Copy dataset from hotchpotch/NanoLongEmbed

README.md

@@ -0,0 +1,295 @@
+---
+configs:
+- config_name: bm25
+  data_files:
+  - split: Nano2WikiMultihopQA
+    path: bm25/Nano2WikiMultihopQA-*
+  - split: NanoNarrativeQA
+    path: bm25/NanoNarrativeQA-*
+  - split: NanoNeedle
+    path: bm25/NanoNeedle-*
+  - split: NanoPasskey
+    path: bm25/NanoPasskey-*
+  - split: NanoQMSum
+    path: bm25/NanoQMSum-*
+  - split: NanoSummScreenFD
+    path: bm25/NanoSummScreenFD-*
+- config_name: corpus
+  data_files:
+  - split: Nano2WikiMultihopQA
+    path: corpus/Nano2WikiMultihopQA-*
+  - split: NanoNarrativeQA
+    path: corpus/NanoNarrativeQA-*
+  - split: NanoNeedle
+    path: corpus/NanoNeedle-*
+  - split: NanoPasskey
+    path: corpus/NanoPasskey-*
+  - split: NanoQMSum
+    path: corpus/NanoQMSum-*
+  - split: NanoSummScreenFD
+    path: corpus/NanoSummScreenFD-*
+- config_name: qrels
+  data_files:
+  - split: Nano2WikiMultihopQA
+    path: qrels/Nano2WikiMultihopQA-*
+  - split: NanoNarrativeQA
+    path: qrels/NanoNarrativeQA-*
+  - split: NanoNeedle
+    path: qrels/NanoNeedle-*
+  - split: NanoPasskey
+    path: qrels/NanoPasskey-*
+  - split: NanoQMSum
+    path: qrels/NanoQMSum-*
+  - split: NanoSummScreenFD
+    path: qrels/NanoSummScreenFD-*
+- config_name: queries
+  data_files:
+  - split: Nano2WikiMultihopQA
+    path: queries/Nano2WikiMultihopQA-*
+  - split: NanoNarrativeQA
+    path: queries/NanoNarrativeQA-*
+  - split: NanoNeedle
+    path: queries/NanoNeedle-*
+  - split: NanoPasskey
+    path: queries/NanoPasskey-*
+  - split: NanoQMSum
+    path: queries/NanoQMSum-*
+  - split: NanoSummScreenFD
+    path: queries/NanoSummScreenFD-*
+language:
+- en
+tags:
+- Long Context
+- retrieval
+- nano
+dataset_info:
+- config_name: bm25
+  features:
+  - name: query-id
+    dtype: string
+  - name: corpus-ids
+    list: string
+  splits:
+  - name: Nano2WikiMultihopQA
+    num_bytes: 216150
+    num_examples: 200
+  - name: NanoNarrativeQA
+    num_bytes: 217115
+    num_examples: 200
+  - name: NanoNeedle
+    num_bytes: 172831
+    num_examples: 98
+  - name: NanoPasskey
+    num_bytes: 162029
+    num_examples: 100
+  - name: NanoQMSum
+    num_bytes: 212111
+    num_examples: 200
+  - name: NanoSummScreenFD
+    num_bytes: 217098
+    num_examples: 200
+  download_size: 130031
+  dataset_size: 1197334
+- config_name: corpus
+  features:
+  - name: _id
+    dtype: string
+  - name: text
+    dtype: string
+  splits:
+  - name: Nano2WikiMultihopQA
+    num_bytes: 11283128
+    num_examples: 300
+  - name: NanoNarrativeQA
+    num_bytes: 116191265
+    num_examples: 355
+  - name: NanoNeedle
+    num_bytes: 28226538
+    num_examples: 800
+  - name: NanoPasskey
+    num_bytes: 23182064
+    num_examples: 800
+  - name: NanoQMSum
+    num_bytes: 10515610
+    num_examples: 197
+  - name: NanoSummScreenFD
+    num_bytes: 10373940
+    num_examples: 336
+  download_size: 95954462
+  dataset_size: 199772545
+- config_name: qrels
+  features:
+  - name: query-id
+    dtype: string
+  - name: corpus-id
+    dtype: string
+  splits:
+  - name: Nano2WikiMultihopQA
+    num_bytes: 4614
+    num_examples: 200
+  - name: NanoNarrativeQA
+    num_bytes: 4616
+    num_examples: 200
+  - name: NanoNeedle
+    num_bytes: 3362
+    num_examples: 98
+  - name: NanoPasskey
+    num_bytes: 3434
+    num_examples: 100
+  - name: NanoQMSum
+    num_bytes: 4576
+    num_examples: 200
+  - name: NanoSummScreenFD
+    num_bytes: 4631
+    num_examples: 200
+  download_size: 18531
+  dataset_size: 25233
+- config_name: queries
+  features:
+  - name: _id
+    dtype: string
+  - name: text
+    dtype: string
+  splits:
+  - name: Nano2WikiMultihopQA
+    num_bytes: 16812
+    num_examples: 200
+  - name: NanoNarrativeQA
+    num_bytes: 13155
+    num_examples: 200
+  - name: NanoNeedle
+    num_bytes: 7953
+    num_examples: 98
+  - name: NanoPasskey
+    num_bytes: 5997
+    num_examples: 100
+  - name: NanoQMSum
+    num_bytes: 92566
+    num_examples: 200
+  - name: NanoSummScreenFD
+    num_bytes: 123428
+    num_examples: 200
+  download_size: 165197
+  dataset_size: 259911
+---
+# NanoLongEmbed
+
+This dataset is a Nano-style retrieval dataset. Nano-series evaluation can
+be run easily with the [HAKARI Benchmark](https://github.com/hotchpotch/hakari-bench).
+
+NanoLongEmbed is derived from dwzhu/LongEmbed. 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).
+
+NanoLongEmbed contains Nano-style long-context retrieval splits derived from LongEmbed tasks.
+
+## Source Links
+
+- [LongEmbed source dataset](https://huggingface.co/datasets/dwzhu/LongEmbed)
+
+## LongEmbed Source Statistics
+
+The source LongEmbed tasks have small corpus sizes by design. This NanoLongEmbed
+build keeps the full source corpus for every task because every source corpus is
+below the 10,000 document cap.
+
+| Nano split | Source task | Domain | Source queries | Source docs | Nano queries | Nano docs |
+|---|---|---|---:|---:|---:|---:|
+| `NanoNarrativeQA` | NarrativeQA | Literature, File | 10,449 | 355 | 200 | 355 |
+| `NanoQMSum` | QMSum | Meeting | 1,527 | 197 | 200 | 197 |
+| `Nano2WikiMultihopQA` | 2WikimQA | Wikipedia | 300 | 300 | 200 | 300 |
+| `NanoSummScreenFD` | SummScreenFD | ScreenWriting | 336 | 336 | 200 | 336 |
+| `NanoPasskey` | Passkey | Synthetic | 400 | 800 | 100 | 800 |
+| `NanoNeedle` | Needle | Synthetic | 400 | 800 | 98 | 800 |
+
+No corpus documents were removed as duplicates in this build:
+
+- duplicate source document ids: 0 for every split
+- duplicate source document texts: 0 for every split
+- corpus text rows skipped as duplicates during generation: 0 for every split
+- qrels rewrites due to duplicate corpus text: 0 for every split
+
+`NanoPasskey` and `NanoNeedle` have fewer than 200 queries because duplicate
+query text is removed within each split. The source `passkey` split has 400
+query rows but 100 unique query texts; the source `needle` split has 400 query
+rows but 98 unique query texts.
+
+## 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 has the same Nano split names. The exact parquet paths are defined
+in the dataset card metadata above. If a regenerated 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 dwzhu/LongEmbed as the upstream benchmark or dataset family.
+2. Load source datasets from `dwzhu/LongEmbed` revision `10039a580487dacecf79db69166e17ace3ede392`.
+3. Use the source evaluation split selected by the generator for each retrieval task.
+4. Create one Nano split for each selected source retrieval task.
+5. Keep up to 200 eligible queries per Nano split.
+6. Include all 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 retained document id.
+9. Store corpus text in the generated document `text` field.
+10. Generate BM25 top-100 candidates with `unknown:Qwen/Qwen3-0.6B` tokenization, or the per-split tokenizer shown below.
+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.
+
+## 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 up to 100 corpus ids.
+
+BM25 candidates are generated from the selected corpus for each split. When a
+qrels-positive document is not present in the raw BM25 top-100 results, 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 |
+|---|---|---|---:|---:|---:|
+| `NanoNarrativeQA` | narrativeqa | dwzhu/LongEmbed | 200 | 355 | 200 |
+| `NanoSummScreenFD` | summ_screen_fd | dwzhu/LongEmbed | 200 | 336 | 200 |
+| `NanoQMSum` | qmsum | dwzhu/LongEmbed | 200 | 197 | 200 |
+| `Nano2WikiMultihopQA` | 2wikimqa | dwzhu/LongEmbed | 200 | 300 | 200 |
+| `NanoPasskey` | passkey | dwzhu/LongEmbed | 100 | 800 | 100 |
+| `NanoNeedle` | needle | dwzhu/LongEmbed | 98 | 800 | 98 |
+
+## BM25 nDCG@10
+
+`nDCG@10` is computed from the included BM25 ranking against the included qrels.
+
+| Nano split | Tokenizer | Forced BM25 positives | BM25 nDCG@10 |
+|---|---|---:|---:|
+| `NanoNarrativeQA` | `stemmer:en/english` | 21 | 0.6910 |
+| `NanoSummScreenFD` | `stemmer:en/english` | 0 | 0.9746 |
+| `NanoQMSum` | `stemmer:en/english` | 0 | 0.7132 |
+| `Nano2WikiMultihopQA` | `stemmer:en/english` | 0 | 0.9515 |
+| `NanoPasskey` | `stemmer:en/english` | 1 | 0.7506 |
+| `NanoNeedle` | `stemmer:en/english` | 2 | 0.6852 |
+
+## Skipped Tasks
+
+None.
+
+## License
+
+NanoLongEmbed is a derived dataset. Users must comply with the licenses, terms, and
+attribution requirements of the upstream datasets listed above.

bm25/Nano2WikiMultihopQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32b860f728774181b5c19358ba0597cfacfa7b9175cee181491e602593bff3e7
+size 26331

bm25/Nano2WikiMultihopQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9408468c309d1cb86a039208e191697b17078d713bb05a6d4b866817fc1b1e5e
+size 8500

bm25/NanoNarrativeQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e5b8099d3e8396303aebdbcd8d877c5c6b06112e7a887bb839217dc1cc1edd2
+size 26581

bm25/NanoNarrativeQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d816c6c76bd582111b2906a15c4f0888fc0bf9f8c01c936a7b635d20c43a5254
+size 8750

bm25/NanoNeedle-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be94e25408e4ebe764858596f9fadda45b9c363438e42293cfae8758becb151e
+size 17138

bm25/NanoNeedle.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b73d93d7f5090f5ec1be7444029ffb27c6792e38f3c853b6d71fb54ce46e522d
+size 11330

bm25/NanoPasskey-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:380d9a96cd18c689e0fa27320044c15791ec34a163fe00fbfc8a84a0418d0ea5
+size 10059

bm25/NanoPasskey.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85f01bdb0188dc67200274f6f5674200652e29966c917fa8ba86cdbb2c51b223
+size 6740

bm25/NanoQMSum-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6451051bb1a9cbe2a49ed21241e5d8750d0c5961ebb50254ad0f304ac339d376
+size 23431

bm25/NanoQMSum.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92e07266179e998230af96574eb069502596785885dbb49c81f7c6c7bac85e1b
+size 7473

bm25/NanoSummScreenFD-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd677e0bc794526278dba94667d755293f93435bb8d5159a3e307a9b264bfb19
+size 26491

bm25/NanoSummScreenFD.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ee2541b96ce29352faf7e41d0143839b27b35d2538ef60959d7b66abd4489b4
+size 8651

build_nanolongembed.py

@@ -0,0 +1,485 @@
+from __future__ import annotations
+
+from collections import defaultdict
+from dataclasses import asdict
+from pathlib import Path
+import json
+import sys
+from typing import Any
+
+from datasets import Dataset, load_dataset
+from huggingface_hub import HfApi
+
+PROJECT_ROOT = Path(__file__).resolve().parents[3]
+if str(PROJECT_ROOT) not in sys.path:
+    sys.path.insert(0, str(PROJECT_ROOT))
+
+from nano_ir_eval.bm25_subset import (  # noqa: E402
+    DEFAULT_TRANSFORMER_TOKENIZER,
+    compute_ndcg_at_k,
+    compute_qrels_coverage,
+    detect_primary_language,
+    generate_bm25_rows,
+    plan_tokenization,
+)
+
+
+SOURCE_DATASET_ID = "dwzhu/LongEmbed"
+OUTPUT_DIR = Path(__file__).resolve().parent
+QUERY_LIMIT = 50
+CORPUS_LIMIT = 10_000
+TOP_K = 100
+LANGUAGE_SEED = 13
+LANGUAGE_SAMPLE_SIZE = 50
+
+TASKS = [
+    ("narrativeqa", "NanoNarrativeQA", "first_valid_queries"),
+    ("summ_screen_fd", "NanoSummScreenFD", "first_valid_queries"),
+    ("qmsum", "NanoQMSum", "first_valid_queries"),
+    ("2wikimqa", "Nano2WikiMultihopQA", "first_valid_queries"),
+    ("passkey", "NanoPasskey", "balanced_context_length"),
+    ("needle", "NanoNeedle", "balanced_context_length"),
+]
+
+
+def _clean_text(value: Any) -> str:
+    return str(value).strip()
+
+
+def _source_revision() -> str | None:
+    try:
+        return str(HfApi().dataset_info(SOURCE_DATASET_ID).sha)
+    except Exception:
+        return None
+
+
+def _load_source_task(config_name: str) -> tuple[list[dict[str, Any]], list[dict[str, Any]], list[dict[str, Any]]]:
+    corpus = list(load_dataset(SOURCE_DATASET_ID, config_name, split="corpus"))
+    queries = list(load_dataset(SOURCE_DATASET_ID, config_name, split="queries"))
+    qrels = list(load_dataset(SOURCE_DATASET_ID, config_name, split="qrels"))
+    return corpus, queries, qrels
+
+
+def _qrels_by_query(qrels: list[dict[str, Any]]) -> dict[str, list[str]]:
+    result: dict[str, list[str]] = defaultdict(list)
+    for row in qrels:
+        query_id = str(row["qid"])
+        corpus_id = str(row["doc_id"])
+        if query_id and corpus_id:
+            result[query_id].append(corpus_id)
+    return dict(result)
+
+
+def _select_first_valid_queries(
+    queries: list[dict[str, Any]],
+    qrels_for_query: dict[str, list[str]],
+    corpus_ids: set[str],
+) -> list[dict[str, Any]]:
+    selected: list[dict[str, Any]] = []
+    for row in queries:
+        query_id = str(row["qid"])
+        if not query_id or not _clean_text(row["text"]):
+            continue
+        positives = qrels_for_query.get(query_id, [])
+        if not positives or not all(corpus_id in corpus_ids for corpus_id in positives):
+            continue
+        selected.append(row)
+        if len(selected) >= QUERY_LIMIT:
+            break
+    return selected
+
+
+def _select_balanced_context_queries(
+    queries: list[dict[str, Any]],
+    qrels_for_query: dict[str, list[str]],
+    corpus_ids: set[str],
+) -> list[dict[str, Any]]:
+    groups: dict[int, list[dict[str, Any]]] = defaultdict(list)
+    for row in queries:
+        query_id = str(row["qid"])
+        if not query_id or not _clean_text(row["text"]):
+            continue
+        positives = qrels_for_query.get(query_id, [])
+        if not positives or not all(corpus_id in corpus_ids for corpus_id in positives):
+            continue
+        groups[int(row["context_length"])].append(row)
+
+    selected: list[dict[str, Any]] = []
+    context_lengths = sorted(groups)
+    index = 0
+    while len(selected) < QUERY_LIMIT:
+        added = False
+        for context_length in context_lengths:
+            bucket = groups[context_length]
+            if index < len(bucket):
+                selected.append(bucket[index])
+                added = True
+                if len(selected) >= QUERY_LIMIT:
+                    break
+        if not added:
+            break
+        index += 1
+    return selected
+
+
+def _build_nano_split(config_name: str, split_name: str, selection_policy: str) -> dict[str, Any]:
+    source_corpus, source_queries, source_qrels = _load_source_task(config_name)
+    corpus_ids = {str(row["doc_id"]) for row in source_corpus}
+    qrels_for_query = _qrels_by_query(source_qrels)
+
+    if selection_policy == "balanced_context_length":
+        selected_queries = _select_balanced_context_queries(source_queries, qrels_for_query, corpus_ids)
+    else:
+        selected_queries = _select_first_valid_queries(source_queries, qrels_for_query, corpus_ids)
+
+    if len(selected_queries) != QUERY_LIMIT:
+        raise RuntimeError(f"{split_name}: selected {len(selected_queries)} queries, expected {QUERY_LIMIT}.")
+
+    selected_query_ids = [str(row["qid"]) for row in selected_queries]
+    selected_query_id_set = set(selected_query_ids)
+    selected_qrels = [
+        {"query-id": query_id, "corpus-id": corpus_id}
+        for query_id in selected_query_ids
+        for corpus_id in qrels_for_query[query_id]
+    ]
+    positive_corpus_ids = {row["corpus-id"] for row in selected_qrels}
+
+    selected_corpus: list[dict[str, str]] = []
+    seen_corpus_ids: set[str] = set()
+    seen_texts: set[str] = set()
+    duplicate_text_skipped = 0
+    for row in source_corpus:
+        corpus_id = str(row["doc_id"])
+        text = _clean_text(row["text"])
+        if not corpus_id or not text:
+            continue
+        if corpus_id in seen_corpus_ids:
+            continue
+        if text in seen_texts:
+            duplicate_text_skipped += 1
+            if corpus_id not in positive_corpus_ids:
+                continue
+        selected_corpus.append({"_id": corpus_id, "text": text})
+        seen_corpus_ids.add(corpus_id)
+        seen_texts.add(text)
+        if len(selected_corpus) >= CORPUS_LIMIT:
+            break
+
+    selected_corpus_ids = {row["_id"] for row in selected_corpus}
+    missing_positive_ids = positive_corpus_ids - selected_corpus_ids
+    if missing_positive_ids:
+        raise RuntimeError(f"{split_name}: qrels positives missing from selected corpus: {sorted(missing_positive_ids)[:5]}")
+
+    nano_queries = [{"_id": str(row["qid"]), "text": _clean_text(row["text"])} for row in selected_queries]
+
+    if len({row["_id"] for row in nano_queries}) != len(nano_queries):
+        raise RuntimeError(f"{split_name}: duplicate query IDs.")
+    if len({row["_id"] for row in selected_corpus}) != len(selected_corpus):
+        raise RuntimeError(f"{split_name}: duplicate corpus IDs.")
+    if any(row["query-id"] not in selected_query_id_set for row in selected_qrels):
+        raise RuntimeError(f"{split_name}: qrels reference an unselected query.")
+    if any(row["corpus-id"] not in selected_corpus_ids for row in selected_qrels):
+        raise RuntimeError(f"{split_name}: qrels reference an unselected document.")
+
+    output_paths = {
+        "corpus": OUTPUT_DIR / "corpus" / f"{split_name}.parquet",
+        "queries": OUTPUT_DIR / "queries" / f"{split_name}.parquet",
+        "qrels": OUTPUT_DIR / "qrels" / f"{split_name}.parquet",
+    }
+    for path in output_paths.values():
+        path.parent.mkdir(parents=True, exist_ok=True)
+
+    Dataset.from_list(selected_corpus).to_parquet(str(output_paths["corpus"]))
+    Dataset.from_list(nano_queries).to_parquet(str(output_paths["queries"]))
+    Dataset.from_list(selected_qrels).to_parquet(str(output_paths["qrels"]))
+
+    context_counts: dict[str, int] | None = None
+    if selected_queries and "context_length" in selected_queries[0]:
+        counts: dict[str, int] = defaultdict(int)
+        for row in selected_queries:
+            counts[str(row["context_length"])] += 1
+        context_counts = dict(sorted(counts.items(), key=lambda item: int(item[0])))
+
+    return {
+        "source_config": config_name,
+        "split_name": split_name,
+        "selection_policy": selection_policy,
+        "source_query_count": len(source_queries),
+        "source_corpus_count": len(source_corpus),
+        "source_qrels_count": len(source_qrels),
+        "selected_query_count": len(nano_queries),
+        "selected_corpus_count": len(selected_corpus),
+        "qrels_count": len(selected_qrels),
+        "duplicate_text_skipped": duplicate_text_skipped,
+        "qrels_rewrite_count": 0,
+        "context_length_query_counts": context_counts,
+    }
+
+
+def _force_qrels_positive_candidates(
+    rows: list[dict[str, Any]],
+    qrels: list[dict[str, Any]],
+    *,
+    top_k: int,
+    corpus_size: int,
+) -> tuple[list[dict[str, Any]], int]:
+    positives_by_query: dict[str, list[str]] = defaultdict(list)
+    for row in qrels:
+        positives_by_query[str(row["query-id"])].append(str(row["corpus-id"]))
+
+    effective_top_k = min(top_k, corpus_size)
+    forced_count = 0
+    forced_rows: list[dict[str, Any]] = []
+    for row in rows:
+        query_id = str(row["query-id"])
+        positives = positives_by_query.get(query_id, [])
+        positive_set = set(positives)
+        candidates = []
+        seen: set[str] = set()
+        for corpus_id in row["corpus-ids"]:
+            corpus_id = str(corpus_id)
+            if corpus_id not in seen:
+                candidates.append(corpus_id)
+                seen.add(corpus_id)
+
+        for positive_id in positives:
+            if positive_id in seen:
+                continue
+            forced_count += 1
+            if len(candidates) < effective_top_k:
+                candidates.append(positive_id)
+                seen.add(positive_id)
+                continue
+            for index in range(len(candidates) - 1, -1, -1):
+                if candidates[index] not in positive_set:
+                    seen.remove(candidates[index])
+                    candidates[index] = positive_id
+                    seen.add(positive_id)
+                    break
+            else:
+                raise RuntimeError(f"{query_id}: cannot force all positives within top-{effective_top_k}.")
+
+        forced_rows.append({"query-id": query_id, "corpus-ids": candidates[:effective_top_k]})
+    return forced_rows, forced_count
+
+
+def _build_bm25_for_split(split_name: str) -> dict[str, Any]:
+    corpus = list(Dataset.from_parquet(str(OUTPUT_DIR / "corpus" / f"{split_name}.parquet")))
+    queries = list(Dataset.from_parquet(str(OUTPUT_DIR / "queries" / f"{split_name}.parquet")))
+    qrels = list(Dataset.from_parquet(str(OUTPUT_DIR / "qrels" / f"{split_name}.parquet")))
+
+    detection = detect_primary_language(
+        [str(row["text"]) for row in corpus],
+        sample_size=LANGUAGE_SAMPLE_SIZE,
+        seed=LANGUAGE_SEED,
+    )
+    plan = plan_tokenization(
+        detection=detection,
+        splitter_mode="auto",
+        tokenizer_name=DEFAULT_TRANSFORMER_TOKENIZER,
+        language_hint="en",
+    )
+
+    raw_rows = generate_bm25_rows(
+        corpus=corpus,
+        queries=queries,
+        plan=plan,
+        top_k=TOP_K,
+        show_progress=False,
+    )
+    rows, forced_count = _force_qrels_positive_candidates(
+        raw_rows,
+        qrels,
+        top_k=TOP_K,
+        corpus_size=len(corpus),
+    )
+    coverage = compute_qrels_coverage(bm25_rows=rows, qrels=qrels, top_k=TOP_K)
+    if coverage.recall != 1.0:
+        raise RuntimeError(f"{split_name}: BM25 qrels coverage is {coverage.recall}, expected 1.0.")
+
+    bm25_path = OUTPUT_DIR / "bm25" / f"{split_name}.parquet"
+    bm25_path.parent.mkdir(parents=True, exist_ok=True)
+    Dataset.from_list(rows).to_parquet(str(bm25_path))
+
+    ndcg_at_10 = compute_ndcg_at_k(bm25_rows=rows, qrels=qrels, k=min(10, TOP_K))
+    ndcg_at_100 = compute_ndcg_at_k(bm25_rows=rows, qrels=qrels, k=min(100, TOP_K))
+    return {
+        "split_name": split_name,
+        "language_detection": asdict(detection),
+        "tokenization_plan": asdict(plan),
+        "top_k": TOP_K,
+        "qrels_coverage": asdict(coverage),
+        "forced_positive_count": forced_count,
+        "ndcg_at_10": ndcg_at_10,
+        "ndcg_at_100": ndcg_at_100,
+    }
+
+
+def _write_readme(split_names: list[str]) -> None:
+    config_order = ["bm25", "corpus", "qrels", "queries"]
+    lines = ["---", "configs:"]
+    for config_name in config_order:
+        lines.append(f"- config_name: {config_name}")
+        lines.append("  data_files:")
+        for split_name in split_names:
+            lines.append(f"  - split: {split_name}")
+            lines.append(f"    path: {config_name}/{split_name}.parquet")
+        if config_name == "queries":
+            lines.append("  default: true")
+    lines.extend(
+        [
+            "language:",
+            "- en",
+            "tags:",
+            "- Long Context",
+            "- retrieval",
+            "- nano",
+            "---",
+            "",
+            "# NanoLongEmbed",
+            "",
+            "NanoLongEmbed is a Nano-style retrieval subset derived from `dwzhu/LongEmbed`.",
+            "It keeps the NanoBEIR-compatible config layout: `corpus`, `queries`, `qrels`, and `bm25`.",
+            "",
+            "## Source",
+            "",
+            "- Source dataset: `dwzhu/LongEmbed`",
+            "- Source tasks: NarrativeQA, SummScreenFD, QMSum, 2WikiMultihopQA, Passkey, Needle",
+            "- Upstream card: https://huggingface.co/datasets/dwzhu/LongEmbed",
+            "",
+            "## Extraction Policy",
+            "",
+            "- 50 queries are selected per task.",
+            "- Real-world tasks use the first valid source queries in source order.",
+            "- Passkey and Needle use deterministic round-robin selection across context lengths.",
+            "- The full source corpus is retained for each task because every corpus has fewer than 10,000 documents.",
+            "- Exact duplicate corpus text is skipped when present; no duplicate corpus text was found in this build.",
+            "- Qrels are limited to the selected queries and retain only `query-id` and `corpus-id`.",
+            "",
+            "## BM25",
+            "",
+            "- BM25 candidates are top-100 per query.",
+            "- Tokenization uses the repository BM25 auto plan with English as a language hint.",
+            "- Any missing qrels-positive document is forced into the candidate list by replacing tail non-positive candidates.",
+            "- Per-split metadata and reproducibility settings are in `metadata/*.json` and `nano_bm25_subset_config.json`.",
+            "",
+            "## Schemas",
+            "",
+            "- `corpus`: `_id: string`, `text: string`",
+            "- `queries`: `_id: string`, `text: string`",
+            "- `qrels`: `query-id: string`, `corpus-id: string`",
+            "- `bm25`: `query-id: string`, `corpus-ids: list[string]`",
+            "",
+            "## License",
+            "",
+            "This derived local dataset does not assign a new license. Users must comply with the upstream LongEmbed dataset and source-data terms.",
+            "",
+        ]
+    )
+    (OUTPUT_DIR / "README.md").write_text("\n".join(lines), encoding="utf-8")
+
+
+def main() -> None:
+    OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+    source_revision = _source_revision()
+    split_metadata = []
+    bm25_metadata = []
+    split_names = [split_name for _, split_name, _ in TASKS]
+
+    for source_config, split_name, selection_policy in TASKS:
+        print(f"Building nano split {split_name} from {source_config}", flush=True)
+        split_metadata.append(_build_nano_split(source_config, split_name, selection_policy))
+
+    for split_name in split_names:
+        print(f"Building BM25 split {split_name}", flush=True)
+        bm25_metadata.append(_build_bm25_for_split(split_name))
+
+    metadata_by_split = {item["split_name"]: item for item in split_metadata}
+    for item in bm25_metadata:
+        split_name = item["split_name"]
+        metadata_by_split[split_name]["bm25"] = item
+
+    metadata_dir = OUTPUT_DIR / "metadata"
+    metadata_dir.mkdir(parents=True, exist_ok=True)
+    for split_name, metadata in metadata_by_split.items():
+        metadata["source_dataset_id"] = SOURCE_DATASET_ID
+        metadata["source_revision"] = source_revision
+        (metadata_dir / f"{split_name}.json").write_text(
+            json.dumps(metadata, ensure_ascii=False, indent=2),
+            encoding="utf-8",
+        )
+
+    manifest = {
+        "dataset_name": "NanoLongEmbed",
+        "source_dataset_id": SOURCE_DATASET_ID,
+        "source_revision": source_revision,
+        "output_dir": str(OUTPUT_DIR),
+        "query_limit_per_split": QUERY_LIMIT,
+        "corpus_limit_per_split": CORPUS_LIMIT,
+        "bm25_top_k": TOP_K,
+        "language_seed": LANGUAGE_SEED,
+        "language_sample_size": LANGUAGE_SAMPLE_SIZE,
+        "split_mapping": [
+            {"source_config": source_config, "split_name": split_name, "selection_policy": selection_policy}
+            for source_config, split_name, selection_policy in TASKS
+        ],
+        "counts": [
+            {
+                "split_name": item["split_name"],
+                "queries": item["selected_query_count"],
+                "corpus": item["selected_corpus_count"],
+                "qrels": item["qrels_count"],
+            }
+            for item in split_metadata
+        ],
+    }
+    (OUTPUT_DIR / "manifest.json").write_text(json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8")
+
+    bm25_summary = {
+        "dataset_id": str(OUTPUT_DIR),
+        "source_dataset_id": SOURCE_DATASET_ID,
+        "source_revision": source_revision,
+        "output_dir": str(OUTPUT_DIR),
+        "corpus_subset_name": "corpus",
+        "queries_subset_name": "queries",
+        "qrels_subset_name": "qrels",
+        "top_k": TOP_K,
+        "sample_size": LANGUAGE_SAMPLE_SIZE,
+        "language_seed": LANGUAGE_SEED,
+        "auto_select_best_splitter": False,
+        "selection_ndcg_k": None,
+        "default_tokenization_config": {
+            "splitter_mode": "auto",
+            "tokenizer_name": DEFAULT_TRANSFORMER_TOKENIZER,
+            "stemmer_algorithm": None,
+            "enable_stemming": True,
+        },
+        "positive_forcing": "replace tail non-positive candidates with missing qrels positives",
+        "splits": [
+            {
+                "split_name": item["split_name"],
+                "tokenization_plan": item["tokenization_plan"],
+                "main_score_name": None,
+                "main_score": None,
+                "selected_evaluation": None,
+                "candidate_evaluations": None,
+                "qrels_coverage": item["qrels_coverage"],
+                "forced_positive_count": item["forced_positive_count"],
+                "ndcg_at_10": item["ndcg_at_10"],
+                "ndcg_at_100": item["ndcg_at_100"],
+            }
+            for item in bm25_metadata
+        ],
+    }
+    (OUTPUT_DIR / "nano_bm25_subset_config.json").write_text(
+        json.dumps(bm25_summary, ensure_ascii=False, indent=2),
+        encoding="utf-8",
+    )
+
+    _write_readme(split_names)
+
+    print(f"Wrote NanoLongEmbed to {OUTPUT_DIR}", flush=True)
+
+
+if __name__ == "__main__":
+    main()

corpus/Nano2WikiMultihopQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b341c7aafd2a9de0fc7b8249fc01c18d382f10f3d2844db3a1426688905ecf92
+size 6695117

corpus/Nano2WikiMultihopQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b341c7aafd2a9de0fc7b8249fc01c18d382f10f3d2844db3a1426688905ecf92
+size 6695117

corpus/NanoNarrativeQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9104e9f90ecd3d18327a75f15ef7fdff669a26702c9e456520b170dbb65ed5c7
+size 63177746

corpus/NanoNarrativeQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9104e9f90ecd3d18327a75f15ef7fdff669a26702c9e456520b170dbb65ed5c7
+size 63177746

corpus/NanoNeedle-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54d9ac0bf63886f7d64be4fcbd0bbcdb77304ccdbcab0ef38eee6e7ebee46ddc
+size 14101095

corpus/NanoNeedle.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54d9ac0bf63886f7d64be4fcbd0bbcdb77304ccdbcab0ef38eee6e7ebee46ddc
+size 14101095

corpus/NanoPasskey-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e5d8318ffb237f9081c2884aadaced87ecff166de736daaf10cdfdd6b41ed37
+size 1165218

corpus/NanoPasskey.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e5d8318ffb237f9081c2884aadaced87ecff166de736daaf10cdfdd6b41ed37
+size 1165218

corpus/NanoQMSum-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed0360982cffa94d7e0281c504e63bb5727cc639c63c149d468d0aa7041e13bd
+size 4798098

corpus/NanoQMSum.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed0360982cffa94d7e0281c504e63bb5727cc639c63c149d468d0aa7041e13bd
+size 4798098

corpus/NanoSummScreenFD-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d69a4df3af4a2079bea2927c3823d51467f6ade8d0ddf7f460465cd20eb57a04
+size 6017188

corpus/NanoSummScreenFD.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d69a4df3af4a2079bea2927c3823d51467f6ade8d0ddf7f460465cd20eb57a04
+size 6017188

manifest.json

@@ -0,0 +1,81 @@
+{
+  "dataset_name": "NanoLongEmbed",
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392",
+  "output_dir": "/home/hotchpotch/src/github.com/hotchpotch/nano-ir-eval/output/nano_datasets/NanoLongEmbed",
+  "query_limit_per_split": 50,
+  "corpus_limit_per_split": 10000,
+  "bm25_top_k": 100,
+  "language_seed": 13,
+  "language_sample_size": 50,
+  "split_mapping": [
+    {
+      "source_config": "narrativeqa",
+      "split_name": "NanoNarrativeQA",
+      "selection_policy": "first_valid_queries"
+    },
+    {
+      "source_config": "summ_screen_fd",
+      "split_name": "NanoSummScreenFD",
+      "selection_policy": "first_valid_queries"
+    },
+    {
+      "source_config": "qmsum",
+      "split_name": "NanoQMSum",
+      "selection_policy": "first_valid_queries"
+    },
+    {
+      "source_config": "2wikimqa",
+      "split_name": "Nano2WikiMultihopQA",
+      "selection_policy": "first_valid_queries"
+    },
+    {
+      "source_config": "passkey",
+      "split_name": "NanoPasskey",
+      "selection_policy": "balanced_context_length"
+    },
+    {
+      "source_config": "needle",
+      "split_name": "NanoNeedle",
+      "selection_policy": "balanced_context_length"
+    }
+  ],
+  "counts": [
+    {
+      "split_name": "NanoNarrativeQA",
+      "queries": 50,
+      "corpus": 355,
+      "qrels": 50
+    },
+    {
+      "split_name": "NanoSummScreenFD",
+      "queries": 50,
+      "corpus": 336,
+      "qrels": 50
+    },
+    {
+      "split_name": "NanoQMSum",
+      "queries": 50,
+      "corpus": 197,
+      "qrels": 50
+    },
+    {
+      "split_name": "Nano2WikiMultihopQA",
+      "queries": 50,
+      "corpus": 300,
+      "qrels": 50
+    },
+    {
+      "split_name": "NanoPasskey",
+      "queries": 50,
+      "corpus": 800,
+      "qrels": 50
+    },
+    {
+      "split_name": "NanoNeedle",
+      "queries": 50,
+      "corpus": 800,
+      "qrels": 50
+    }
+  ]
+}

metadata/Nano2WikiMultihopQA.json

@@ -0,0 +1,42 @@
+{
+  "source_config": "2wikimqa",
+  "split_name": "Nano2WikiMultihopQA",
+  "selection_policy": "first_valid_queries",
+  "source_query_count": 300,
+  "source_corpus_count": 300,
+  "source_qrels_count": 300,
+  "selected_query_count": 50,
+  "selected_corpus_count": 300,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": null,
+  "bm25": {
+    "split_name": "Nano2WikiMultihopQA",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.8060238008596459,
+      "support_ratio": 0.98,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 0,
+    "ndcg_at_10": 0.9319038568095249,
+    "ndcg_at_100": 0.9406546234742748
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

metadata/NanoNarrativeQA.json

@@ -0,0 +1,42 @@
+{
+  "source_config": "narrativeqa",
+  "split_name": "NanoNarrativeQA",
+  "selection_policy": "first_valid_queries",
+  "source_query_count": 10449,
+  "source_corpus_count": 355,
+  "source_qrels_count": 10449,
+  "selected_query_count": 50,
+  "selected_corpus_count": 355,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": null,
+  "bm25": {
+    "split_name": "NanoNarrativeQA",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.6008472263813018,
+      "support_ratio": 1.0,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 4,
+    "ndcg_at_10": 0.6081141940840378,
+    "ndcg_at_100": 0.6638750650343473
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

metadata/NanoNeedle.json

@@ -0,0 +1,51 @@
+{
+  "source_config": "needle",
+  "split_name": "NanoNeedle",
+  "selection_policy": "balanced_context_length",
+  "source_query_count": 400,
+  "source_corpus_count": 800,
+  "source_qrels_count": 400,
+  "selected_query_count": 50,
+  "selected_corpus_count": 800,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": {
+    "256": 7,
+    "512": 7,
+    "1024": 6,
+    "2048": 6,
+    "4096": 6,
+    "8192": 6,
+    "16384": 6,
+    "32768": 6
+  },
+  "bm25": {
+    "split_name": "NanoNeedle",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.8917386984825134,
+      "support_ratio": 1.0,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 2,
+    "ndcg_at_10": 0.3839632728352887,
+    "ndcg_at_100": 0.43128913271775404
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

metadata/NanoPasskey.json

@@ -0,0 +1,51 @@
+{
+  "source_config": "passkey",
+  "split_name": "NanoPasskey",
+  "selection_policy": "balanced_context_length",
+  "source_query_count": 400,
+  "source_corpus_count": 800,
+  "source_qrels_count": 400,
+  "selected_query_count": 50,
+  "selected_corpus_count": 800,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": {
+    "256": 7,
+    "512": 7,
+    "1024": 6,
+    "2048": 6,
+    "4096": 6,
+    "8192": 6,
+    "16384": 6,
+    "32768": 6
+  },
+  "bm25": {
+    "split_name": "NanoPasskey",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.9547083139419555,
+      "support_ratio": 1.0,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 0,
+    "ndcg_at_10": 0.4924688774379761,
+    "ndcg_at_100": 0.502481048776537
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

metadata/NanoQMSum.json

@@ -0,0 +1,42 @@
+{
+  "source_config": "qmsum",
+  "split_name": "NanoQMSum",
+  "selection_policy": "first_valid_queries",
+  "source_query_count": 1527,
+  "source_corpus_count": 197,
+  "source_qrels_count": 1527,
+  "selected_query_count": 50,
+  "selected_corpus_count": 197,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": null,
+  "bm25": {
+    "split_name": "NanoQMSum",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.7766384822130203,
+      "support_ratio": 1.0,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 0,
+    "ndcg_at_10": 0.7921034475051778,
+    "ndcg_at_100": 0.815163333552449
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

metadata/NanoSummScreenFD.json

@@ -0,0 +1,42 @@
+{
+  "source_config": "summ_screen_fd",
+  "split_name": "NanoSummScreenFD",
+  "selection_policy": "first_valid_queries",
+  "source_query_count": 336,
+  "source_corpus_count": 336,
+  "source_qrels_count": 336,
+  "selected_query_count": 50,
+  "selected_corpus_count": 336,
+  "qrels_count": 50,
+  "duplicate_text_skipped": 0,
+  "qrels_rewrite_count": 0,
+  "context_length_query_counts": null,
+  "bm25": {
+    "split_name": "NanoSummScreenFD",
+    "language_detection": {
+      "language": "en",
+      "confidence": 0.7837093853950501,
+      "support_ratio": 1.0,
+      "ambiguous": false,
+      "sample_count": 50
+    },
+    "tokenization_plan": {
+      "mode": "stemmer",
+      "language": "en",
+      "stemmer_algorithm": "english",
+      "tokenizer_name": null,
+      "reason": "language 'en' mapped to stemmer 'english'"
+    },
+    "top_k": 100,
+    "qrels_coverage": {
+      "total": 50,
+      "hits": 50,
+      "recall": 1.0
+    },
+    "forced_positive_count": 0,
+    "ndcg_at_10": 0.9678557852142875,
+    "ndcg_at_100": 0.9678557852142875
+  },
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392"
+}

nano_bm25_subset_config.json

@@ -0,0 +1,155 @@
+{
+  "dataset_id": "/home/hotchpotch/src/github.com/hotchpotch/nano-ir-eval/output/nano_datasets/NanoLongEmbed",
+  "source_dataset_id": "dwzhu/LongEmbed",
+  "source_revision": "10039a580487dacecf79db69166e17ace3ede392",
+  "output_dir": "/home/hotchpotch/src/github.com/hotchpotch/nano-ir-eval/output/nano_datasets/NanoLongEmbed",
+  "corpus_subset_name": "corpus",
+  "queries_subset_name": "queries",
+  "qrels_subset_name": "qrels",
+  "top_k": 100,
+  "sample_size": 50,
+  "language_seed": 13,
+  "auto_select_best_splitter": false,
+  "selection_ndcg_k": null,
+  "default_tokenization_config": {
+    "splitter_mode": "auto",
+    "tokenizer_name": "Qwen/Qwen3-0.6B",
+    "stemmer_algorithm": null,
+    "enable_stemming": true
+  },
+  "positive_forcing": "replace tail non-positive candidates with missing qrels positives",
+  "splits": [
+    {
+      "split_name": "NanoNarrativeQA",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 4,
+      "ndcg_at_10": 0.6081141940840378,
+      "ndcg_at_100": 0.6638750650343473
+    },
+    {
+      "split_name": "NanoSummScreenFD",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 0,
+      "ndcg_at_10": 0.9678557852142875,
+      "ndcg_at_100": 0.9678557852142875
+    },
+    {
+      "split_name": "NanoQMSum",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 0,
+      "ndcg_at_10": 0.7921034475051778,
+      "ndcg_at_100": 0.815163333552449
+    },
+    {
+      "split_name": "Nano2WikiMultihopQA",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 0,
+      "ndcg_at_10": 0.9319038568095249,
+      "ndcg_at_100": 0.9406546234742748
+    },
+    {
+      "split_name": "NanoPasskey",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 0,
+      "ndcg_at_10": 0.4924688774379761,
+      "ndcg_at_100": 0.502481048776537
+    },
+    {
+      "split_name": "NanoNeedle",
+      "tokenization_plan": {
+        "mode": "stemmer",
+        "language": "en",
+        "stemmer_algorithm": "english",
+        "tokenizer_name": null,
+        "reason": "language 'en' mapped to stemmer 'english'"
+      },
+      "main_score_name": null,
+      "main_score": null,
+      "selected_evaluation": null,
+      "candidate_evaluations": null,
+      "qrels_coverage": {
+        "total": 50,
+        "hits": 50,
+        "recall": 1.0
+      },
+      "forced_positive_count": 2,
+      "ndcg_at_10": 0.3839632728352887,
+      "ndcg_at_100": 0.43128913271775404
+    }
+  ]
+}

qrels/Nano2WikiMultihopQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91798d58c063fa890172e6668d5443560377e5e9b83da995955c7c835d6f6614
+size 3394

qrels/Nano2WikiMultihopQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc8acbd4648e7d2f6af4b243a7d2ba32fff922b8303004b84d6c1585fe1f4803
+size 1805

qrels/NanoNarrativeQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3d16b7209b76b88f746a7ec264c6cdaab99e06666f7e412a6adb7525828c25f
+size 3283

qrels/NanoNarrativeQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d9e394db4dd0b34522ffd15cdd2ca7ac2f29ffdfe0e3d22c90cbb4bae8f4eab
+size 1792

qrels/NanoNeedle-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aec65028b6f844eba4fcfef9cf7966256d17ee2c6e3f67a4c573324eb6aabdb0
+size 2697

qrels/NanoNeedle.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e67bb799d76587c6b0f05cb3d05f4b9c92339acf8f401cdf9de9808269957840
+size 2094

qrels/NanoPasskey-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:498d68f321e5850f43a31910cea0baa8621dfcc7b6145629b1dc5815dd02f56f
+size 2719

qrels/NanoPasskey.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de1e6aa7208010befa074526f054b93d1e83ce5f32cf32535de7c84791e81e99
+size 2101

qrels/NanoQMSum-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8299b482abd97753a3e7372ca1020f0e0a548759d39d9408dadf39c3e7155df
+size 3041

qrels/NanoQMSum.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29f48478d8aaaef77b71dcd736c598302ad8507319e43b8b4210aa6219d84bb0
+size 1763

qrels/NanoSummScreenFD-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4c352a4f4f1c2ccdcf02f64cd8b805b05c2e999bb4ff866699fe99fb2ca74b7
+size 3397

qrels/NanoSummScreenFD.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba86b500834476527c7eb85d99f358c04ce8f4d08a6c0cb3eb70dde2952f29b6
+size 1806

queries/Nano2WikiMultihopQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cde4086a7420542ff101afc03770e4a6cdb540e8ec5b5523c15c37aac817a67e
+size 10626

queries/Nano2WikiMultihopQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50b3f3e7ee7d40147ee9c0ce395a4228bbca19324d443e0c2108afc022858827
+size 3829

queries/NanoNarrativeQA-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cccda52c46e2564e7d4dfeec3789d156d48022d5e1920109f90c6af521311c22
+size 10492

queries/NanoNarrativeQA.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:430ab576a0fdb48b2a646d5bba43660d8b12684b693bc440cec9ea28acfd3b20
+size 3784