datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
xlam-function-calling-60k-raw
XLAM Function Calling 60k Raw Dataset
This dataset includes train and test splits derived from Salesforce/xlam-function-calling-60k.
Train split size: 95% of the original dataset
Test split size: 5% of the original dataset
babi_tasks
bAbI Dataset : Per-Task Splits
The examples in this dataset were obtained from "Muennighoff/babi". This dataset offer the same examples but it can be loaded using a per-task basis:
from datasets import load_dataset
ds = load_dataset('RawthiL/babi_tasks', 'task_1-single_supporting_fact')
The available tasks are:
Task ID
Task Name
Split Name
1
Single Supporting Fact
task_1-single_supporting_fact
2
Two Supporting Facts
task_2-two_supporting_facts
3
Three Supporting… See the full description on the dataset page: https://huggingface.co/datasets/RawthiL/babi_tasks.swedish-legal-decisions-raw-v1
Swedish Court Decisions — Svenska Domstolsavgöranden
55,096 court decisions spanning 45 years of Swedish case law, purpose-built for LLM training.
The most comprehensive open dataset of Swedish appellate court decisions available for AI development. Sourced directly from the official Swedish Courts case law database via their public REST API and preprocessed into three ready-to-use training configurations.
Why This Dataset
Scale and depth: 55,096 decisions covering… See the full description on the dataset page: https://huggingface.co/datasets/nexoneAB/swedish-legal-decisions-raw-v1.xlam-function-calling-60k-raw-augmented
XLAM Function Calling 60k Raw Augmented Dataset
This dataset includes augmented train and test splits derived from product-science/xlam-function-calling-60k-raw.
Train split size: Original size plus augmented data
Test split size: Original size plus augmented data
Augmentation Details
This dataset has been augmented by modifying function names in the original data. Randomly selected function names have underscores replaced with periods at random positions… See the full description on the dataset page: https://huggingface.co/datasets/product-science/xlam-function-calling-60k-raw-augmented.triveni-raw
📦 Pretraining Corpus
📊 Dataset Overview
This dataset combines data from two major sources—Vaani and Flickr30k—to support multilingual and multimodal model pretraining.
Source
Languages
Samples per Language
Total Samples
Vaani
Hindi, English, Hinglish
30,195
90,585
Flickr30k
Hindi, English, Hinglish
31,014
93,042
Total
—
—
183,627
📁 Dataset Sources
🗣️ Vaani Dataset
License: CC-BY-4.0
Description:
VAANI is an… See the full description on the dataset page: https://huggingface.co/datasets/LingoIITGN/triveni-raw.train_raw_video
ShareGPTVideo Raw ActivityNet Videos for Train data
All dataset and models can be found at ShareGPTVideo.
Contents:
Due to our scene split, we provide our processed activityNet videos corresponding to test frames in
train video frames
the processing script is process_activitynet.py
mmlu_pro_categories
MMLU-Pro Dataset : Per-Category Splits
This dataset was created from TIGER-Lab/MMLU-Pro, by dividing the original dataset into a dataset per different category. The objective is to make it easier to work on sub-categories.
from datasets import load_dataset
ds = load_dataset('RawthiL/mmlu_pro_categories', 'category_biology')
The available tasks are:
Category Name
Split Name
Biology
category_biology
Business
category_business
Chemistry
category_chemistry
Computer… See the full description on the dataset page: https://huggingface.co/datasets/RawthiL/mmlu_pro_categories.Stress-17K-raw
🧪 Synthetic Training Data — Stress-17K
Stress-17K is a synthetic dataset designed to support instruction tuning for the StresSLM model tackling Sentence Stress Detection (SSD) and Sentence Stress Reasoning (SSR) tasks, as introduced in the StressTest paper.
It is built by augmenting a raw dataset of stressed speech samples with task-specific prompts, enabling models to learn SSD and SSR more effectively. This README explains how to explore the raw version and how to prepare it for… See the full description on the dataset page: https://huggingface.co/datasets/slprl/Stress-17K-raw.Data-Prep-Bench
Data-Prep-Bench
Dataset Overview
This dataset is a comprehensive resource built for Supervised Fine-Tuning (SFT) and evaluation of Large Language Models (LLMs), covering six domains: Finance, Medicine, Law, Mathematics, Science, and General.
A key feature of this dataset is that we employed 12 different data generation methods (including Agent-based methods, DataFlow series, pure LLM-based generation, and a SKILL method) using multiple cutting-edge models (such as… See the full description on the dataset page: https://huggingface.co/datasets/RAWENTER/Data-Prep-Bench.commonsense_cot_partial_raw
Commonsense QA CoT (Partial, Raw, No Human Annotation)
Dataset Summary
Seeded by the CommonsenseQA dataset (tau/commonsense_qa) this preliminary set randomly samples 1,000 question-answer
entries and uses Mixtral (mistralai/Mixtral-8x7B-Instruct-v0.1) to generate 3 unique CoT (Chain-of-Thought) rationales.
This was created as the preliminary step towards fine-tuning a LM (language model) to specialize on commonsense reasoning.
The working hypothesis, inspired by the… See the full description on the dataset page: https://huggingface.co/datasets/peterkchung/commonsense_cot_partial_raw.test_raw_video_data
ShareGPTVideo Raw Videos for Testing data
All dataset and models can be found at ShareGPTVideo.
Contents:
In case of need, this contains raw videos corresponding to test frames in
Test video frames
Raw-RhinoRhino dataset before doing AI-guided deep cleaning. Contains 1,960,351 examples
aka-llama-korean-dataset-multiturn-raw
Aka-LLAMA Korean Multi-Turn Dataset (Raw)
This dataset is a raw version of a multi-turn Korean conversation dataset generated using kordinal. It is designed for research and development in Korean natural language processing (NLP), specifically in multi-turn dialogue generation.
License
This dataset is released under the CC BY-NC 4.0 license. It is strictly for non-commercial research and educational purposes. Commercial usage is prohibited.
Additionally, some data… See the full description on the dataset page: https://huggingface.co/datasets/snupilab/aka-llama-korean-dataset-multiturn-raw.ruozhiba_raw
Note
预处理方式
from datasets import load_dataset
import jsonlines
import matplotlib.pyplot as plt
ds_ruozhiba = load_dataset("kirp/wisdomBar")
_data = []
for item in ds_ruozhiba["train"]:
instruct = item["title"] if item["detail"] is None else item["title"] + ("," if item["title"][-1] not in [",", "，","。", ".", "！", "!", "？", "?"] else "") + item["detail"]
if instruct:
_data.append(instruct)
_data_to_dump = [[{"from": "human", "value": value}] for value in… See the full description on the dataset page: https://huggingface.co/datasets/ticoAg/ruozhiba_raw.myawady-raw-dataset
Myawady Raw News Corpus 🇲🇲
This dataset contains over 59,000 full-text Burmese news articles scraped from the Myawady News Portal, the official media outlet of the Myanmar military government.
Unlike the title-only version, this dataset includes complete article content, with metadata fields such as category, publication date, and image URLs. It is intended for use in Myanmar NLP and AI research, including:
🧠 Language modeling
📰 Text summarization
🏷️ Named entity… See the full description on the dataset page: https://huggingface.co/datasets/freococo/myawady-raw-dataset.sata-bench-raw
Cite
@misc{xu2025satabenchselectapplybenchmark,
title={SATA-BENCH: Select All That Apply Benchmark for Multiple Choice Questions},
author={Weijie Xu and Shixian Cui and Xi Fang and Chi Xue and Stephanie Eckman and Chandan Reddy},
year={2025},
eprint={2506.00643},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2506.00643},
}
Select-All-That-Apply Benchmark (SATA-bench) Dataset Desciption… See the full description on the dataset page: https://huggingface.co/datasets/sata-bench/sata-bench-raw.10-X-raw-v1
SEC 10-X Filings Dataset
This dataset contains processed SEC 10-X (10-K, 10-Q) filings, focusing on Risk Factors and Management Discussion & Analysis (MD&A) sections from corporate financial reports from 1993-2023.
🔗 Original Dataset: [SEC-EDGAR-10X] contains stripped down versions of the original filings, details about which can be found here.
This dataset is a further cleaned tabulated version of the original stripped down version making it more suitable for training tasks.… See the full description on the dataset page: https://huggingface.co/datasets/theaayushbajaj/10-X-raw-v1.indo-bloom-raw-bse
📚 Indo-Bloom BSE RAW Corpus
⚠️ RESEARCH ARTIFACT STATUS: RAW CORPUS (Stage 0)
This dataset serves as the raw material corpus for the Indo-Bloom research project at Universitas Negeri Malang (UM).
Current State: Extracted & Cleaned Context from BSE Textbooks
Next Stage: QA Pair Generation (Stage 1) → Silver Corpus
🔒 FROZEN — Raw v1.0
This version is permanently frozen to ensure reproducibility. This corpus will be used as input for QA generation pipeline.
📄… See the full description on the dataset page: https://huggingface.co/datasets/Alwiiiiiiiiii/indo-bloom-raw-bse.indo-bloom-raw-bse
📚 Indo-Bloom BSE RAW Corpus
⚠️ RESEARCH ARTIFACT STATUS: RAW CORPUS (Stage 0)
This dataset serves as the raw material corpus for the Indo-Bloom research project at Universitas Negeri Malang (UM).
Current State: Extracted & Cleaned Context from BSE Textbooks
Next Stage: QA Pair Generation (Stage 1) → Silver Corpus
🔒 FROZEN — Raw v1.0
This version is permanently frozen to ensure reproducibility. This corpus will be used as input for QA generation pipeline.
📄 Associated… See the full description on the dataset page: https://huggingface.co/datasets/Firmansyah-Ibrahim/indo-bloom-raw-bse.pl-eureka-ipbox-raw
Interpretacje podatkowe KIS — IP Box (Eureka) — korpus źródłowy
Podzbiór indywidualnych interpretacji podatkowych Krajowej Informacji Skarbowej (KIS)
dotyczących ulgi IP Box — preferencyjnej stawki 5% od dochodu z kwalifikowanych praw
własności intelektualnej (art. 30ca–30cb ustawy o PIT oraz art. 24d–24e ustawy o CIT) —
wyodrębniony z urzędowej bazy Eureka. Jest to nieprzetworzony zbiór bez ręcznych etykiet,
przeznaczony do zadań NLP w obszarze polskiego prawa podatkowego… See the full description on the dataset page: https://huggingface.co/datasets/AI-TAX/pl-eureka-ipbox-raw.commonsense_cot_partial_raw
Commonsense QA CoT (Partial, Raw, No Human Annotation)
Dataset Summary
Seeded by the CommonsenseQA dataset (tau/commonsense_qa) this preliminary set randomly samples 1,000 question-answer
entries and uses Mixtral (mistralai/Mixtral-8x7B-Instruct-v0.1) to generate 3 unique CoT (Chain-of-Thought) rationales.
This was created as the preliminary step towards fine-tuning a LM (language model) to specialize on commonsense reasoning.
The working hypothesis, inspired by the… See the full description on the dataset page: https://huggingface.co/datasets/spacekat99/commonsense_cot_partial_raw.turkish-cuisine-qa_raw
Turkish Cuisine Canonical Dataset (RAW)
Bu veri seti, Uunan/turkish-cuisine-qa reposunda bulunan LLM Instruction Tuning veri setinin ham (canonical JSON) kaynağıdır.
Türk mutfağına ait 2.700'den fazla yöresel ve geleneksel yemeğin; malzemeleri, yapılış aşamaları, yöresi ve coğrafi işaret durumu gibi verileri yapılandırılmış bir JSON objesi içerisinde tutmaktadır.
Araştırmacıların veriyi kendi ihtiyaçlarına göre işlemesi ve farklı formatlara dönüştürebilmesi amacıyla ham kaynak… See the full description on the dataset page: https://huggingface.co/datasets/Uunan/turkish-cuisine-qa_raw.texthumanizer-raw-dataTil-Web-Raw-KK-v1
Til-Web-Raw-KK-v1
Білім беру сайттарының бастапқы веб-айнасы · Исходное зеркало образовательных сайтов · Raw mirror of educational websites
Қазақша · Русский · English
Қазақша
Til-Web-Raw-KK-v1 — қазақ тіліндегі білім беру және сұрақ-жауап сайттарының бастапқы веб-айнасы, көлемі 19.50 ГБ. Репозиторий сервер берген HTML беттерін, суреттерді және жүктелетін тіркемелерді мәтін тазалауға дейінгі күйінде сақтайды.
Құрамы
Сайт
Беттер
HTML көлемі… See the full description on the dataset page: https://huggingface.co/datasets/TilQazyna/Til-Web-Raw-KK-v1.pl-eureka-raw
Interpretacje podatkowe KIS (Eureka) — korpus źródłowy
Bazowy korpus indywidualnych interpretacji podatkowych wydawanych przez Krajową
Informację Skarbową (KIS), pozyskany z urzędowej bazy Eureka. Jest to nieprzetworzony zbiór bez ręcznych etykiet, przeznaczony do zadań NLP
w obszarze polskiego prawa podatkowego, takich jak wyszukiwanie
semantyczne, klasyfikacja, ekstrakcja informacji, odpowiadanie na pytania i streszczanie.
Liczba dokumentów: 488 917 interpretacji
Język:… See the full description on the dataset page: https://huggingface.co/datasets/AI-TAX/pl-eureka-raw.