Hyukkyu/train-agnews

AGNews — Training, unified schema

A normalised copy of the dataset behind the mteb task AGNews, a retrieval training set built from sentence-transformers/agnews. The benchmark's queries, documents and relevance judgements, less the non-English rows (see Language filtering) — reshaped into one strict schema shared by every dataset in this collection.

Source	sentence-transformers/agnews @ e8a1e3d19a3c (the revision pinned in mteb)
Domain · languages	news · eng
Queries / documents / qrels	639,160 / 690,784 / 714,695 (English only, see Language filtering)
Qrels per query	min 1 · mean 1.12 · max 162
Score values	1 ×714,695
Layout	queries · corpus · qrels · hard-negatives · teacher-scores, split train
Splits	corpus: train · hard-negatives: train · judgments: train · qrels: train · queries: train · teacher-scores: train
Hard negatives	sources: dense · 8,595,207 rows
Teacher scores	jinaai/jina-reranker-v3.5 · 8,684,464 rows (positives included)
Judgments	judgments: typesafe/jev-1.13.0 · 2,460,794 rows
Ids	sha1(text)[:20]; identical texts collapse to one document (387,363 collapsed)
Pair recovery	794,921 of 794,921 source pairs reconstructed from queries × qrels × corpus with byte-equal text
License	unspecified

Language filtering

The source carries non-English (almost all Italian) wire stories, which revision 1f43702090d1 still holds. This revision drops the queries whose headline and positive article together, and the passages whose text, both fastText lid.176 and langid identify as non-English; every qrels, hard-negatives and teacher-scores row that references a dropped query or passage; and the queries left without a positive (scripts/preprocess/dataset/filter_language.py in the SPARSE repository).

	before	after
queries rows	715,929	639,160
corpus rows	770,382	690,784
qrels rows	794,921	714,695
hard-negatives rows	9,920,606	8,595,207
teacher-scores rows	10,020,606	8,684,464

76,700 non-English queries and 79,598 non-English passages; 69 queries lost their only positive.

Schema

config	columns	rules
queries	id: string, text: string	ids unique and non-empty; every query has ≥ 1 qrel
corpus	id: string, title: string, text: string	title is always present ("" when the source has none)
qrels	query-id: string, corpus-id: string, score: int32	referential integrity to both tables; no duplicate pairs; no floats
hard-negatives	query-id: string, corpus-id: string, rank: int32, source: string	one row per mined negative; (query-id, corpus-id, source) unique; never a labelled positive of the same query
teacher-scores	query-id: string, corpus-id: string, teacher: string, score: float32	one row per scored pair (positives included); a row means scored — never a placeholder
judgments	query-id: string, corpus-id: string, judge: string, role: string, p_yes: float64, round: int32	one row per judged pair; role is positive (the training positive) or candidate (a mined candidate, never a labelled positive or a labelled negative); p_yes in [0, 1]; round 0 the first request, 1.. the top-ups

Files are Parquet, sorted by id, zstd-compressed, sharded at 500 MB. Every rule above is enforced by a validator before publishing; provenance.json records the source file hashes, what changed, and the output file hashes.

What changed from the source

• byte-preserved all text — no whitespace, newline, or control-character normalisation

Hard negatives and teacher scores

Filled by the owner's annotation pipeline (annotation=jina35) for the train split of the query set(s) below; queries without a labelled positive are left out.

• Candidates: dense retrieval with jinaai/jina-embeddings-v5-text-small over the full corpus to depth 1,000; 100 candidates per query drawn from the rank windows 1–30 (30), 31–100 (30), 101–300 (20), 301–1000 (20), the query's labelled positives excluded. rank is the dense rank; source is dense for a mined row and dataset for a negative the source labels itself (those are kept for every query of the split, sampled or not).
• Teacher: jinaai/jina-reranker-v3.5, listwise: a query's positive and all of its candidates are scored together in one context of up to 32,768 tokens. score is the raw cosine score, one row per (query, positive) and per (query, candidate); a labelled negative that was also mined is scored once. No filtering is applied to the tables.

configs	queries	hard negatives	teacher scores
hard-negatives · teacher-scores	89,257 (seeded sample, seed 1)	8,595,207 (8,595,207 dense)	8,684,464

from datasets import load_dataset
negatives = load_dataset("Hyukkyu/train-agnews", "hard-negatives", split="train")
scores    = load_dataset("Hyukkyu/train-agnews", "teacher-scores", split="train")

Jev judgments

judgments holds, for every query of the training sample (the queries with teacher scores), whether TypeSafe's Jev (jev-1.13.0) judged its training positive and its mined candidates relevant: p_yes is Jev's P(yes) for the source's question (e.g. does the passage answer the query?). They locate the false negatives among the mined candidates and the mislabelled positives.

• Requests. One request per query (round 0): its training positive, its candidates whose teacher score taken as (cos + 1) / 2 is at least 0.85 × the positive's (at most 24, the highest scores) and 8 random candidates below that, shuffled under neutral ids, one yes/no question per passage. Queries left with fewer than 7 candidates under their source's cutoff got their next hardest unjudged candidates in rounds 1–2 (25 per request), those still under 10 in rounds 3–8 (12 per request). The dataset's own labelled negatives were never sent. Texts were cut to 256 (query) and 512 (passage) tokens of the jina-embeddings-v5 small tokenizer. Jev answers a request's passages in one context, so P(yes) is calibrated to these groups: the thresholds below apply to this table, not to single-pair calls.
• Accuracy (an audit of 329 pairs from one third of the queries, labelled blind by an LLM (Claude), at the pilot's fixed thresholds 0.35 and 0.15): a candidate at P(yes) ≥ 0.35 was relevant 78% of the time inside the band (n = 89) and 60% below it (n = 30); one under 0.35 was relevant 9% (band, n = 90) and 2% (below the band, n = 45) of the time. A positive under 0.15 was mislabelled 92% of the time (n = 36) in the sources that keep the check; in agnews, hotpotqa, npr, whose positives are right by construction, a further check found Jev's flags right 0%–17% of the time, so their positives are not checked.
• Use (the SPARSE loader, annotation.filter.judge): a candidate at P(yes) ≥ its source's cutoff (below; fitted on 1,946 labelled pairs) is never a negative; a positive under 0.15 is replaced by the candidate Jev scores highest if that is ≥ 0.8, else the query is dropped; a candidate at ≥ 0.9 can become an extra positive. Compare p_yes as a float64 (it is stored as one).

config	queries	rows	candidates per query	top-up rows	candidate cutoff	positive check
judgments	89,257	2,460,794	26.6	709,019	0.30	skipped

Rows of queries or passages this revision no longer holds were left out: judgments 376,131 rows (10,743 queries removed entirely, 34,861 more lost some candidates).

Load it

from datasets import load_dataset
queries   = load_dataset("Hyukkyu/train-agnews", "queries", split="train")
corpus    = load_dataset("Hyukkyu/train-agnews", "corpus", split="train")
qrels     = load_dataset("Hyukkyu/train-agnews", "qrels", split="train")
judgments = load_dataset("Hyukkyu/train-agnews", "judgments", split="train")

License and attribution

The data is redistributed under the source's terms — unspecified. All credit belongs to the original authors; see the source repository and the references in mteb's task metadata (https://huggingface.co/datasets/sentence-transformers/agnews). This repository is an independent repackaging and is not affiliated with the RTEB or MTEB maintainers.

No license declared upstream. Neither the source repo nor its origin states redistribution terms. This copy is published for reproducibility of the RTEB benchmark only; check with the original authors before any other use.