datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
wikidata-extraction
Wikidata Extraction
This dataset contains all RDF triples extracted from the latest Wikidata, converted from the N-Triples format to Parquet.
The data originates from Wikidata, a free and open knowledge base that acts as central storage for structured data used by Wikipedia and other Wikimedia projects. The source file is the "truthy" N-Triples dump (latest-truthy.nt.bz2), which contains only the current, non-deprecated statements.
The code to extract this data is available at… See the full description on the dataset page: https://huggingface.co/datasets/piebro/wikidata-extraction.vi-books_extractionll books_text_final/ | wc -l
=> 13968 text
After filter (not vi, too small)
wc -l vi-books.jsonl
=> 12832 text
=> !!! Cần lọc truyện từ dtv ra !!!
xzcat dtv_ebooks_details.jsonl.xz | wc -l
# => 13486
ll books_text_final/ | grep dtv_ebooks_ | wc -l
# => 3535 từ dtv
lấy các whitelist categories tại dtv_categories_whitelist.jsonl
lần theo từng ebook một trong dtv_ebooks_details.jsonl.xz trường {"cat": "Self Help - Khởi nghiệp", "cat_url":… See the full description on the dataset page: https://huggingface.co/datasets/tiendung/vi-books_extraction.tweet_sentiment_extraction
TweetSentimentExtractionClassification
An MTEB dataset
Massive Text Embedding Benchmark
Task category
t2c
Domains
Social, Written
Reference
https://www.kaggle.com/competitions/tweet-sentiment-extraction/overview
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_tasks(["TweetSentimentExtractionClassification"])
evaluator = mteb.MTEB(task)
model =… See the full description on the dataset page: https://huggingface.co/datasets/mteb/tweet_sentiment_extraction.sec-contracts-financial-extraction-instructions
S&P 500 SEC Financial Extraction Instructions
Dataset Summary
7,683 instruction-tuning examples for training LLMs to extract structured financial data from SEC filings. Covers two filing types across S&P 500 companies:
Split
Examples
Filing Type
Description
train
3,430
Exhibit 10 + DEF 14A
Positive examples with validated outputs
corrective
4,253
Exhibit 10 + DEF 14A
Corrective, rescued, and negative examples
Exhibit 10 — Material Contracts (2… See the full description on the dataset page: https://huggingface.co/datasets/TheTokenFactory/sec-contracts-financial-extraction-instructions.tool-output-extraction-swebench
Tool Output Extraction Dataset
Paper | Code
Training data for squeez — a small model that prunes verbose coding agent tool output to only the evidence the agent needs next.
Task
Task-conditioned context pruning of a single tool observation for coding agents.
Given a focused extraction query and one verbose tool output, return the smallest verbatim evidence block(s) the agent should read next.
The model copies lines from the tool output — it never rewrites, summarizes, or… See the full description on the dataset page: https://huggingface.co/datasets/KRLabsOrg/tool-output-extraction-swebench.affirming-review-extractionscrapinghub-article-extraction-benchmark
Scrapinghub Article Extraction Benchmark
This dataset was originally created and distributed under MIT License by Scrapinghub on GitHub: github.com/scrapinghub/article-extraction-benchmark
It is mirrored on the HuggingFace Hub as a convenience.
DRIVE-digital-retinal-images-for-vessel-extractionarXiv:2501.18921https://arxiv.org/abs/2501.18921
paper_extractionprice-tag-extraction
Price tag extraction dataset
This dataset contains images of price tags extracted from the Open Prices dataset, along with information extracted from this dataset.
It is intended to be used to train large visual language models (LVLMs) to extract information from price tag images, as part of the Open Prices project.
For more information about the format of this dataset, please refer to the documentation on llm-image-extraction datasets.
Dataset creation
A detailed… See the full description on the dataset page: https://huggingface.co/datasets/openfoodfacts/price-tag-extraction.supercon-extraction-harbor-tasksglobal-openstreetmap-extraction-slippy-tiles-tarraw-fact-extractiontweet_sentiment_extraction
Tweet Sentiment Extraction
Source: https://www.kaggle.com/c/tweet-sentiment-extraction/data
cdw-extraction-harbor-tasksskill-extraction-tech
Skill Extraction with ESCO skills - TECH subset
Dataset Summary
This dataset contains an extension of the TECH subset form the SkillSpan dataset, in which spans of skill mentions in sentences have been labeled with corresponding ESCO skills (ESCO v1.1.0).
This dataset is part of a three-part evaluation dataset for skill extraction:
skill-extraction-tech
skill-extraction-house
skill-extraction-techwolf
Citation Information
If you use this dataset, please include… See the full description on the dataset page: https://huggingface.co/datasets/TechWolf/skill-extraction-tech.funding-extraction-harness-benchmarkbiosurfactants-extraction-harbor-tasksmall_receipt_extraction_datasetinsurance-claims-extractionThis dataset can be used for benchmarking LLM Structured Outputs via the code here:
https://github.com/cleanlab/structured-output-benchmark/
web-content-extraction-benchmark
WCXB: Web Content Extraction Benchmark
The largest open benchmark for evaluating web content extraction, boilerplate removal, and main content detection across diverse page types.
WCXB provides 2,008 human-reviewed web pages spanning 7 page types and 1,613 domains, with ground truth annotations, HTML source files, and baseline results from 14 extraction systems. Unlike existing benchmarks that focus exclusively on news articles, WCXB evaluates extraction across the full diversity of… See the full description on the dataset page: https://huggingface.co/datasets/murrough-foley/web-content-extraction-benchmark.json_data_extraction
Diverse Restricted JSON Data Extraction
Curated by: The paraloq analytics team.
Uses
Benchmark restricted JSON data extraction (text + JSON schema -> JSON instance)
Fine-Tune data extraction model (text + JSON schema -> JSON instance)
Fine-Tune JSON schema Retrieval model (text -> retriever -> most adequate JSON schema)
Out-of-Scope Use
Intended for research purposes only.
Dataset Structure
The data comes with the following fields:
title: The… See the full description on the dataset page: https://huggingface.co/datasets/paraloq/json_data_extraction.danish-dynaword-extractionsSkill-extraction-Tech-graded
skill-extraction-tech-graded
Graded-relevance annotations for sentences from
TechWolf/skill-extraction-tech
against the ESCO v1.1.0 skill taxonomy. Layout follows the
BEIR convention so it is drop-in for
MTEB-style retrieval evaluators.
This dataset was created for the RecSys-HR 2026 WorkRB challenge.
Configs
config
split
rows
columns
queries
validation
75
_id (sentence id), text (sentence)
queries
test
338
_id (sentence id), text (sentence)
corpus… See the full description on the dataset page: https://huggingface.co/datasets/TechWolf/Skill-extraction-Tech-graded.doab-metadata-extraction
DOAB Open Access Books - Metadata Extraction Dataset
Dataset Description
This dataset contains 9,363 open access books with page images and rich bibliographic metadata extracted from MARC21 records, curated specifically for training and evaluating Vision Language Models (VLMs) on automatic metadata extraction from scholarly monographs.
The dataset is derived from the Penn State ScholarSphere DOAB collection (Directory of Open Access Books), focusing on books with Creative… See the full description on the dataset page: https://huggingface.co/datasets/biglam/doab-metadata-extraction.task181_outcome_extraction
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task181_outcome_extraction
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/task181_outcome_extraction.paper_extraction_v1tooth_extraction_4This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.0",
"robot_type": "so100",
"total_episodes": 200,
"total_frames": 76053,
"total_tasks": 1,
"total_videos": 400,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:200"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/samsam0510/tooth_extraction_4.danish-extraction-v1
danish-extraction-v1
Danish information-extraction rows over real prose, where the schema is
proposed per passage rather than fixed. Built from
danish-foundation-models/danish-dynaword
by scripts/gen_extraction_da.py.
Each source passage got its own field set: an LLM proposed 3-6 fields for that
text without seeing any values, then filled them in a separate turn. Roughly a
quarter of proposed fields come back empty, which are genuine abstention
targets rather than annotation… See the full description on the dataset page: https://huggingface.co/datasets/jensjepsen/danish-extraction-v1.Skill-extraction-SkillSkape-graded
skill-extraction-skillskape-graded
Graded-relevance annotations for sentences from
jjzha/skillskape
against the ESCO v1.1.0 skill taxonomy. Layout follows the
BEIR convention so it is drop-in for
MTEB-style retrieval evaluators.
This dataset was created for the RecSys-HR 2026 WorkRB challenge.
Configs
config
split
rows
columns
queries
validation
100
_id (sentence id), text (sentence)
queries
test
500
_id (sentence id), text (sentence)
corpus
corpus… See the full description on the dataset page: https://huggingface.co/datasets/TechWolf/Skill-extraction-SkillSkape-graded.