datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
hotpot_qa
Dataset Card for "hotpot_qa"
Dataset Summary
HotpotQA is a new dataset with 113k Wikipedia-based question-answer pairs with four key features: (1) the questions require finding and reasoning over multiple supporting documents to answer; (2) the questions are diverse and not constrained to any pre-existing knowledge bases or knowledge schemas; (3) we provide sentence-level supporting facts required for reasoning, allowingQA systems to reason… See the full description on the dataset page: https://huggingface.co/datasets/hotpotqa/hotpot_qa.HotpotQA_test_top_250_only_w_correct-v2
HotpotQAHardNegatives
An MTEB dataset
Massive Text Embedding Benchmark
HotpotQA is a question answering dataset featuring natural, multi-hop questions, with strong supervision for supporting facts to enable more explainable question answering systems. The hard negative version has been created by pooling the 250 top documents per query from BM25, e5-multilingual-large and e5-mistral-instruct.
Task category
t2t
Domains
Web, Written
Reference
https://hotpotqa.github.io/… See the full description on the dataset page: https://huggingface.co/datasets/mteb/HotpotQA_test_top_250_only_w_correct-v2.nq_hotpotqa_trainhotpotqa
HotpotQA
An MTEB dataset
Massive Text Embedding Benchmark
HotpotQA is a question answering dataset featuring natural, multi-hop questions, with strong supervision for supporting facts to enable more explainable question answering systems.
Task category
t2t
Domains
Web, Written
Reference
https://hotpotqa.github.io/
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task =… See the full description on the dataset page: https://huggingface.co/datasets/mteb/hotpotqa.hotpotqa
Dataset Card for BEIR Benchmark
hotpotqa is one of the datasets from the Question Answering task within BEIR, measuring Wikipedia article retrieval for a given multi-hop query.
Dataset Summary
BEIR is a heterogeneous benchmark built from 18 diverse datasets representing 9 information retrieval tasks.
Fact-checking: FEVER, Climate-FEVER, SciFact
Question-Answering: NQ, HotpotQA, FiQA-2018
Bio-Medical IR: TREC-COVID, BioASQ, NFCorpus
News Retrieval: TREC-NEWS, Robust04… See the full description on the dataset page: https://huggingface.co/datasets/BeIR/hotpotqa.bactrainus-hotpotqa
Bactrainus HotpotQA Training Suite
COMPLETE RELEASE
One canonical HotpotQA source, eight clean and ID-aligned training views.
90,447 SOURCE IDs
8 CONFIGURATIONS
SFT READY
ID-ALIGNED
CC BY-SA 4.0
The Bactrainus training suite turns the complete 90,447-example HotpotQA distractor/train split into a canonical structured dataset and seven ready-to-train chat-formatted SFT configurations. Every row keeps the official HotpotQA… See the full description on the dataset page: https://huggingface.co/datasets/bactrianus/bactrainus-hotpotqa.hotpot_qaThis dataset is a modified version of the HotPotQA distractor dataset, which contains factual questions requiring multi-hop reasoning.
In the original HotPotQA dataset, each example presents ten paragraphs, only two of which contain the information necessary to answer the question; the remaining eight paragraphs include closely related but irrelevant details.
Consequently, solving this task requires the model to identify and reason over the pertinent passages.
To more strongly develop… See the full description on the dataset page: https://huggingface.co/datasets/mehuldamani/hotpot_qa.hotpotqa_verified_logits_with_base_50khotpotqa-qrels
Dataset Card for BEIR Benchmark
Dataset Summary
BEIR is a heterogeneous benchmark that has been built from 18 diverse datasets representing 9 information retrieval tasks:
Fact-checking: FEVER, Climate-FEVER, SciFact
Question-Answering: NQ, HotpotQA, FiQA-2018
Bio-Medical IR: TREC-COVID, BioASQ, NFCorpus
News Retrieval: TREC-NEWS, Robust04
Argument Retrieval: Touche-2020, ArguAna
Duplicate Question Retrieval: Quora, CqaDupstack
Citation-Prediction: SCIDOCS
Tweet… See the full description on the dataset page: https://huggingface.co/datasets/BeIR/hotpotqa-qrels.hotpotqa_verified_logits_with_128_distractors_preprocessedhotpotqa-decontaminated
hotpotqa (Decontaminated)
A decontaminated version of the hotpotqa dataset from the BEIR benchmark, with samples found in the mgte-en pre-training dataset removed.
Decontamination methodology
Contamination was detected using a two-pass approach against the full mgte-en dataset (484 GB, 1,235 parquet files):
Pass 1: Exact hash matching
All texts (queries and corpus documents) were normalized (lowercased, unicode NFKD, whitespace collapsed) and hashed with… See the full description on the dataset page: https://huggingface.co/datasets/lightonai/hotpotqa-decontaminated.hotpotqa
Dataset Card for HotpotQA
This is a reformatting of the HotpotQA dataset used to train the BGE-M3 model. See the full BGE-M3 dataset in Shitao/bge-m3-data.
Dataset Subsets
triplet subset
Columns: "anchor", "positive", "negative"
Column types: str, str, str
Examples:{
'anchor': 'What country of origin does House of Cosbys and Bill Cosby have in common?',
'positive': 'House of Cosbys House of Cosbys is an American animated sitcom created by Justin Roiland… See the full description on the dataset page: https://huggingface.co/datasets/sentence-transformers/hotpotqa.hotpotqa-distractor-lance
HotpotQA distractor (Lance Format)
A Lance-formatted version of HotpotQA using the distractor config — multi-hop reading-comprehension questions where each answer requires combining facts from two Wikipedia paragraphs, with 10 candidate paragraphs per question (gold + 8 distractors). The dataset ships with MiniLM question embeddings, flattened context text for full-text search, and pre-built ANN/FTS indices, available directly from the Hub at… See the full description on the dataset page: https://huggingface.co/datasets/lance-format/hotpotqa-distractor-lance.hotpot_qa
HotpotQA with Full Wikipedia Articles
This dataset extends the original HotpotQA dataset by including complete Wikipedia article text for all referenced articles in each example.
Dataset Structure
This dataset contains two configurations matching the original HotpotQA:
distractor: 97,940 examples with 10 paragraphs each (2 gold + 8 distractor)
fullwiki: 105,257 examples requiring retrieval from full Wikipedia
New Feature: full_articles
Each example now… See the full description on the dataset page: https://huggingface.co/datasets/ParthMandaliya/hotpot_qa.msa-hotpotqa-qa-with-idsbactrainus-hotpotqa-teacher-traces
Bactrainus HotpotQA Teacher Traces
SOURCE-LINKED v1.0.0
Archived Llama 3.1 rationale and question-decomposition supervision, paired with complete SFT conversations and stable HotpotQA identities.
198,660 ROWS
4 CONFIGURATIONS
SFT MESSAGES
8B + 70B LABELS
CC BY-SA 4.0
A focused release of recovered teacher-generated supervision for multi-hop question answering. Every row contains the normalized annotation, an ordered… See the full description on the dataset page: https://huggingface.co/datasets/bactrianus/bactrainus-hotpotqa-teacher-traces.hotpotqa-fr-abstracts
HotpotQA-fr (abstracts)
65 565 questions multi-sauts en français, au format de
HotpotQA (Yang et al., 2018). Les questions sont
construites directement sur Wikipédia français, sans traduction. L'annotation
humaine du jeu original est remplacée par une génération par modèle de langue
suivie d'une validation automatique par ablations.
Chaque contexte contient dix paragraphes : les deux abstracts nécessaires à la
réponse et huit distracteurs. Tous les paragraphes sont des abstracts… See the full description on the dataset page: https://huggingface.co/datasets/Mvanypersele/hotpotqa-fr-abstracts.triviaqa-hotpotqa-nq-squad-msmarco-hard-neg-sft4bmsa-hotpotqa-docs-with-idsSplits_Subset_HotpotQa
Dataset Card for "Splits_Subset_HotpotQa"
More Information needed
hotpot_qa_modifiedThis dataset is a modified version of the HotPotQA distractor dataset, which contains factual questions requiring multi-hop reasoning.
In the original HotPotQA dataset, each example presents ten paragraphs, only two of which contain the information necessary to answer the question; the remaining eight paragraphs include closely related but irrelevant details.
Consequently, solving this task requires the model to identify and reason over the pertinent passages.
To more strongly develop… See the full description on the dataset page: https://huggingface.co/datasets/markstanl/hotpot_qa_modified.hotpotqaProxyCoT-HotpotQAThis is the HotpotQA data that we used in our ProxyCoT project (https://aclanthology.org/2026.acl-long.1917/), and it is based on long-context reasoning (32K-128K tokens).
HotpotQA here is a new version originally from https://aclanthology.org/2026.acl-long.1917/ with extended contexts.
For more details on the context extension, refer to the ProxyCoT paper.
To use our dataset, please follow the code below.
train_samples = load_dataset("oaimli/proxycot-hotpotqa", split="train")
dev_samples =… See the full description on the dataset page: https://huggingface.co/datasets/oaimli/ProxyCoT-HotpotQA.task192_hotpotqa_sentence_generation
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task192_hotpotqa_sentence_generation
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks}… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task192_hotpotqa_sentence_generation.hotpotqahotpotqa
Dataset Card for "hotpotqa"
Split taken from the MRQA 2019 Shared Task, formatted and filtered for Question Answering. For the original dataset, have a look here.
hotpotqa-wiki
HotpotQA Wikipedia Corpus
This dataset is a processed Wikipedia corpus derived from the HotpotQA Wikipedia dump.
The original dump was downloaded from:
https://nlp.stanford.edu/projects/hotpotqa/enwiki-20171001-pages-meta-current-withlinks-processed.tar.bz2
Additional details about the structure and preprocessing of the original HotpotQA Wikipedia dump are available here:
https://hotpotqa.github.io/wiki-readme.html
Dataset Description
This dataset is built on top… See the full description on the dataset page: https://huggingface.co/datasets/ParthMandaliya/hotpotqa-wiki.rollouts-hotpotqahotpotqa-vn
HotpotQA-VN
An MTEB dataset
Massive Text Embedding Benchmark
A translated dataset from HotpotQA is a question answering dataset featuring natural, multi-hop questions, with strong supervision for supporting facts to enable more explainable question answering systems. The process of creating the VN-MTEB (Vietnamese Massive Text Embedding Benchmark) from English samples involves a new automated system: - The system uses large language models (LLMs), specifically Coherence's Aya model… See the full description on the dataset page: https://huggingface.co/datasets/GreenNode/hotpotqa-vn.beir-hotpotqa
HotpotQA — BEIR, unified schema
A normalised copy of the dataset behind the mteb task HotpotQA, one of the tasks of the BEIR benchmark as mteb defines it. Same queries, documents
and relevance judgements as the benchmark evaluates — reshaped into one strict schema shared by every dataset
in this collection.
Source
mteb/hotpotqa @ ab518f4d6fcc (the revision pinned in mteb)
Domain · languages
wikipedia · eng
Queries / documents / qrels (all splits)
97,852 / 5,233… See the full description on the dataset page: https://huggingface.co/datasets/Hyukkyu/beir-hotpotqa.