/metadata/test.json b/NanoCmedqa/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..a6d8635636030ff1bfc3b13896cf087a7ef89917 --- /dev/null +++ b/NanoCmedqa/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoCmedqa", + "source_task": "CmedqaRetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/CmedqaRetrieval", + "revision": "c476f85bf03d6642ec66bf54b9a551c88108bbb4" + }, + "source_eval_splits": [ + "dev" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 324, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 324, + "hits": 324, + "recall": 1.0 + }, + "ndcg_at_10": 0.16678868356587398, + "ndcg_at_100": 0.30693075139953757, + "source_container": "task.dataset", + "source_eval_split_used": "dev", + "source_query_count": 3999, + "source_corpus_count": 100001, + "source_qrels_query_count": 3999, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 134, + "forced_doc_count": 202, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoCmedqa/qrels/test.parquet b/NanoCmedqa/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..6ffad5e83ecd37173ffaac4255d4bccce1a60c46 --- /dev/null +++ b/NanoCmedqa/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e71c742a0d0f1073e5cb66afa16f624fe79536f138be5b3c8e698ac1ebd96cdd +size 19624 diff --git a/NanoCmedqa/queries/test.parquet b/NanoCmedqa/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..a2404ddacd90c63f06040820b909a4cd9b296875 --- /dev/null +++ b/NanoCmedqa/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4d9738887ad6d87f5863d3dfb35ae6d318a7e8000e8075968bd68e29bdc7355 +size 30084 diff --git a/NanoMedicalQA/corpus/test.parquet b/NanoMedicalQA/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..cf9816fdf460c497725c706e0294da3d3bf74052 --- /dev/null +++ b/NanoMedicalQA/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:524d587d8dc23bf3898c9a576df69d3aab04d0c172d9b3833f61cbdb5a6a70d3 +size 1169252 diff --git a/NanoMedicalQA/metadata/test.json b/NanoMedicalQA/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..b425166ad063a0d891637060a18136665b04af2a --- /dev/null +++ b/NanoMedicalQA/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoMedicalQA", + "source_task": "MedicalQARetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/medical_qa", + "revision": "ae763399273d8b20506b80cf6f6f9a31a6a2b238" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 2007, + "qrels": 200, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 200, + "hits": 200, + "recall": 1.0 + }, + "ndcg_at_10": 0.4735533646828049, + "ndcg_at_100": 0.5241369411712118, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 2048, + "source_corpus_count": 2048, + "source_qrels_query_count": 2048, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 7, + "duplicate_doc_texts_removed": 42, + "qrels_rewritten_for_duplicate_text": 1, + "forced_queries": 41, + "forced_doc_count": 41, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoMedicalQA/qrels/test.parquet b/NanoMedicalQA/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..830bcf67b95808eda0fce7d0e77660a83c5d97be --- /dev/null +++ b/NanoMedicalQA/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69445fbcad8dedae89b2281c5fd2f35b6d161b5440af28d7f3fc99b82f06e602 +size 16805 diff --git a/NanoMedicalQA/queries/test.parquet b/NanoMedicalQA/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..e068022421119e564815a55601b413cbeef95e6d --- /dev/null +++ b/NanoMedicalQA/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ff4a969992932ce1603fd7200053606d9bbf9db3cc1299bdd5ac12461284eba +size 12403 diff --git a/NanoNFCorpus/corpus/test.parquet b/NanoNFCorpus/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..ab7379e2c523f13a94f0a796b298dff21fd4e819 --- /dev/null +++ b/NanoNFCorpus/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1f90a0ec952f85e1f0578789f0271114685aeca3e82b4809cfbc1406513d513 +size 3094743 diff --git a/NanoNFCorpus/metadata/test.json b/NanoNFCorpus/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..14833cf96d4884764d8be053352b65c65c080274 --- /dev/null +++ b/NanoNFCorpus/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoNFCorpus", + "source_task": "NFCorpus", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/nfcorpus", + "revision": "ec0fa4fe99da2ff19ca1214b7966684033a58814" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 3593, + "qrels": 3718, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 3718, + "hits": 3718, + "recall": 1.0 + }, + "ndcg_at_10": 0.24341837520694568, + "ndcg_at_100": 0.49341846897457736, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 323, + "source_corpus_count": 3633, + "source_qrels_query_count": 323, + "source_queries_with_more_than_bm25_top_k_positives": 22, + "capped_queries_with_more_than_bm25_top_k_positives": 22, + "dropped_positive_qrels_by_capping": 5753, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 65, + "qrels_rewritten_for_duplicate_text": 25, + "forced_queries": 185, + "forced_doc_count": 3029, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoNFCorpus/qrels/test.parquet b/NanoNFCorpus/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..2e8f53ef4bbe9e6c026c64bb5d93f263af6fdeaf --- /dev/null +++ b/NanoNFCorpus/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60491f056b34eea0fecf2aabcb6c368250a8702983efb3b21b1debc348ebeddf +size 17226 diff --git a/NanoNFCorpus/queries/test.parquet b/NanoNFCorpus/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..4a6fdc947363a0e280f0eb6e1b222eb4410ddbd7 --- /dev/null +++ b/NanoNFCorpus/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32b2e06eab0915bdc46487d383448b451d65775a37f0135960d755d3fa6ca1b3 +size 6269 diff --git a/NanoPublicHealthQA/corpus/test.parquet b/NanoPublicHealthQA/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..ae2086b14f439d33abf54a1a003e4272021efa98 --- /dev/null +++ b/NanoPublicHealthQA/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b178830d234ea7b1927df97b1297d42e078155fa6abefff73adde900a917306 +size 62204 diff --git a/NanoPublicHealthQA/metadata/test.json b/NanoPublicHealthQA/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..c53784416f66548e1f8a62eab12628fcc8f7a5fd --- /dev/null +++ b/NanoPublicHealthQA/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoPublicHealthQA", + "source_task": "PublicHealthQA", + "source_type": "Retrieval", + "source_dataset": { + "path": "xhluca/publichealth-qa", + "revision": "main" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 86, + "corpus": 86, + "qrels": 86, + "bm25_rows": 86, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "ar", + "stemmer_algorithm": "arabic", + "tokenizer_name": null, + "reason": "language 'ar' mapped to stemmer 'arabic'" + }, + "qrels_coverage": { + "total": 86, + "hits": 86, + "recall": 1.0 + }, + "ndcg_at_10": 0.6722409775506696, + "ndcg_at_100": 0.708832504784472, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "test", + "source_query_count": 86, + "source_corpus_count": 86, + "source_qrels_query_count": 86, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 0, + "forced_doc_count": 0, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoPublicHealthQA/qrels/test.parquet b/NanoPublicHealthQA/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..20dfb274629c7c6d4687746d17de1ca51167b555 --- /dev/null +++ b/NanoPublicHealthQA/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be04d8801105f1e0ba141941b504362630a91ed025e89457b7e056cbdcd35d93 +size 2110 diff --git a/NanoPublicHealthQA/queries/test.parquet b/NanoPublicHealthQA/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..fcf337faafd6cd5d2a5b2ef0612b0a2099cdb9a3 --- /dev/null +++ b/NanoPublicHealthQA/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fed606333c6cba8178cab46dccaec961e6b5ecaa79c88502b7b46c7aa16ba085 +size 7750 diff --git a/NanoSciFact/corpus/test.parquet b/NanoSciFact/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..d39dfd68c96c6a2758557c81bad0c8733fe1ef42 --- /dev/null +++ b/NanoSciFact/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d54c52138f32a1fe75ccf8bdb7863924b4fbdd6bb0a39ac5d492cb9b082c15f6 +size 4382658 diff --git a/NanoSciFact/metadata/test.json b/NanoSciFact/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..83ea5fc615595105b8ffc6662bdeaa5b0fdfef52 --- /dev/null +++ b/NanoSciFact/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoSciFact", + "source_task": "SciFact", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/scifact", + "revision": "d56462d0e63a25450459c4f213e49ffdb866f7f9" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.6288504435215211, + "ndcg_at_100": 0.672445425266301, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 26, + "forced_doc_count": 28, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoSciFact/qrels/test.parquet b/NanoSciFact/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..79b10ecae7763cfe1c3ad80a1717c3e1423e8a14 --- /dev/null +++ b/NanoSciFact/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e760b306e8aacb875c040d48f9f71a9561a55b060618f0761092eb6303fd784c +size 4320 diff --git a/NanoSciFact/queries/test.parquet b/NanoSciFact/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..19840b07f9ca449dd9567fdf8e02d3b774e47957 --- /dev/null +++ b/NanoSciFact/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a373f20a1ffce418e2a6733841a92493f4f17fcc7c6a574f700a78e6e453fe3d +size 14663 diff --git a/NanoSciFactPL/corpus/test.parquet b/NanoSciFactPL/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..9dfeae45a0e1e48adba57f942190f425ecc8b1de --- /dev/null +++ b/NanoSciFactPL/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c00f21256cb339c1f6aa8208cc4ef3ea5a0bf251bd18df7880b2ee852d21b256 +size 5038845 diff --git a/NanoSciFactPL/metadata/test.json b/NanoSciFactPL/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..5d74a3a05768dfb93c17417961ad4c64e92f97e1 --- /dev/null +++ b/NanoSciFactPL/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoSciFactPL", + "source_task": "SciFact-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/SciFact-PL", + "revision": "d9707265734052a21d5892b6300ca11028fa067b" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.512080183670099, + "ndcg_at_100": 0.5813036932580986, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 43, + "forced_doc_count": 46, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoSciFactPL/qrels/test.parquet b/NanoSciFactPL/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..79b10ecae7763cfe1c3ad80a1717c3e1423e8a14 --- /dev/null +++ b/NanoSciFactPL/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e760b306e8aacb875c040d48f9f71a9561a55b060618f0761092eb6303fd784c +size 4320 diff --git a/NanoSciFactPL/queries/test.parquet b/NanoSciFactPL/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..535369aefb804f2f4526f873749fc28563966af2 --- /dev/null +++ b/NanoSciFactPL/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a8b623069d4647fb3738f0006987b0e4c8414a0f22fa0a88d37f53f92361dda +size 16857 diff --git a/NanoTRECCOVID/corpus/test.parquet b/NanoTRECCOVID/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..2d3aa6ebe5d6f729b9b854cd6609451df1bfc477 --- /dev/null +++ b/NanoTRECCOVID/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1981985632143208390620a9b6a4806798a60b3f5870308f840f82930698c1fc +size 6917408 diff --git a/NanoTRECCOVID/metadata/test.json b/NanoTRECCOVID/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..231dfb5c3bde9aa89a4c4042f169d91bbe9f933a --- /dev/null +++ b/NanoTRECCOVID/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoTRECCOVID", + "source_task": "TRECCOVID", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/trec-covid", + "revision": "bb9466bac8153a0349341eb1b22e06409e78ef4e" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.19660899617590905, + "ndcg_at_100": 0.316419378930131, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 20, + "forced_doc_count": 20, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoTRECCOVID/qrels/test.parquet b/NanoTRECCOVID/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..5e0538bf3a447181114ec459af86f970133a0d9e --- /dev/null +++ b/NanoTRECCOVID/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:192f6bdc7f601009b0dcda720df49b2e17de5360aa8b7fe19475b66d4d396f85 +size 1921 diff --git a/NanoTRECCOVID/queries/test.parquet b/NanoTRECCOVID/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..112b988f1af5cad31db69c2755d4f7b9dcc93c9d --- /dev/null +++ b/NanoTRECCOVID/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:405cc352921321c00856f43d321194a16db97e95ad91e961d2152615fe2366c7 +size 4014 diff --git a/NanoTRECCOVIDPL/corpus/test.parquet b/NanoTRECCOVIDPL/corpus/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..03c01652277bb93976ddf96132f00df2a35a0296 --- /dev/null +++ b/NanoTRECCOVIDPL/corpus/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7afa571680d8e2d5c7c6c9f1d7b10b53fd5d113372810758e89764b17c1140ff +size 7871336 diff --git a/NanoTRECCOVIDPL/metadata/test.json b/NanoTRECCOVIDPL/metadata/test.json new file mode 100644 index 0000000000000000000000000000000000000000..02b7b5a78cafbb5d3b610da06d656faf9a702aaa --- /dev/null +++ b/NanoTRECCOVIDPL/metadata/test.json @@ -0,0 +1,48 @@ +{ + "split": "NanoTRECCOVIDPL", + "source_task": "TRECCOVID-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/TRECCOVID-PL", + "revision": "1e710582482d4199ff690e5dd2491a70627523f5" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.10975964202381241, + "ndcg_at_100": 0.24460587901998618, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 3, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 30, + "forced_doc_count": 30, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/NanoTRECCOVIDPL/qrels/test.parquet b/NanoTRECCOVIDPL/qrels/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..5e0538bf3a447181114ec459af86f970133a0d9e --- /dev/null +++ b/NanoTRECCOVIDPL/qrels/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:192f6bdc7f601009b0dcda720df49b2e17de5360aa8b7fe19475b66d4d396f85 +size 1921 diff --git a/NanoTRECCOVIDPL/queries/test.parquet b/NanoTRECCOVIDPL/queries/test.parquet new file mode 100644 index 0000000000000000000000000000000000000000..d6923f6f9d8eab9e71969b79f8ac19fcacba1f18 --- /dev/null +++ b/NanoTRECCOVIDPL/queries/test.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:550975cd56fa77ae352e589564b77d72db8b5ad51cd599a7af99dd806500ee81 +size 4409 diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..3dac593136bb65c4eb54c7af9f0ce5235ddeebb1 --- /dev/null +++ b/README.md @@ -0,0 +1,204 @@ +--- +configs: +- config_name: bm25 + data_files: + - split: NanoCUREv1 + path: bm25/NanoCUREv1.parquet + - split: NanoNFCorpus + path: bm25/NanoNFCorpus.parquet + - split: NanoTRECCOVID + path: bm25/NanoTRECCOVID.parquet + - split: NanoTRECCOVIDPL + path: bm25/NanoTRECCOVIDPL.parquet + - split: NanoSciFact + path: bm25/NanoSciFact.parquet + - split: NanoSciFactPL + path: bm25/NanoSciFactPL.parquet + - split: NanoMedicalQA + path: bm25/NanoMedicalQA.parquet + - split: NanoPublicHealthQA + path: bm25/NanoPublicHealthQA.parquet + - split: NanoCmedqa + path: bm25/NanoCmedqa.parquet + - split: NanoCMedQAv2reranking + path: bm25/NanoCMedQAv2reranking.parquet +- config_name: corpus + data_files: + - split: NanoCUREv1 + path: NanoCUREv1/corpus/test.parquet + - split: NanoNFCorpus + path: NanoNFCorpus/corpus/test.parquet + - split: NanoTRECCOVID + path: NanoTRECCOVID/corpus/test.parquet + - split: NanoTRECCOVIDPL + path: NanoTRECCOVIDPL/corpus/test.parquet + - split: NanoSciFact + path: NanoSciFact/corpus/test.parquet + - split: NanoSciFactPL + path: NanoSciFactPL/corpus/test.parquet + - split: NanoMedicalQA + path: NanoMedicalQA/corpus/test.parquet + - split: NanoPublicHealthQA + path: NanoPublicHealthQA/corpus/test.parquet + - split: NanoCmedqa + path: NanoCmedqa/corpus/test.parquet + - split: NanoCMedQAv2reranking + path: NanoCMedQAv2reranking/corpus/test.parquet +- config_name: qrels + data_files: + - split: NanoCUREv1 + path: NanoCUREv1/qrels/test.parquet + - split: NanoNFCorpus + path: NanoNFCorpus/qrels/test.parquet + - split: NanoTRECCOVID + path: NanoTRECCOVID/qrels/test.parquet + - split: NanoTRECCOVIDPL + path: NanoTRECCOVIDPL/qrels/test.parquet + - split: NanoSciFact + path: NanoSciFact/qrels/test.parquet + - split: NanoSciFactPL + path: NanoSciFactPL/qrels/test.parquet + - split: NanoMedicalQA + path: NanoMedicalQA/qrels/test.parquet + - split: NanoPublicHealthQA + path: NanoPublicHealthQA/qrels/test.parquet + - split: NanoCmedqa + path: NanoCmedqa/qrels/test.parquet + - split: NanoCMedQAv2reranking + path: NanoCMedQAv2reranking/qrels/test.parquet +- config_name: queries + data_files: + - split: NanoCUREv1 + path: NanoCUREv1/queries/test.parquet + - split: NanoNFCorpus + path: NanoNFCorpus/queries/test.parquet + - split: NanoTRECCOVID + path: NanoTRECCOVID/queries/test.parquet + - split: NanoTRECCOVIDPL + path: NanoTRECCOVIDPL/queries/test.parquet + - split: NanoSciFact + path: NanoSciFact/queries/test.parquet + - split: NanoSciFactPL + path: NanoSciFactPL/queries/test.parquet + - split: NanoMedicalQA + path: NanoMedicalQA/queries/test.parquet + - split: NanoPublicHealthQA + path: NanoPublicHealthQA/queries/test.parquet + - split: NanoCmedqa + path: NanoCmedqa/queries/test.parquet + - split: NanoCMedQAv2reranking + path: NanoCMedQAv2reranking/queries/test.parquet + default: true +language: +- ar +- en +- pl +- zh +tags: +- information-retrieval +- retrieval +- nano +- bm25 +--- + +# NanoMedical + +This dataset is a Nano-style retrieval dataset. Nano-series evaluation can be run easily with the [HAKARI-Bench](https://github.com/hotchpotch/hakari-bench). + +NanoMedical is derived from MTEB(Medical, v1). It follows the Hugging Face Datasets layout convention used by [sentence-transformers/NanoBEIR-en](https://huggingface.co/datasets/sentence-transformers/NanoBEIR-en): each Nano split has separate `corpus`, `queries`, and `qrels` tables, and BM25 candidates are provided separately in a `bm25` table. This layout follows the NanoBEIR-style evaluation approach summarized in [NanoBEIR](https://huggingface.co/blog/sionic-ai/eval-sionic-nano-beir). + +NanoMedical contains 10 Nano retrieval splits derived from MTEB(Medical, v1). Each split keeps up to 200 eligible queries and up to 10000 corpus documents, with exact duplicate query and document text removed where the generator records that policy. + +## Source Links + +- Source benchmark: `MTEB(Medical, v1)` +- `clinia/CUREv1`: https://huggingface.co/datasets/clinia/CUREv1 +- `mteb/CMedQAv2-reranking`: https://huggingface.co/datasets/mteb/CMedQAv2-reranking +- `mteb/CmedqaRetrieval`: https://huggingface.co/datasets/mteb/CmedqaRetrieval +- `mteb/SciFact-PL`: https://huggingface.co/datasets/mteb/SciFact-PL +- `mteb/TRECCOVID-PL`: https://huggingface.co/datasets/mteb/TRECCOVID-PL +- `mteb/medical_qa`: https://huggingface.co/datasets/mteb/medical_qa +- `mteb/nfcorpus`: https://huggingface.co/datasets/mteb/nfcorpus +- `mteb/scifact`: https://huggingface.co/datasets/mteb/scifact +- `mteb/trec-covid`: https://huggingface.co/datasets/mteb/trec-covid +- `xhluca/publichealth-qa`: https://huggingface.co/datasets/xhluca/publichealth-qa + +## Data Layout + +This dataset uses four Hugging Face Datasets configs: + +- `corpus`: documents with `_id` and `text` +- `queries`: queries with `_id` and `text` +- `qrels`: positive relevance labels with `query-id` and `corpus-id` +- `bm25`: BM25 candidate lists with `query-id` and `corpus-ids` + +Each config uses the same Nano split names. If the actual generated dataset uses a different schema, config name, path layout, or field name, revise this section before publishing the README. + +## Construction Steps + +This dataset was built as follows. If the actual generation procedure differs, revise this section before publishing the README. + +1. Use MTEB(Medical, v1) as the upstream benchmark or dataset family. +2. Load the source datasets recorded in `manifest.json` and per-split metadata files. +3. Use the source benchmark evaluation split, preferring `test` when available as the source evaluation split policy. +4. Create one Nano split for each selected source retrieval task. +5. Keep up to 200 eligible queries per Nano split. +6. Include qrels-positive documents for the selected queries. +7. Fill the corpus from source corpus order up to 10000 documents. +8. Remove exact duplicate document text within each split. If a removed duplicate was referenced by qrels, rewrite qrels to the kept document id when the generator records that policy. +9. Store document title and body as a single `text` field when the source provides both. +10. Generate BM25 top-100 candidates with the tokenization policy recorded per split. +11. If a qrels-positive document is missing from the raw BM25 result, insert it into the final `bm25` candidate list by replacing a tail non-positive candidate. + +Some source queries had more positives than the BM25 candidate cap. For those cases, the Nano qrels keep the first available positive document so BM25 top-k remains comparable; 140 selected queries were capped and 57784 source positive qrels were omitted from this Nano slice. + +## BM25 Subset Policy + +The `bm25` config is a candidate subset for first-stage retrieval and reranking. It is not a separate source dataset. Each row contains one query id and a ranked list of corpus ids. + +BM25 candidates are generated from the selected corpus for each split. The configured candidate cap is top-100. When a qrels-positive document is not present in the raw BM25 result, the missing positive is forced into the final candidate list by replacing a tail candidate that is not positive for that query. Candidate ids are kept unique after replacement. + +## Split Mapping + +| Nano split | Source task | Source dataset | Queries | Corpus | Qrels | +|---|---|---|---:|---:|---:| +| `NanoCUREv1` | `CUREv1` | `clinia/CUREv1` | 200 | 10000 | 5181 | +| `NanoNFCorpus` | `NFCorpus` | `mteb/nfcorpus` | 200 | 3593 | 3718 | +| `NanoTRECCOVID` | `TRECCOVID` | `mteb/trec-covid` | 50 | 10000 | 50 | +| `NanoTRECCOVIDPL` | `TRECCOVID-PL` | `mteb/TRECCOVID-PL` | 50 | 10000 | 50 | +| `NanoSciFact` | `SciFact` | `mteb/scifact` | 200 | 5183 | 226 | +| `NanoSciFactPL` | `SciFact-PL` | `mteb/SciFact-PL` | 200 | 5183 | 226 | +| `NanoMedicalQA` | `MedicalQARetrieval` | `mteb/medical_qa` | 200 | 2007 | 200 | +| `NanoPublicHealthQA` | `PublicHealthQA` | `xhluca/publichealth-qa` | 86 | 86 | 86 | +| `NanoCmedqa` | `CmedqaRetrieval` | `mteb/CmedqaRetrieval` | 200 | 10000 | 324 | +| `NanoCMedQAv2reranking` | `CMedQAv2-reranking` | `mteb/CMedQAv2-reranking` | 200 | 10000 | 377 | + +## BM25 nDCG@10 + +`nDCG@10` is computed from the included BM25 ranking against the included qrels. + +Tokenizer policy summary: `stemmer:ar, stemmer:en, whitespace:pl, wordseg:zh`. + +| Nano split | Tokenizer | Forced BM25 positives | BM25 nDCG@10 | +|---|---|---:|---:| +| `NanoCUREv1` | `stemmer:en` | 2977 | 0.3699 | +| `NanoNFCorpus` | `stemmer:en` | 3029 | 0.2434 | +| `NanoTRECCOVID` | `stemmer:en` | 20 | 0.1966 | +| `NanoTRECCOVIDPL` | `whitespace:pl` | 30 | 0.1098 | +| `NanoSciFact` | `stemmer:en` | 28 | 0.6289 | +| `NanoSciFactPL` | `whitespace:pl` | 46 | 0.5121 | +| `NanoMedicalQA` | `stemmer:en` | 41 | 0.4736 | +| `NanoPublicHealthQA` | `stemmer:ar` | 0 | 0.6722 | +| `NanoCmedqa` | `wordseg:zh` | 202 | 0.1668 | +| `NanoCMedQAv2reranking` | `wordseg:zh` | 258 | 0.1500 | + +## Skipped Tasks + +| Source task | Source type or dataset | Reason | +|---|---|---| +| `MedrxivClusteringP2P.v2` | `Clustering` | not a retrieval-shaped task for Nano IR conversion | +| `MedrxivClusteringS2S.v2` | `Clustering` | not a retrieval-shaped task for Nano IR conversion | + +## License + +NanoMedical is a derived dataset. Users must comply with the licenses, terms, and attribution requirements of the upstream datasets and benchmarks. diff --git a/bm25/NanoCMedQAv2reranking.parquet b/bm25/NanoCMedQAv2reranking.parquet new file mode 100644 index 0000000000000000000000000000000000000000..0912c5877b8e86d3f48ef75a106996fde08696b5 --- /dev/null +++ b/bm25/NanoCMedQAv2reranking.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5894839509b6b804f7de7abe9efa0dd6d14d38f1a826eb7e94332750f435e86 +size 72030 diff --git a/bm25/NanoCUREv1.parquet b/bm25/NanoCUREv1.parquet new file mode 100644 index 0000000000000000000000000000000000000000..5f48b9d950c71bc219b453c0cf8b70251812edfd --- /dev/null +++ b/bm25/NanoCUREv1.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de6b7f0ba7393c110cf4657ea9a57370653b58cb2bcff8dcbc25244c92674a5b +size 263356 diff --git a/bm25/NanoCmedqa.parquet b/bm25/NanoCmedqa.parquet new file mode 100644 index 0000000000000000000000000000000000000000..390eb02f83e0167a8482ef8e5e0f9b932ae491a8 --- /dev/null +++ b/bm25/NanoCmedqa.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f3a85401a79fe4059d851ce87edfec9c5d04ba2874c88f10695474397cc666a +size 225376 diff --git a/bm25/NanoMedicalQA.parquet b/bm25/NanoMedicalQA.parquet new file mode 100644 index 0000000000000000000000000000000000000000..5463869d6a4fb2bf7311ce36f27540708145bad1 --- /dev/null +++ b/bm25/NanoMedicalQA.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62863be127dfa55b0d39e073e073f57459fd5f3d3498594a0d8696b436ec5376 +size 76187 diff --git a/bm25/NanoNFCorpus.parquet b/bm25/NanoNFCorpus.parquet new file mode 100644 index 0000000000000000000000000000000000000000..bab9a234b64739d55199559dd8e2fbd9c9cf8ffa --- /dev/null +++ b/bm25/NanoNFCorpus.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f92d48b253bcb5ef56b7b61c67e85dd1b1af66268d4c97afc8972cd2563427a +size 44283 diff --git a/bm25/NanoPublicHealthQA.parquet b/bm25/NanoPublicHealthQA.parquet new file mode 100644 index 0000000000000000000000000000000000000000..754743f3fc0d18678e5627b50741ff0e4e6e7c13 --- /dev/null +++ b/bm25/NanoPublicHealthQA.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ed987a7b7b26f0da2ec8ceb5ded10baec3c3bf98bbb389540ab83af041cb5a1 +size 8727 diff --git a/bm25/NanoSciFact.parquet b/bm25/NanoSciFact.parquet new file mode 100644 index 0000000000000000000000000000000000000000..6fd4a601d764e0f8d1f7fcf860564402f707447d --- /dev/null +++ b/bm25/NanoSciFact.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fdbbd3f18bfed338b9097d1e18be7228df2bc1014f112a577e383c86027558b +size 72103 diff --git a/bm25/NanoSciFactPL.parquet b/bm25/NanoSciFactPL.parquet new file mode 100644 index 0000000000000000000000000000000000000000..6ace3e59f8df8c2660c92c95d9289abf953da4a0 --- /dev/null +++ b/bm25/NanoSciFactPL.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a2f5bc1b2a7bff8d85d3f92adf4eaba87f463ee90729da5603711cb4983591f +size 72532 diff --git a/bm25/NanoTRECCOVID.parquet b/bm25/NanoTRECCOVID.parquet new file mode 100644 index 0000000000000000000000000000000000000000..496007df7ee6f05d063781fb6c22d56a02f94a13 --- /dev/null +++ b/bm25/NanoTRECCOVID.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:316b7274deb6d4d24fb27d4edf6c3e9d619ab498bdd1d0681e6ad3cd1ee3431b +size 35118 diff --git a/bm25/NanoTRECCOVIDPL.parquet b/bm25/NanoTRECCOVIDPL.parquet new file mode 100644 index 0000000000000000000000000000000000000000..d27e4894bb61f5d8b7157d4007ddd8054b8c1e49 --- /dev/null +++ b/bm25/NanoTRECCOVIDPL.parquet @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b9e3b13c788ce1e0ebf18926ceaed1432f2c1b9bdf336a5b1e57c0ec096f213 +size 39366 diff --git a/manifest.json b/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..e7ef4953388b11c75f92186cc5a92c1f5e4658fb --- /dev/null +++ b/manifest.json @@ -0,0 +1,519 @@ +{ + "dataset": "NanoMedical", + "benchmark": "MTEB(Medical, v1)", + "query_limit": 200, + "doc_limit": 10000, + "bm25_top_k": 100, + "splits": [ + { + "split": "NanoCUREv1", + "source_task": "CUREv1", + "source_type": "Retrieval", + "source_dataset": { + "path": "clinia/CUREv1", + "revision": "3bcf51c91e04d04a8a3329dfbe988b964c5cbe83" + }, + "source_eval_splits": [ + "all", + "dentistry_and_oral_health", + "dermatology", + "gastroenterology", + "genetics", + "neuroscience_and_neurology", + "orthopedic_surgery", + "otorhinolaryngology", + "plastic_surgery", + "psychiatry_and_psychology", + "pulmonology" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 5181, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 5181, + "hits": 5181, + "recall": 1.0 + }, + "ndcg_at_10": 0.3698959661937695, + "ndcg_at_100": 0.6160108835808653, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "all", + "source_query_count": 2000, + "source_corpus_count": 244600, + "source_qrels_query_count": 2000, + "source_queries_with_more_than_bm25_top_k_positives": 18, + "capped_queries_with_more_than_bm25_top_k_positives": 18, + "dropped_positive_qrels_by_capping": 2785, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 171, + "forced_doc_count": 2977, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoNFCorpus", + "source_task": "NFCorpus", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/nfcorpus", + "revision": "ec0fa4fe99da2ff19ca1214b7966684033a58814" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 3593, + "qrels": 3718, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 3718, + "hits": 3718, + "recall": 1.0 + }, + "ndcg_at_10": 0.24341837520694568, + "ndcg_at_100": 0.49341846897457736, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 323, + "source_corpus_count": 3633, + "source_qrels_query_count": 323, + "source_queries_with_more_than_bm25_top_k_positives": 22, + "capped_queries_with_more_than_bm25_top_k_positives": 22, + "dropped_positive_qrels_by_capping": 5753, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 65, + "qrels_rewritten_for_duplicate_text": 25, + "forced_queries": 185, + "forced_doc_count": 3029, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoTRECCOVID", + "source_task": "TRECCOVID", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/trec-covid", + "revision": "bb9466bac8153a0349341eb1b22e06409e78ef4e" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.19660899617590905, + "ndcg_at_100": 0.316419378930131, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 20, + "forced_doc_count": 20, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoTRECCOVIDPL", + "source_task": "TRECCOVID-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/TRECCOVID-PL", + "revision": "1e710582482d4199ff690e5dd2491a70627523f5" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.10975964202381241, + "ndcg_at_100": 0.24460587901998618, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 3, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 30, + "forced_doc_count": 30, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoSciFact", + "source_task": "SciFact", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/scifact", + "revision": "d56462d0e63a25450459c4f213e49ffdb866f7f9" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.6288504435215211, + "ndcg_at_100": 0.672445425266301, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 26, + "forced_doc_count": 28, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoSciFactPL", + "source_task": "SciFact-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/SciFact-PL", + "revision": "d9707265734052a21d5892b6300ca11028fa067b" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.512080183670099, + "ndcg_at_100": 0.5813036932580986, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 43, + "forced_doc_count": 46, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoMedicalQA", + "source_task": "MedicalQARetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/medical_qa", + "revision": "ae763399273d8b20506b80cf6f6f9a31a6a2b238" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 2007, + "qrels": 200, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 200, + "hits": 200, + "recall": 1.0 + }, + "ndcg_at_10": 0.4735533646828049, + "ndcg_at_100": 0.5241369411712118, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 2048, + "source_corpus_count": 2048, + "source_qrels_query_count": 2048, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 7, + "duplicate_doc_texts_removed": 42, + "qrels_rewritten_for_duplicate_text": 1, + "forced_queries": 41, + "forced_doc_count": 41, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoPublicHealthQA", + "source_task": "PublicHealthQA", + "source_type": "Retrieval", + "source_dataset": { + "path": "xhluca/publichealth-qa", + "revision": "main" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 86, + "corpus": 86, + "qrels": 86, + "bm25_rows": 86, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "ar", + "stemmer_algorithm": "arabic", + "tokenizer_name": null, + "reason": "language 'ar' mapped to stemmer 'arabic'" + }, + "qrels_coverage": { + "total": 86, + "hits": 86, + "recall": 1.0 + }, + "ndcg_at_10": 0.6722409775506696, + "ndcg_at_100": 0.708832504784472, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "test", + "source_query_count": 86, + "source_corpus_count": 86, + "source_qrels_query_count": 86, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 0, + "forced_doc_count": 0, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoCmedqa", + "source_task": "CmedqaRetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/CmedqaRetrieval", + "revision": "c476f85bf03d6642ec66bf54b9a551c88108bbb4" + }, + "source_eval_splits": [ + "dev" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 324, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 324, + "hits": 324, + "recall": 1.0 + }, + "ndcg_at_10": 0.16678868356587398, + "ndcg_at_100": 0.30693075139953757, + "source_container": "task.dataset", + "source_eval_split_used": "dev", + "source_query_count": 3999, + "source_corpus_count": 100001, + "source_qrels_query_count": 3999, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 134, + "forced_doc_count": 202, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split": "NanoCMedQAv2reranking", + "source_task": "CMedQAv2-reranking", + "source_type": "Reranking", + "source_dataset": { + "path": "mteb/CMedQAv2-reranking", + "revision": "01399b6956a4c6bf9edc5b6b90e176aeeb399686" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 377, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 377, + "hits": 377, + "recall": 1.0 + }, + "ndcg_at_10": 0.15003900776298235, + "ndcg_at_100": 0.29903015701374225, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 1000, + "source_corpus_count": 100000, + "source_qrels_query_count": 1000, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 406, + "qrels_rewritten_for_duplicate_text": 7, + "forced_queries": 148, + "forced_doc_count": 258, + "missing_positive_doc_count_after_forcing": 0 + } + ], + "skipped": [ + { + "task": "MedrxivClusteringP2P.v2", + "source_type": "Clustering", + "source_dataset": { + "path": "mteb/medrxiv-clustering-p2p", + "revision": "e7a26af6f3ae46b30dde8737f02c07b1505bcc73" + }, + "reason": "not a retrieval-shaped task for Nano IR conversion" + }, + { + "task": "MedrxivClusteringS2S.v2", + "source_type": "Clustering", + "source_dataset": { + "path": "mteb/medrxiv-clustering-s2s", + "revision": "35191c8c0dca72d8ff3efcd72aa802307d469663" + }, + "reason": "not a retrieval-shaped task for Nano IR conversion" + } + ] +} diff --git a/metadata/NanoCMedQAv2reranking.json b/metadata/NanoCMedQAv2reranking.json new file mode 100644 index 0000000000000000000000000000000000000000..705b6fa87c14dfda7ed18ee5f4b6058d05c18bba --- /dev/null +++ b/metadata/NanoCMedQAv2reranking.json @@ -0,0 +1,48 @@ +{ + "split": "NanoCMedQAv2reranking", + "source_task": "CMedQAv2-reranking", + "source_type": "Reranking", + "source_dataset": { + "path": "mteb/CMedQAv2-reranking", + "revision": "01399b6956a4c6bf9edc5b6b90e176aeeb399686" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 377, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 377, + "hits": 377, + "recall": 1.0 + }, + "ndcg_at_10": 0.15003900776298235, + "ndcg_at_100": 0.29903015701374225, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 1000, + "source_corpus_count": 100000, + "source_qrels_query_count": 1000, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 406, + "qrels_rewritten_for_duplicate_text": 7, + "forced_queries": 148, + "forced_doc_count": 258, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoCUREv1.json b/metadata/NanoCUREv1.json new file mode 100644 index 0000000000000000000000000000000000000000..7943d1233834ffdba912181ce10c91ba8cbb613f --- /dev/null +++ b/metadata/NanoCUREv1.json @@ -0,0 +1,58 @@ +{ + "split": "NanoCUREv1", + "source_task": "CUREv1", + "source_type": "Retrieval", + "source_dataset": { + "path": "clinia/CUREv1", + "revision": "3bcf51c91e04d04a8a3329dfbe988b964c5cbe83" + }, + "source_eval_splits": [ + "all", + "dentistry_and_oral_health", + "dermatology", + "gastroenterology", + "genetics", + "neuroscience_and_neurology", + "orthopedic_surgery", + "otorhinolaryngology", + "plastic_surgery", + "psychiatry_and_psychology", + "pulmonology" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 5181, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 5181, + "hits": 5181, + "recall": 1.0 + }, + "ndcg_at_10": 0.3698959661937695, + "ndcg_at_100": 0.6160108835808653, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "all", + "source_query_count": 2000, + "source_corpus_count": 244600, + "source_qrels_query_count": 2000, + "source_queries_with_more_than_bm25_top_k_positives": 18, + "capped_queries_with_more_than_bm25_top_k_positives": 18, + "dropped_positive_qrels_by_capping": 2785, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 171, + "forced_doc_count": 2977, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoCmedqa.json b/metadata/NanoCmedqa.json new file mode 100644 index 0000000000000000000000000000000000000000..a6d8635636030ff1bfc3b13896cf087a7ef89917 --- /dev/null +++ b/metadata/NanoCmedqa.json @@ -0,0 +1,48 @@ +{ + "split": "NanoCmedqa", + "source_task": "CmedqaRetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/CmedqaRetrieval", + "revision": "c476f85bf03d6642ec66bf54b9a551c88108bbb4" + }, + "source_eval_splits": [ + "dev" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 10000, + "qrels": 324, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "qrels_coverage": { + "total": 324, + "hits": 324, + "recall": 1.0 + }, + "ndcg_at_10": 0.16678868356587398, + "ndcg_at_100": 0.30693075139953757, + "source_container": "task.dataset", + "source_eval_split_used": "dev", + "source_query_count": 3999, + "source_corpus_count": 100001, + "source_qrels_query_count": 3999, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 134, + "forced_doc_count": 202, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoMedicalQA.json b/metadata/NanoMedicalQA.json new file mode 100644 index 0000000000000000000000000000000000000000..b425166ad063a0d891637060a18136665b04af2a --- /dev/null +++ b/metadata/NanoMedicalQA.json @@ -0,0 +1,48 @@ +{ + "split": "NanoMedicalQA", + "source_task": "MedicalQARetrieval", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/medical_qa", + "revision": "ae763399273d8b20506b80cf6f6f9a31a6a2b238" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 2007, + "qrels": 200, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 200, + "hits": 200, + "recall": 1.0 + }, + "ndcg_at_10": 0.4735533646828049, + "ndcg_at_100": 0.5241369411712118, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 2048, + "source_corpus_count": 2048, + "source_qrels_query_count": 2048, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 7, + "duplicate_doc_texts_removed": 42, + "qrels_rewritten_for_duplicate_text": 1, + "forced_queries": 41, + "forced_doc_count": 41, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoNFCorpus.json b/metadata/NanoNFCorpus.json new file mode 100644 index 0000000000000000000000000000000000000000..14833cf96d4884764d8be053352b65c65c080274 --- /dev/null +++ b/metadata/NanoNFCorpus.json @@ -0,0 +1,48 @@ +{ + "split": "NanoNFCorpus", + "source_task": "NFCorpus", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/nfcorpus", + "revision": "ec0fa4fe99da2ff19ca1214b7966684033a58814" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 3593, + "qrels": 3718, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 3718, + "hits": 3718, + "recall": 1.0 + }, + "ndcg_at_10": 0.24341837520694568, + "ndcg_at_100": 0.49341846897457736, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 323, + "source_corpus_count": 3633, + "source_qrels_query_count": 323, + "source_queries_with_more_than_bm25_top_k_positives": 22, + "capped_queries_with_more_than_bm25_top_k_positives": 22, + "dropped_positive_qrels_by_capping": 5753, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 65, + "qrels_rewritten_for_duplicate_text": 25, + "forced_queries": 185, + "forced_doc_count": 3029, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoPublicHealthQA.json b/metadata/NanoPublicHealthQA.json new file mode 100644 index 0000000000000000000000000000000000000000..c53784416f66548e1f8a62eab12628fcc8f7a5fd --- /dev/null +++ b/metadata/NanoPublicHealthQA.json @@ -0,0 +1,48 @@ +{ + "split": "NanoPublicHealthQA", + "source_task": "PublicHealthQA", + "source_type": "Retrieval", + "source_dataset": { + "path": "xhluca/publichealth-qa", + "revision": "main" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 86, + "corpus": 86, + "qrels": 86, + "bm25_rows": 86, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "ar", + "stemmer_algorithm": "arabic", + "tokenizer_name": null, + "reason": "language 'ar' mapped to stemmer 'arabic'" + }, + "qrels_coverage": { + "total": 86, + "hits": 86, + "recall": 1.0 + }, + "ndcg_at_10": 0.6722409775506696, + "ndcg_at_100": 0.708832504784472, + "source_container": "task.corpus/queries/relevant_docs", + "source_eval_split_used": "test", + "source_query_count": 86, + "source_corpus_count": 86, + "source_qrels_query_count": 86, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 0, + "forced_doc_count": 0, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoSciFact.json b/metadata/NanoSciFact.json new file mode 100644 index 0000000000000000000000000000000000000000..83ea5fc615595105b8ffc6662bdeaa5b0fdfef52 --- /dev/null +++ b/metadata/NanoSciFact.json @@ -0,0 +1,48 @@ +{ + "split": "NanoSciFact", + "source_task": "SciFact", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/scifact", + "revision": "d56462d0e63a25450459c4f213e49ffdb866f7f9" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.6288504435215211, + "ndcg_at_100": 0.672445425266301, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 26, + "forced_doc_count": 28, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoSciFactPL.json b/metadata/NanoSciFactPL.json new file mode 100644 index 0000000000000000000000000000000000000000..5d74a3a05768dfb93c17417961ad4c64e92f97e1 --- /dev/null +++ b/metadata/NanoSciFactPL.json @@ -0,0 +1,48 @@ +{ + "split": "NanoSciFactPL", + "source_task": "SciFact-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/SciFact-PL", + "revision": "d9707265734052a21d5892b6300ca11028fa067b" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 200, + "corpus": 5183, + "qrels": 226, + "bm25_rows": 200, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 226, + "hits": 226, + "recall": 1.0 + }, + "ndcg_at_10": 0.512080183670099, + "ndcg_at_100": 0.5813036932580986, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 300, + "source_corpus_count": 5183, + "source_qrels_query_count": 300, + "source_queries_with_more_than_bm25_top_k_positives": 0, + "capped_queries_with_more_than_bm25_top_k_positives": 0, + "dropped_positive_qrels_by_capping": 0, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 43, + "forced_doc_count": 46, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoTRECCOVID.json b/metadata/NanoTRECCOVID.json new file mode 100644 index 0000000000000000000000000000000000000000..231dfb5c3bde9aa89a4c4042f169d91bbe9f933a --- /dev/null +++ b/metadata/NanoTRECCOVID.json @@ -0,0 +1,48 @@ +{ + "split": "NanoTRECCOVID", + "source_task": "TRECCOVID", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/trec-covid", + "revision": "bb9466bac8153a0349341eb1b22e06409e78ef4e" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.19660899617590905, + "ndcg_at_100": 0.316419378930131, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 0, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 20, + "forced_doc_count": 20, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/metadata/NanoTRECCOVIDPL.json b/metadata/NanoTRECCOVIDPL.json new file mode 100644 index 0000000000000000000000000000000000000000..02b7b5a78cafbb5d3b610da06d656faf9a702aaa --- /dev/null +++ b/metadata/NanoTRECCOVIDPL.json @@ -0,0 +1,48 @@ +{ + "split": "NanoTRECCOVIDPL", + "source_task": "TRECCOVID-PL", + "source_type": "Retrieval", + "source_dataset": { + "path": "mteb/TRECCOVID-PL", + "revision": "1e710582482d4199ff690e5dd2491a70627523f5" + }, + "source_eval_splits": [ + "test" + ], + "query_limit": 200, + "doc_limit": 10000, + "queries": 50, + "corpus": 10000, + "qrels": 50, + "bm25_rows": 50, + "bm25_top_k": 100, + "bm25_tokenization": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "qrels_coverage": { + "total": 50, + "hits": 50, + "recall": 1.0 + }, + "ndcg_at_10": 0.10975964202381241, + "ndcg_at_100": 0.24460587901998618, + "source_container": "task.dataset", + "source_eval_split_used": "test", + "source_query_count": 50, + "source_corpus_count": 171332, + "source_qrels_query_count": 50, + "source_queries_with_more_than_bm25_top_k_positives": 50, + "capped_queries_with_more_than_bm25_top_k_positives": 50, + "dropped_positive_qrels_by_capping": 24623, + "skipped_queries_missing_text_or_qrels": 0, + "skipped_duplicate_query_texts": 0, + "duplicate_doc_texts_removed": 3, + "qrels_rewritten_for_duplicate_text": 0, + "forced_queries": 30, + "forced_doc_count": 30, + "missing_positive_doc_count_after_forcing": 0 +} diff --git a/nano_bm25_subset_config.json b/nano_bm25_subset_config.json new file mode 100644 index 0000000000000000000000000000000000000000..db520a9156885cc46a610ca7739205069bd84379 --- /dev/null +++ b/nano_bm25_subset_config.json @@ -0,0 +1,200 @@ +{ + "dataset": "NanoMedical", + "benchmark": "MTEB(Medical, v1)", + "query_limit": 200, + "doc_limit": 10000, + "bm25_top_k": 100, + "candidate_policy": "Rank the selected split corpus with BM25, keep up to top-100 candidates, and force qrels-positive documents into the final candidate list when missing.", + "splits": [ + { + "split_name": "NanoCUREv1", + "source_task": "CUREv1", + "queries": 200, + "corpus": 10000, + "qrels": 5181, + "tokenization_plan": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "ndcg_at_10": 0.3698959661937695, + "ndcg_at_100": 0.6160108835808653, + "forced_queries": 171, + "forced_doc_count": 2977, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoNFCorpus", + "source_task": "NFCorpus", + "queries": 200, + "corpus": 3593, + "qrels": 3718, + "tokenization_plan": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "ndcg_at_10": 0.24341837520694568, + "ndcg_at_100": 0.49341846897457736, + "forced_queries": 185, + "forced_doc_count": 3029, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoTRECCOVID", + "source_task": "TRECCOVID", + "queries": 50, + "corpus": 10000, + "qrels": 50, + "tokenization_plan": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "ndcg_at_10": 0.19660899617590905, + "ndcg_at_100": 0.316419378930131, + "forced_queries": 20, + "forced_doc_count": 20, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoTRECCOVIDPL", + "source_task": "TRECCOVID-PL", + "queries": 50, + "corpus": 10000, + "qrels": 50, + "tokenization_plan": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "ndcg_at_10": 0.10975964202381241, + "ndcg_at_100": 0.24460587901998618, + "forced_queries": 30, + "forced_doc_count": 30, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoSciFact", + "source_task": "SciFact", + "queries": 200, + "corpus": 5183, + "qrels": 226, + "tokenization_plan": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "ndcg_at_10": 0.6288504435215211, + "ndcg_at_100": 0.672445425266301, + "forced_queries": 26, + "forced_doc_count": 28, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoSciFactPL", + "source_task": "SciFact-PL", + "queries": 200, + "corpus": 5183, + "qrels": 226, + "tokenization_plan": { + "mode": "whitespace", + "language": "pl", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'pl' falls back to whitespace split" + }, + "ndcg_at_10": 0.512080183670099, + "ndcg_at_100": 0.5813036932580986, + "forced_queries": 43, + "forced_doc_count": 46, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoMedicalQA", + "source_task": "MedicalQARetrieval", + "queries": 200, + "corpus": 2007, + "qrels": 200, + "tokenization_plan": { + "mode": "stemmer", + "language": "en", + "stemmer_algorithm": "english", + "tokenizer_name": null, + "reason": "language 'en' mapped to stemmer 'english'" + }, + "ndcg_at_10": 0.4735533646828049, + "ndcg_at_100": 0.5241369411712118, + "forced_queries": 41, + "forced_doc_count": 41, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoPublicHealthQA", + "source_task": "PublicHealthQA", + "queries": 86, + "corpus": 86, + "qrels": 86, + "tokenization_plan": { + "mode": "stemmer", + "language": "ar", + "stemmer_algorithm": "arabic", + "tokenizer_name": null, + "reason": "language 'ar' mapped to stemmer 'arabic'" + }, + "ndcg_at_10": 0.6722409775506696, + "ndcg_at_100": 0.708832504784472, + "forced_queries": 0, + "forced_doc_count": 0, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoCmedqa", + "source_task": "CmedqaRetrieval", + "queries": 200, + "corpus": 10000, + "qrels": 324, + "tokenization_plan": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "ndcg_at_10": 0.16678868356587398, + "ndcg_at_100": 0.30693075139953757, + "forced_queries": 134, + "forced_doc_count": 202, + "missing_positive_doc_count_after_forcing": 0 + }, + { + "split_name": "NanoCMedQAv2reranking", + "source_task": "CMedQAv2-reranking", + "queries": 200, + "corpus": 10000, + "qrels": 377, + "tokenization_plan": { + "mode": "wordseg", + "language": "zh", + "stemmer_algorithm": null, + "tokenizer_name": null, + "reason": "language 'zh' mapped to word-segmentation tokenizer" + }, + "ndcg_at_10": 0.15003900776298235, + "ndcg_at_100": 0.29903015701374225, + "forced_queries": 148, + "forced_doc_count": 258, + "missing_positive_doc_count_after_forcing": 0 + } + ] +}