NanoIFIR

This dataset is a Nano-style retrieval dataset for HAKARI-bench.

NanoIFIR contains 7 Nano retrieval splits derived from IFIR. 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.

Usage

from datasets import load_dataset

dataset_id = "hakari-bench/NanoIFIR"
split = "NanoIFIRAila"

queries = load_dataset(dataset_id, "queries", split=split)
corpus = load_dataset(dataset_id, "corpus", split=split)
qrels = load_dataset(dataset_id, "qrels", split=split)
reranking_candidates = load_dataset(dataset_id, "reranking_hybrid", split=split)

Data Layout

This dataset uses six 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
• harrier_oss_v1_270m: dense candidate lists from microsoft/harrier-oss-v1-270m
• reranking_hybrid: RRF candidate lists built from bm25 and harrier_oss_v1_270m

Each config has the same Nano split names.

Candidate Construction

• bm25: local BM25 top-500 with automatic language-aware tokenization. The resolved tokenizer is shown in the Candidate Quality table, for example wordseg@ja.
• harrier_oss_v1_270m: dense top-500 from microsoft/harrier-oss-v1-270m. In tables this is shown as Dense; Dense means microsoft/harrier-oss-v1-270m with the web_search_query prompt for queries and cosine similarity over normalized embeddings.
• reranking_hybrid: RRF over bm25 and harrier_oss_v1_270m using rrf_k=100, keeping the RRF top-100.

Safeguard means rank 101 is appended only when RRF top-100 contains no qrels-positive document.

Split Statistics

Length statistics are character counts computed with len(str(text)).

Nano split	Queries	Corpus	Qrels	Query chars avg	Query chars p50	Query chars p75	Doc chars avg	Doc chars p50	Doc chars p75
NanoIFIRAila	40	2914	119	2890.1	3012.0	3581.5	19998.1	15303.0	24733.2
NanoIFIRCds	42	10000	466	225.2	213.0	260.2	1630.2	1604.0	1983.0
NanoIFIRFiQA	200	10000	1010	65.8	64.5	78.2	791.9	541.0	975.0
NanoIFIRFire	167	1739	563	3283.8	3242.0	3574.5	27167.7	24688.0	36394.5
NanoIFIRNFCorpus	86	3593	242	37.8	37.0	44.8	1589.5	1612.0	1868.0
NanoIFIRPm	59	10000	1217	145.7	146.0	157.0	2244.9	1651.5	2792.5
NanoIFIRScifact	43	10000	255	73.6	73.0	90.5	1452.6	1454.0	1819.0

Candidate Quality

nDCG@10 and Recall@100 are computed from the included candidate rankings against the included qrels, then reported as 0-100 scores such as 52.45. Recall@100 uses only the top 100 candidates; an optional rank-101 safeguard positive is not counted in Recall@100.

Dense means microsoft/harrier-oss-v1-270m with the web_search_query prompt and cosine similarity.

Nano split	BM25 tokenizer	BM25 nDCG@10	Dense nDCG@10	Hybrid nDCG@10	BM25 Recall@100	Dense Recall@100	Hybrid Recall@100	Hybrid candidates	Safeguard positives
Mean	-	37.84	46.06	44.97	59.74	72.00	72.71	-	49
NanoIFIRAila	english_porter_stop	9.88	8.78	7.98	28.30	35.19	35.35	100-101	21
NanoIFIRCds	english_porter_stop	22.58	40.73	33.76	35.88	72.48	67.06	100-101	3
NanoIFIRFiQA	english_porter_stop	34.22	53.28	46.78	62.14	79.86	77.92	100-101	3
NanoIFIRFire	english_porter_stop	35.66	34.21	39.96	74.36	69.22	76.63	100-101	12
NanoIFIRNFCorpus	english_porter_stop	33.38	45.80	41.08	63.58	83.40	78.27	100-101	9
NanoIFIRPm	english_porter_stop	42.32	54.48	54.68	55.85	68.25	74.18	100-101	1
NanoIFIRScifact	english_porter_stop	86.82	85.16	90.55	98.10	95.56	99.57	100	0

Hybrid Safeguard Summary

• Safeguard positives: 49
• Rows limited by corpus size: 0
• Metadata file: reranking_hybrid_metadata.json

Source Links

• Source benchmark: IFIR
• if-ir/aila: https://huggingface.co/datasets/if-ir/aila
• if-ir/cds: https://huggingface.co/datasets/if-ir/cds
• if-ir/fiqa: https://huggingface.co/datasets/if-ir/fiqa
• if-ir/fire: https://huggingface.co/datasets/if-ir/fire
• if-ir/nfcorpus: https://huggingface.co/datasets/if-ir/nfcorpus
• if-ir/pm: https://huggingface.co/datasets/if-ir/pm
• if-ir/scifact_open: https://huggingface.co/datasets/if-ir/scifact_open

License

NanoIFIR is a derived dataset. Users must comply with the licenses, terms, and attribution requirements of the upstream datasets and benchmarks.