bactrianus/bactrainus-hotpotqa

<div style="font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif; border: 1px solid #cbd5e1; border-radius: 16px; overflow: hidden; background: #ffffff; margin-bottom: 28px;"> <div style="background: linear-gradient(135deg, #1f4e79 0%, #12263a 100%); padding: 26px; color: white;"> <div style="display: flex; align-items: center; justify-content: space-between; flex-wrap: wrap; gap: 10px;"> <h1 style="margin: 0; font-size: 28px; font-weight: 800; color: white; border: none;">Bactrainus HotpotQA Training Suite</h1> <span style="background: #0f766e; color: white; font-size: 11px; font-weight: 700; padding: 5px 11px; border-radius: 20px; letter-spacing: 0.5px;">COMPLETE RELEASE</span> </div> <p style="margin: 9px 0 0 0; font-size: 15px; color: #dbeafe; font-weight: 500;">One canonical HotpotQA source, eight clean and ID-aligned training views.</p> </div> <div style="display: flex; gap: 8px; flex-wrap: wrap; padding: 13px 24px; background: #f8fafc; border-bottom: 1px solid #e2e8f0;"> <span style="background: #dbeafe; color: #1e40af; font-size: 11px; font-weight: 700; padding: 4px 10px; border-radius: 20px; border: 1px solid #bfdbfe;">90,447 SOURCE IDs</span> <span style="background: #ccfbf1; color: #115e59; font-size: 11px; font-weight: 700; padding: 4px 10px; border-radius: 20px; border: 1px solid #99f6e4;">8 CONFIGURATIONS</span> <span style="background: #fef3c7; color: #92400e; font-size: 11px; font-weight: 700; padding: 4px 10px; border-radius: 20px; border: 1px solid #fde68a;">SFT READY</span> <span style="background: #ede9fe; color: #5b21b6; font-size: 11px; font-weight: 700; padding: 4px 10px; border-radius: 20px; border: 1px solid #ddd6fe;">ID-ALIGNED</span> <span style="background: #dcfce7; color: #166534; font-size: 11px; font-weight: 700; padding: 4px 10px; border-radius: 20px; border: 1px solid #bbf7d0;">CC BY-SA 4.0</span> </div> <div style="padding: 22px 24px; color: #334155; line-height: 1.65; font-size: 14px;"> <p style="margin: 0;">The Bactrainus training suite turns the complete <strong>90,447-example</strong> HotpotQA <code>distractor/train</code> split into a canonical structured dataset and seven ready-to-train chat-formatted SFT configurations. Every row keeps the official HotpotQA identifier as <code>source_id</code>, so the same example can be followed across paragraph selection, decomposition, sentence selection, evidence-grounded reasoning, and answer generation without relying on row order.</p> <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(190px, 1fr)); gap: 12px; margin-top: 18px;"> <div style="border: 1px solid #bfdbfe; border-radius: 9px; background: #eff6ff; padding: 13px;"> <strong style="display: block; color: #1f4e79; margin-bottom: 4px;">Traceable provenance</strong> <span style="font-size: 13px;">Pinned upstream revision, machine-readable source manifest, and reviewed annotation patches.</span> </div> <div style="border: 1px solid #99f6e4; border-radius: 9px; background: #f0fdfa; padding: 13px;"> <strong style="display: block; color: #115e59; margin-bottom: 4px;">Complete modular coverage</strong> <span style="font-size: 13px;">Reader, CoT, paragraph selector, decomposer, sentence selector, and joint SFT data.</span> </div> <div style="border: 1px solid #fde68a; border-radius: 9px; background: #fffbeb; padding: 13px;"> <strong style="display: block; color: #92400e; margin-bottom: 4px;">Strict integrity checks</strong> <span style="font-size: 13px;">Schema, evidence bounds, task labels, row counts, ID equality, and 80 shard checksums.</span> </div> </div> </div> </div>

<p align="center"> <img src="assets/data.png" alt="Bactrainus HotpotQA training-data artwork" width="507"> </p>

At a glance

How the data follows the Bactrainus pipeline

The eight configurations expose the same evidence flow in progressively more specialized training views. The structured records preserve the question, all supplied candidate paragraphs, answer, and supporting facts. The SFT views then supervise paragraph selection, optional question decomposition, sentence-level evidence selection, and answer generation, either as separate modules or as a joint selector-reader task. Every view retains the same source_id, so examples can be joined without relying on row position.

<p align="center"> <img src="architecture.svg" alt="Bactrainus modular architecture" width="900"> </p>

<p align="center"><em>Canonical records are transformed into task-specific conversations while preserving a shared source identity across the modular pipeline.</em></p>

Load the data

from datasets import load_dataset

dataset = load_dataset(
    "bactrianus/bactrainus-hotpotqa",
    "cot-reader-sft",
    split="train",
    revision="v1.0.0",
)

print(dataset.num_rows)          # 90447
print(dataset[0]["source_id"])
print(dataset[0]["messages"])

For a reproducible experiment, pin the immutable release tag:

dataset = load_dataset(
    "bactrianus/bactrainus-hotpotqa",
    "structured",
    split="train",
    revision="v1.0.0",
)

Configurations

Each configuration contains exactly 90,447 unique source_id values.

The seven SFT views are deterministic transformations of structured. The CoT view is a traceable evidence trace: each step copies an annotated supporting sentence and records its exact title and zero-based sentence index. The decomposition view is generated from the paragraphs available after paragraph selection and is therefore complete, stable, and reproducible.

Record formats

Canonical structured record

{
  "source_id": "official-hotpotqa-id",
  "question": "...",
  "answer": "...",
  "question_type": "bridge",
  "difficulty": "hard",
  "candidate_paragraphs": [
    {"title": "...", "sentences": ["...", "..."]}
  ],
  "supporting_facts": [
    {"title": "...", "sentence_index": 1}
  ],
  "gold_paragraph_titles": ["..."]
}

The candidate paragraphs preserve upstream count and order. Every supporting title must resolve to one of those paragraphs and every sentence index is checked against the corresponding sentence list.

SFT record

{
  "source_id": "official-hotpotqa-id",
  "task": "cot_reader",
  "messages": [
    {"role": "system", "content": "..."},
    {"role": "user", "content": "..."},
    {"role": "assistant", "content": "...training target..."}
  ]
}

The data stores framework-neutral chat turns. Apply the native chat template of the chosen base model during tokenization; no tokenizer-specific control tokens are embedded in the Parquet files.

CoT target shape

rationale:
1. [Paragraph title, sentence 0] Exact annotated evidence sentence.
2. [Second title, sentence 2] Exact annotated evidence sentence.
answer: ***FINAL ANSWER***

Decomposition target shape

sub-questions:
1. What information in "First paragraph" is needed to answer the original question?
2. How does the relevant information in "Second paragraph" combine with the previous evidence to determine the answer?

How the suite is built

The release builder downloads the immutable upstream revision by default, normalizes the Hugging Face feature layout, constructs every view from the same typed record, writes Parquet shards in a private staging directory, and installs them only after strict validation succeeds.

git clone https://github.com/Iman998/bactrainus.git
cd bactrainus
python -m pip install -e ".[data]"

# From the dataset-package directory:
python scripts/build_release.py --root .
python scripts/validate_release.py --root .

A complete official JSON or JSONL export can also be used:

python scripts/build_release.py hotpot_train_v1.1.json --root .

SOURCE_MANIFEST.json records the upstream repository, immutable revision, configuration, split, row count, identity key, and published configurations. CHECKSUMS.sha256 records every Parquet shard digest.

The pinned upstream split contains 22 supporting-fact entries whose sentence indices are outside their annotated paragraphs. SOURCE_PATCHES.json records each affected source_id, original index, action, and replacement when needed. This reviewed repair covers 0.010% of supporting-fact entries, retains all 90,447 questions, and is applied before any SFT view is constructed.

Integrity guarantees

The validator enforces:

• —exactly 90,447 rows and 90,447 unique IDs in every configuration;
• —identical source_id sets across all eight configurations;
• —the exact upstream two-to-ten paragraph distribution, including 89,609 ten-paragraph records;
• —valid evidence titles and zero-based sentence bounds;
• —exact task identifiers and ordered system / user / assistant messages;
• —the documented 17,972 easy, 56,814 medium, and 15,661 hard examples; and
• —SHA-256 coverage for every published Parquet shard.

No development examples, model predictions, score tables, notebooks, credentials, or checkpoints are bundled with the training suite.

Recommended use

Use structured when designing a new task representation or auditing evidence. Use an SFT configuration when training the corresponding modular component. The IDs make it straightforward to combine configurations without fuzzy joins:

reader = load_dataset(
    "bactrianus/bactrainus-hotpotqa",
    "reader-sft",
    split="train",
    revision="v1.0.0",
)
decomposer = load_dataset(
    "bactrianus/bactrainus-hotpotqa",
    "question-decomposer-sft",
    split="train",
    revision="v1.0.0",
)

assert set(reader["source_id"]) == set(decomposer["source_id"])

This repository is training data, not an independent held-out benchmark. Follow the official HotpotQA protocol for benchmark reporting and disclose preprocessing, prompt, retrieval, and answer-normalization choices.

Limitations

• —The suite inherits factual, coverage, annotation, and social biases from Wikipedia and HotpotQA.
• —The distractor setting supplies a bounded candidate set and does not measure open-corpus retrieval.
• —Supporting facts are dataset annotations and may not exhaust every relevant sentence.
• —Gold-grounded CoT and decomposition targets favor traceability and reproducibility over stylistic diversity.
• —Prompt wording is one clean task formulation and may need adaptation for a different model family.

License and attribution

This dataset is released under CC BY-SA 4.0, consistent with the upstream HotpotQA license. Redistributions and adaptations must preserve attribution and use a compatible ShareAlike license. See `ATTRIBUTION.md` for the complete attribution statement.

Citation

If you use this training suite, cite the Bactrainus article. Retain the upstream HotpotQA attribution described in `ATTRIBUTION.md` when redistributing or adapting the data.

@article{barati2025bactrainus,
  title   = {Bactrainus: Optimizing Large Language Models for Multi-hop Complex Question Answering Tasks},
  author  = {Barati, Iman and Ghafouri, Arash and Minaei-Bidgoli, Behrouz},
  journal = {arXiv preprint arXiv:2501.06286},
  year    = {2025},
  url     = {https://arxiv.org/abs/2501.06286}
}

Links

• —Models: <https://huggingface.co/bactrianus>
• —Clean code: <https://github.com/Iman998/bactrainus>
• —Article archive: <https://arxiv.org/abs/2501.06286>

For the full method, experiments, and interpretation, read the Bactrainus article on arXiv.