datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Fineweb-Edu-Chinese-V2.2
Chinese Fineweb Edu Dataset V2.2 (Instruct & Pre-train)
[[中文]] | [[English]]
OpenCSG Community | 👾 GitHub | 📖 Technical Report
Dataset Introduction: Filling the Data Puzzle for Chinese Education LLMs
Chinese Fineweb Edu Dataset V2.2is a rare high-quality dataset in the open-source community that covers the full process from Pre-training to Supervised Fine-Tuning (SFT) for the Chinese education domain.
This project aims to solve the core pain point of… See the full description on the dataset page: https://huggingface.co/datasets/opencsg/Fineweb-Edu-Chinese-V2.2.fineweb-edu
FineWeb-Edu (Lance Format)
A Lance-formatted version of FineWeb-Edu — over 1.5 billion educational web passages with cleaned text, source metadata, language detection signals, and 384-dim text embeddings — available directly from the Hub at hf://datasets/lance-format/fineweb-edu/data/train.lance.
Key features
Cleaned passage text in the text column with the source url and title carried alongside.
Language detection signals (language, language_probability) for filtered… See the full description on the dataset page: https://huggingface.co/datasets/lance-format/fineweb-edu.Fineweb-Edu-Chinese-V2.2
Chinese Fineweb Edu Dataset V2.2 (Instruct & Pre-train)
[[中文]] | [[English]]
OpenCSG Community | 👾 GitHub | 📖 Technical Report
Dataset Introduction: Filling the Data Puzzle for Chinese Education LLMs
Chinese Fineweb Edu Dataset V2.2is a rare high-quality dataset in the open-source community that covers the full process from Pre-training to Supervised Fine-Tuning (SFT) for the Chinese education domain.
This project aims to solve the core pain point of… See the full description on the dataset page: https://huggingface.co/datasets/enche1561/Fineweb-Edu-Chinese-V2.2.DeepSWEGym2-Edu
Dataset Description
This dataset is a filtered and deduplicated version of a merge containing many high quality SWE datasets, it aims to improve benchmark results on DeepSWE-style problems, benchmarks, and general coding skills.
It is specifically filtered for rows with complex/long code problems in the original datasets, having an average row size of 291.74kb, a total uncompressed size of 14.93GB, and a total of 53649 examples.
Dataset Details
Curated by:… See the full description on the dataset page: https://huggingface.co/datasets/MoreThought/DeepSWEGym2-Edu.Omni-Edu
Omni-Edu — Core V6 SFT mixture
69,999 supervised instruction examples (~158M characters) covering K-12 subject
competence, curriculum grounding, diagnostic reasoning, pedagogical action and
general-purpose instruction. 12,146 rows (17.4%) are multimodal; every image
referenced by the JSONL ships in this repository under images/.
This is the system-prompted assembly of the v6 core mixture: every row carries
an explicit system message, and the non-system turns are byte-identical… See the full description on the dataset page: https://huggingface.co/datasets/lhpku20010120/Omni-Edu.DeepSWEGym-Edu
Dataset Description
This dataset is a heavily filtered version of all the SWE-bench/SWE-smith-lang datasets (expect php) merged together (originally 88k rows total), it aims to improve benchmark results on DeepSWE-style problems, benchmarks, and general coding skills.
It is specifically filtered for rows with very complex/long code problems in the original datasets, having an average row size of 101.81kb, a total uncompressed size of 4.29GB, and 42529 examples total.… See the full description on the dataset page: https://huggingface.co/datasets/MoreThought/DeepSWEGym-Edu.Fineweb-Edu-Chinese-V2.3
Chinese Fineweb Edu Dataset V2.3
中文 | English
OpenCSG 社区 | GitHub | 数据集许可协议
数据集简介
Chinese Fineweb Edu Dataset V2.3 是 OpenCSG 面向中文教育、知识问答、指令微调和文本生成场景构建的高质量中文教育 SFT 数据集。
该版本包含 23.04 万条高质量中文教育 QA pairs，并将同一批问答对发布为 Alpaca、Messages、Messages-no-system 三种训练格式。三种格式面向不同训练模板，建议训练时按模型和框架选择其中一种格式使用，而不是将不同格式简单相加作为独立知识规模。
V2.3 是在 V2.2 基础上的质量升级版本。针对 V2.2 社区反馈和内部质量审计中出现的重复模式、异常中英文混入、噪声片段、弱证据支撑回答和低质量合成输出等问题，V2.3 提高了源文本进入生成环节的门槛，并优化了问答生成与过滤逻辑。
在数据构建上，V2.3 从约 2.3T… See the full description on the dataset page: https://huggingface.co/datasets/opencsg/Fineweb-Edu-Chinese-V2.3.cc_news_pt_v2
Dataset Summary
This version of the dataset is the portuguese subset from stanford-oval/ccnews.
CC-News-PT v2 is a curation of +11 million news articles from CommonCrawl News in the Portuguese language, from the beginning (2016) to June of 2024.
The data has been cleaned and deduplicated, and language of articles have been detected and filtered. The process is similar to what HuggingFace's DataTrove does.
For license information, please refer to CommonCrawl's Terms of Use.… See the full description on the dataset page: https://huggingface.co/datasets/eduagarcia/cc_news_pt_v2.IndustryInstruction_Subject-Education
IndustryInstruction: Subject Education
This repository contains the IndustryInstruction: Subject Education domain subset of BAAI/IndustryInstruction.
Refer to the parent dataset card for data construction, intended use, limitations,
and licensing details.
Citation
If you use this dataset in your work, please cite IndustryInstruction:
@misc{shi2024industryinstruction,
title = {IndustryInstruction},
author = {Xiaofeng Shi and Lulu Zhao and Hua Zhou and… See the full description on the dataset page: https://huggingface.co/datasets/BAAI/IndustryInstruction_Subject-Education.Fineweb-Edu-Chinese-V3
Fineweb-Edu-Chinese-V3
中文 | English
OpenCSG 社区 | 数据集许可协议
数据集简介
Fineweb-Edu-Chinese-V3 是 OpenCSG 面向学科知识问答、教材理解和推理型指令微调场景构建的高质量中英双语教育 SFT 数据集，也是 Fineweb-Edu-Chinese 系列的最新版本。
该版本包含 18.81 万条 SFT 样本，来自 100,442 篇高质量图书、教材、学科文献与技术长文，覆盖计算机、自然科学、社科人文、法学、经济五大学科方向，并同步提供 Messages、Messages-no-system、Alpaca 三种训练格式。三种格式是同一批问答对的不同导出视图，训练时应按模型模板选择其中一种，而不是简单相加作为独立数据规模。
V3 是 Fineweb-Edu-Chinese 系列的一次数据源与构造范式的整体切换。V1.0 至 V2.3 均以大规模中文网页语料为基础：通过打分器筛选出具备教育属性的网页文本，再由大模型生成问答。V3… See the full description on the dataset page: https://huggingface.co/datasets/opencsg/Fineweb-Edu-Chinese-V3.eduadapt
EduAdapt: Grade-Level Adaptive Science Education Dataset
Dataset Description
Dataset Summary
EduAdapt is a comprehensive educational dataset designed for adaptive learning systems and K-12 science education. The dataset contains question-answer pairs and multiple-choice questions organized by grade level (Grades 1-12) across various scientific disciplines including biology, chemistry, physics, computer science, ecology, medicine, geography, geology, and… See the full description on the dataset page: https://huggingface.co/datasets/notefill/eduadapt.EducationQ
EducationQ Dataset: A Balanced Teaching-Oriented Testbed
🌟 Definition and Value
The EducationQ Dataset is a high-quality, balanced, and teaching-oriented testbed designed to evaluate the pedagogical capabilities of Large Language Models (LLMs). As detailed in our ACL 2025 paper, this dataset serves as the foundational benchmark for the EducationQ multi-agent dialogue framework.
Why "High-Quality and Balanced"?
A "teaching-oriented" evaluation requires… See the full description on the dataset page: https://huggingface.co/datasets/SunriserFuture/EducationQ.EduBench
EduBench 📚
EduBench é um benchmark em português brasileiro para avaliação de Large Language Models (LLMs) em tarefas educacionais, composto por 3,149 questões discursivas extraídas de vestibulares de alta competitividade.
GitHub
Paper
Dataset Description
Fontes
USP: Universidade de São Paulo
UNICAMP: Universidade Estadual de Campinas
UNESP: Universidade Estadual Paulista
Período
2015-2025 (11 anos de provas)
Áreas do… See the full description on the dataset page: https://huggingface.co/datasets/recogna-nlp/EduBench.Fineweb-Edu-Chinese-V2.2
Chinese Fineweb Edu Dataset V2.2 (Instruct & Pre-train)
[[中文]] | [[English]]
OpenCSG Community | 👾 GitHub | 📖 Technical Report
Dataset Introduction: Filling the Data Puzzle for Chinese Education LLMs
Chinese Fineweb Edu Dataset V2.2is a rare high-quality dataset in the open-source community that covers the full process from Pre-training to Supervised Fine-Tuning (SFT) for the Chinese education domain.
This project aims to solve the core pain point of… See the full description on the dataset page: https://huggingface.co/datasets/lvmx12344545366/Fineweb-Edu-Chinese-V2.2.InstructGpt-educational
LuminaSFT
LuminaSFT is a synthetic SFT dataset suite specifically designed to improve both general-purpose and task-specific SLMs. LuminaSFT consists of multiple curated splits that target diverse capabilities:
UltraChat200K-DeepSeek - A regenerated base SFT dataset for broad instruction following.
InstructGPT-NaturalQA and InstructGPT-TriviaQA - Factual question answering datasets to strengthen knowledge recall and answer accuracy.
CoT-Drop - A reading comprehension dataset with… See the full description on the dataset page: https://huggingface.co/datasets/amd/InstructGpt-educational.LLM_Electrical_Engineering_Educational_Synthetic_DialogDataset Card for LLM_Electrical_Engineering_Educational_Synthetic_Dialog
The full CJ Jones' synthetic dataset catalog is available at: https://datadeveloper1.gumroad.com
Want more? 🚀 Get the AI Startup Bundle from Gumroad.
Dataset Description
The LLM_Electrical_Engineering_Educational_Synthetic_Dialog dataset contains AI-generated conversational interactions designed for training large language models in electrical engineering education. This synthetic dialogue corpus simulates tutor-student… See the full description on the dataset page: https://huggingface.co/datasets/CJJones/LLM_Electrical_Engineering_Educational_Synthetic_Dialog.eduagentbench
EduAgentBench
EduAgentBench is a 150-task benchmark release for evaluating AI tutor agents. It targets three complementary capabilities: professional pedagogical judgment, situated multi-turn tutoring, and Canvas-style teaching workflow execution.
This release is prepared for anonymous research review. It contains transformed benchmark task definitions, synthetic/mock course state, compact metadata, and machine-readable dataset documentation. It does not redistribute raw source PDFs… See the full description on the dataset page: https://huggingface.co/datasets/eduagentbench/eduagentbench.EduMUSE
OpenStax Multimodal Exercise Dataset
A multimodal dataset of textbook exercises scraped from OpenStax, each aligned to its most relevant textbook subsection and scored under several open vision-language models. The dataset enables research on retrieval-augmented question answering, the contribution of visual context to scientific QA, and ablation studies on text-only vs. multimodal context.
Contents
final_EduMUSE_dataset.json — the unified dataset (nested by book →… See the full description on the dataset page: https://huggingface.co/datasets/upb-nlp/EduMUSE.RusLang-edu-1000
RusLang-Edu-1000 — an educational Russian-language QA dataset
RusLang-Edu-1000 is an expert-curated dataset of 1,000 instruction-format records ("question — detailed educational answer") covering the Russian language and linguistics: from phonetics and orthography to dialectology and theoretical linguistics. Every record contains a detailed answer (on average ≈1,100 characters), a short reference answer, a concise statement of the rule, and rich annotation (subject area, task… See the full description on the dataset page: https://huggingface.co/datasets/DatasetsEval/RusLang-edu-1000.EduViQA
Dataset Card for EduVQA-Alpha
Dataset Summary
EduVQA-Alpha is a multilingual educational dataset designed for video question-answering (VideoQA). It consists of academic videos, annotated with synthetic question-answer (QA) pairs, in English and Persian. Videos are curated to reflect diverse academic topics and teaching styles, supporting multilingual Retrieval-Augmented Generation (RAG) tasks.
The dataset employs CLIP-SSIM Adaptive Chunking for video segmentation… See the full description on the dataset page: https://huggingface.co/datasets/UIAIC/EduViQA.EduFeedback
EduFeedback
Alternating dataset example: a single curated multi-turn conversation yields a complete (prompt, chosen, rejected) triplet on its own — the direct early response becomes chosen and a later, less-direct response becomes rejected. Both sides come from the same real dialog, so no synthetic LLM generation is needed to fill in the rejected side.
EduFeedback is a synthetically generated, multi-turn conversational
preference dataset in an educational tutoring setting… See the full description on the dataset page: https://huggingface.co/datasets/miria0/EduFeedback.eduintel-faiss
EduIntel FAISS Index v1
A FAISS vector index of Sepedi (Sesotho sa Leboa) text chunks for the
EduIntel RAG pipeline, built by Sediba AI.
This is not a model — it is a retrieval index. It stores chunked Sepedi text and their
embeddings so that a retriever can find relevant Sepedi content for a query, which is then
fed to a generative model (e.g. Sedibaai/SedibaLM)
for answer generation.
Contents
File
Purpose
eduintel.index
The FAISS index (flat or IVF… See the full description on the dataset page: https://huggingface.co/datasets/Sediba-AI/eduintel-faiss.EduData
Dataset Card for EduData
Dataset Summary
EduData is a large-scale Chinese educational question-answering dataset released with our AAAI 2026 paper, "From Diagnosis to Generalization: A Cognitive Approach to Data Selection for Educational LLMs".
The dataset is designed to support the training and evaluation of educational large language models, with a particular focus on data selection, cross-subject transfer, and generalization in exam-style reasoning settings.… See the full description on the dataset page: https://huggingface.co/datasets/guoyx18/EduData.cs50-educational-rag
CS50 Pedagogical RAG Dataset
📜 Dataset Description
This repository contains the data artifacts for the undergraduate thesis, which explores the use of a pedagogical chatbot with Retrieval-Augmented Generation (RAG) for Harvard's CS50: Introduction to Computer Science course.
The project involved several stages of data processing, from raw content collection to the generation and curation of a high-quality evaluation dataset. To ensure full transparency and… See the full description on the dataset page: https://huggingface.co/datasets/dev-jonathanb/cs50-educational-rag.adaption-urdu-edu-cultural-reasoning
This dataset is a remastered version prepared using Adaption's Adaptive Data platform.
adaption-urdu_edu_cultural_reasoning
This dataset contains a mixed collection of question-answer pairs and linguistic tasks presented in both English and Urdu. The content spans multiple domains including history, biology, geography, and Urdu literature, featuring multiple-choice questions, translation exercises, and poetic composition prompts. Samples include historical treaty analysis… See the full description on the dataset page: https://huggingface.co/datasets/abdullah693/adaption-urdu-edu-cultural-reasoning.edubloom-question-bank
🎓 EduBloom: Academic Question Bank Dataset
📌 Overview
EduBloom Academic Question Bank Dataset is the official academic question dataset developed for EduBloom: Bloom's Taxonomy-Based Academic Intelligence Platform under NEP-2020.
The dataset is designed to support intelligent academic assessment systems, automated question-paper generation, semantic question retrieval, difficulty prediction, and cognitive-level classification.
It organizes academic questions… See the full description on the dataset page: https://huggingface.co/datasets/Uzaib52/edubloom-question-bank.k12-special-education-accommodations
K-12 Special Education Accommodations
A small reference dataset of special education accommodations and the federal IDEA
disability taxonomy. This is a reference table, not a corpus - 50 accommodation
records plus two small lookup tables.
Loading
from datasets import load_dataset
ds = load_dataset("robworks-software/k12-special-education-accommodations")
Contents
Table
Rows
Contents
train / validation / test
40 / 5 / 5
accommodation… See the full description on the dataset page: https://huggingface.co/datasets/robworks-software/k12-special-education-accommodations.crypto-education-en-corpus
Crypto Education Corpus (EN)
A curated educational corpus about cryptocurrency and blockchain technology, designed for building and evaluating RAG (Retrieval-Augmented Generation) systems.
Source Distribution
Source
Documents
Share
iqwiki.com
2,379
68.2%
academy.binance.com
581
16.7%
kraken.com
183
5.2%
coinbase.com
125
3.6%
gemini.com
94
2.7%
ethereum.org
74
2.1%
investopedia.com
51
1.5%
Word Count Statistics
Metric
Value… See the full description on the dataset page: https://huggingface.co/datasets/kskada/crypto-education-en-corpus.Annex_7_Educational_Indicators_2024-2025
Grounded Educational Indicators — Nepali SFT Dataset
File: grounded_education_indicators_nepali_sft.jsonl
Records: 10,091
Language: Nepali (ne / ISO 639-3 npi), Devanagari script
License: Apache-2.0 (permissive)
Format: JSON Lines, ShareGPT-style conversational (human / gpt turns)
Task type: Grounded question answering over structured (tabular) educational statistics
Size on disk: ~15.5 MB
1. Overview
This dataset is a supervised fine-tuning (SFT) corpus of 10… See the full description on the dataset page: https://huggingface.co/datasets/Somtharu181coder/Annex_7_Educational_Indicators_2024-2025.MilkQA
The MilkQA Dataset
I am not the author of this dataset, this is a reproduction of the MilkQA dataset on HuggingFace, the original data can be downloaded from the link: http://nilc.icmc.usp.br/nilc/index.php/milkqa
MilkQA is a dataset of dense questions for the task of answer selection. It contains questions and answers of the dairy farming domain that were collected by the customer service of Embrapa Dairy Cattle between the years of 2003 and 2012.
The dataset currently contains 2… See the full description on the dataset page: https://huggingface.co/datasets/eduagarcia/MilkQA.