datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Nemotron-AIQ-Agentic-Safety-Dataset-1.0
Nemotron-AIQ Agentic Safety Dataset
Dataset Summary
Nemotron-AIQ-Agentic-Safety-Dataset is a comprehensive dataset that captures a broad range of novel safety and security contextual risks that can emerge within agentic systems. It highlights the robustness of NVIDIA's open model, llama-3.3-nemotron-super-49b-v1, when deployed as a research assistant inside AIQ, demonstrating its ability to handle a diverse spectrum of agentic safety and security challenges. The dataset… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-AIQ-Agentic-Safety-Dataset-1.0.Pluto-Nano-1.0-Pretrain-v2
ASTRAI Pluto Nano 1.0 — Pretrain Mix (v2)
Curated multilingual pretraining corpus (~50 GB parquet, ~12 B tokens after tokenization) used for ASTRAI Pluto Nano 1.0, a 1 B-total / 50 M-active MoE model with 64 k vocabulary and 5 target languages (EN, PT, ES, ZH, HI).
v2 additions vs v1: OpenThoughts3 (CoT reasoning), openstax textbooks + peS2o (science), and reweighting for better balance. NOTE: factsense (openbmb) was used at training time but is not redistributed here due to its… See the full description on the dataset page: https://huggingface.co/datasets/ASTRAI-labs/Pluto-Nano-1.0-Pretrain-v2.ifstruct-v1.0
IFStruct v1.0
[!Note]
📝 Blog post: https://www.liquid.ai/blog/ifstruct-v1.0
💻 GitHub: https://github.com/Liquid4All/ifstruct
IFStruct is a benchmark for structured-output compliance: can a model produce valid JSON/YAML that follows a requested schema, when the requirements are phrased the many different ways real users phrase them? It is scored without constrained decoding, and only the structure is judged (not content quality, extraction accuracy, or reasoning) so the… See the full description on the dataset page: https://huggingface.co/datasets/LiquidAI/ifstruct-v1.0.Turkish-SFT-Dataset-v1.0
Turkish-SFT-Dataset-v1.01
Repo: AlicanKiraz0/Turkish-SFT-Dataset-v1.0Sürüm: v1.01Lisans: MITBiçim: jsonl (kolonlar: system, user, assistant)Boyut: ~5500 satır ve satır başına 3.000–4.500 token/satır (≈ 20M+ token)Dil: Türkçe (tr)Görevler: talimat izleme, SFT, muhakeme, güvenli ret, uzun-bağlam ve araç kullanım bilinci
🔎 Özet
Bu veri kümesi, Türkçe Denetimli İnce Ayar (SFT) için tasarlanmış, yüksek kaliteli ve uzun çıktılar içeren örneklerden oluşur. İçerik 12 ana… See the full description on the dataset page: https://huggingface.co/datasets/AlicanKiraz0/Turkish-SFT-Dataset-v1.0.Solace-1.0-Omni
Project Solace
The largest verified frontier-model distillation corpus ever released.
60 datasets · 7 frontier model families · 12,586,893 unique conversations · One file · Zero filler
The short version
This is synthetic data. The best kind of synthetic data.
Every example was generated by a verified 2026 frontier model — GLM-5.2, Claude Fable 5, Mythos 5, GPT-5.6 Sol, GPT-5.5 Codex, DeepSeek V4 Pro 0813, Qwen 3.8-Max, and Kimi K3 — then… See the full description on the dataset page: https://huggingface.co/datasets/Solstice-AI/Solace-1.0-Omni.AgentDoG1.0-Training-Data
AgentDoG1.0 Training Data
[💻 GitHub] | [📊 ATBench Dataset] | [📄 ATBench Paper] | [📄 AgentDoG Paper] | [🤗 Collection]
AgentDoG1.0 Training Data releases supervised instruction-tuning data for trajectory-level AI-agent safety modeling. It is paired with the AgentDoG and ATBench line of work: ATBench is the benchmark release, while this repository contains training-oriented data for binary safety classification and fine-grained taxonomy diagnosis.
Introduction… See the full description on the dataset page: https://huggingface.co/datasets/AI45Research/AgentDoG1.0-Training-Data.LatamGPT-Corpus-1.0
LatamGPT-Corpus-1.0
🌐 Language versions: English | Español | Português
🔗 Project links: Official LatamGPT website | Corpus dashboard
🤖 Associated model: The complete LatamGPT corpus—of which this repository contains the openly released portion—was used in the training process of Llama-3.1-70B-LatamGPT-SFT-1.0.
Dataset description
Summary
LatamGPT-Corpus-1.0 is the open release of the data corpus assembled for the continued pretraining of… See the full description on the dataset page: https://huggingface.co/datasets/latam-gpt/LatamGPT-Corpus-1.0.magpie-sft-v1.0
magpie-sft-v1.0
This repository provides an instruction-tuning dataset developed by LLM-jp, a collaborative project launched in Japan.
This is a dataset of instruction and response pairs created using the Magpie method.
cyberagent/calm3-22b-chat was used for generating the instructions, and Qwen/Qwen2.5-32B-Instruct was used for generating the responses.
Send Questions to
llm-jp(at)nii.ac.jp
Model Card Authors
The names are listed in alphabetical order.… See the full description on the dataset page: https://huggingface.co/datasets/llm-jp/magpie-sft-v1.0.Luau-Coder-1.0-Preview-SFT
Luau Coder 1.0 Preview SFT 🦭
This dataset is exceptionally high-quality supervised fine-tuning conversations for a highly capable coding model in Roblox Luau domain.
It prioritize technical correctness, useful engineering judgment, realistic interaction, and efficient explanations over output volume.
This dataset includes & covering:
Multi-turns (4-10 turns)
Dynamic CoT (length)
Dynamic Interleaved Reasoning
Long Context Session
Q/A
Review
Debugging
Bug Fix… See the full description on the dataset page: https://huggingface.co/datasets/khtsly/Luau-Coder-1.0-Preview-SFT.dclm-baseline-1.0-parquet_urls
Dataset Card for dclm-baseline-1.0-parquet_urls
This dataset provides the URLs and top-level domains associated with training records in mlfoundations/dclm-baseline-1.0-parquet. It is part of a collection of datasets curated to make exploring LLM training datasets more straightforward and accessible.
Dataset Details
Dataset Description
This dataset was created by downloading the source data, extracting URLs and top-level domains, and retaining only… See the full description on the dataset page: https://huggingface.co/datasets/nhagar/dclm-baseline-1.0-parquet_urls.Ko-Agent-Trajectories-1.0
Ko-Agent-Trajectories-1.0
Dataset card v1.1.1 (2026-09-22). The pipeline code is now released in this repository
under pipeline/, together with the API catalogue, the scenario templates and the complete
prompt set. The card reports the completed human review study and the v1.1 artefacts
(behaviour DPO config, per-item validation scores, manifest, filter asset).
Korean edition: README.ko.md.
TL;DR
A Korean multi-turn agent ↔ tool trajectory corpus synthesized… See the full description on the dataset page: https://huggingface.co/datasets/taejoon89/Ko-Agent-Trajectories-1.0.Qiita-1.07MThis dataset contains 1,074,174 articles published on Qiita.
midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs
midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs
Pre-tokenized MIDI pieces for IsoFLOP scaling-law runs. Each row is one full
piece (no time-windowing); training crops sequences from packed token bins.
The source column is the original piece metadata as JSON so a row can be
traced back to its EPR Labs source dataset.
Based on MIDI datasets gathered by EPR Labs.
Codec
name: dyadic
tokenizer vocab size: 512
max_time_step: 1.0
n_velocity_bins: 32… See the full description on the dataset page: https://huggingface.co/datasets/wmatejuk/midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512-epr-labs.Nemotron-AIQ-Agentic-Safety-Dataset-1.0
Nemotron-AIQ Agentic Safety Dataset
Dataset Summary
Nemotron-AIQ-Agentic-Safety-Dataset is a comprehensive dataset that captures a broad range of novel safety and security contextual risks that can emerge within agentic systems. It highlights the robustness of NVIDIA's open model, llama-3.3-nemotron-super-49b-v1, when deployed as a research assistant inside AIQ, demonstrating its ability to handle a diverse spectrum of agentic safety and security challenges. The dataset… See the full description on the dataset page: https://huggingface.co/datasets/yuqing1207/Nemotron-AIQ-Agentic-Safety-Dataset-1.0.tactical-military-reasoning-v.1.0
Tactical Military Reasoning Dataset v1.0
A curated collection of 150 rich tactical military scenarios with LLM-generated reasoning strategies for both attacking and defending forces.
📝 Preface
Oncologists do not study cancer because they love cancer and wish for it to occur more frequently. They study cancer to better understand its causes, progression, and consequences in order to therefore eradicate it from the earth more effectively. A distaste for something… See the full description on the dataset page: https://huggingface.co/datasets/ZennyKenny/tactical-military-reasoning-v.1.0.cyberusecase-v1.0
Cybersecurity SOC Fine-Tuning Dataset — 17.5k Real CVEs (2018–2026) + SOC Knowledge
A large supervised fine-tuning (SFT) dataset for teaching an LLM expert-level
cybersecurity reasoning across vulnerability management, SOC alert triage, detection
engineering, threat intelligence & hunting, incident response, and cloud/DevSecOps.
It combines 17,590 real CVEs (2018–2026) pulled from the NIST NVD data feeds with a
hand-curated set of 65 landmark CVEs (rich, multi-angle coverage)… See the full description on the dataset page: https://huggingface.co/datasets/ronaldocloud/cyberusecase-v1.0.stem-reasoning-v1.0.0-ccbysa-001
YouAI Data — stem-reasoning-v1.0.0-ccbysa-001
Dataset Description
YouAI Data — 1,000 STEM training examples extracted from verified CC-BY-SA expert sources — real domain experts solving real problems, not synthetic LLM generation. Contains 394 step-by-step reasoning chains and 569 instruction/response pairs across 332 unique sources. Every example traces to a source URL, available source metadata, and verified license. Includes 364 DPO preference pairs as a free companion… See the full description on the dataset page: https://huggingface.co/datasets/YouAIData/stem-reasoning-v1.0.0-ccbysa-001.ROMB-1.0
♦ ROMB
Русское описание и инструкция
ROMB (Russian Olympiad Math Benchmark) evaluates models on Russian-language
school olympiad mathematics. The test set contains 2552 text-only tasks:
1716 arithmetic/other tasks, 644 logic tasks, and 192 geometry tasks. Tasks have
typed answers, answer-format notes, and per-task checking rules.
The evaluator also supports configurable v3 runs: native thinking, optional
JSON Schema constrained decoding, plain or \boxed{…} answers, and… See the full description on the dataset page: https://huggingface.co/datasets/d0rj/ROMB-1.0.ManipuriGPT-Corpus-v1.0
ManipuriGPT Corpus v1.0
ManipuriGPT Corpus v1.0 is a research-grade, multi-script, deduplicated, and quality-scored corpus specifically engineered for pretraining Manipuri (Meiteilon) language foundation models.
Quick Summary
Total Sequences: 147,956
Total Tokens (ManipuriGPT-Tokenizer-v1.0): 4,347,075
Total Characters: 16,019,401
Pipeline Version: 5.6
Release Version: v1.0.0
Build Timestamp: 2026-07-25T09:17:21.960438Z
Primary Writing Systems… See the full description on the dataset page: https://huggingface.co/datasets/nanskong/ManipuriGPT-Corpus-v1.0.qa-expert-multi-hop-qa-V1.0
Dataset Card for QA-Expert-multi-hop-qa-V1.0
This dataset aims to provide multi-domain training data for the task: Question Answering, with a focus on Multi-hop Question Answering.
In total, this dataset contains 25.5k for training and 3.19k for evaluation.
You can take a look at the model we trained on this data: https://huggingface.co/khaimaitien/qa-expert-7B-V1.0
The dataset is mostly generated using the OpenAPI model (gpt-3.5-turbo-instruct). Please read more information about… See the full description on the dataset page: https://huggingface.co/datasets/khaimaitien/qa-expert-multi-hop-qa-V1.0.JL-ActionBoundary-1K-v1.0.0
JL-ActionBoundary-1K v1.0.0
Counterfactual Ask–Inspect–Act–Defer supervision for coding agents
JL-ActionBoundary-1K teaches a coding agent to choose the correct next policy before changing code:
ACT: the task is sufficiently specified for bounded repository work;
INSPECT: missing information can be recovered from the repository;
ASK: a material product decision belongs to the user;
DEFER: live execution authority or rollback ownership is missing.… See the full description on the dataset page: https://huggingface.co/datasets/jumplander/JL-ActionBoundary-1K-v1.0.0.OpenOcypus-1.0
🪲 OpenOcypus-1.0
📖 Description
OpenOcypus-1.0 is the first release in the OpenOcypus series — a collection of high‑quality SFT datasets designed to evolve over time. Future versions will introduce additional sources, refined filtering, and expanded task coverage.
Total size: 1,157,428 examples.
The dataset is designed to create a versatile assistant capable of:
🗣️ Engaging in natural conversations
🧮 Solving math problems with step‑by‑step explanations
💻… See the full description on the dataset page: https://huggingface.co/datasets/zait-ai/OpenOcypus-1.0.Tucan-BG-v1.0
Tucan-BG Dataset v1.0
Bilingual Function Calling Training Dataset for Bulgarian Language Models 🇧🇬
📄 Supporting the Tucan model series
Paper: https://arxiv.org/abs/2506.23394
Overview 🚀
Tucan-BG-v1.0 is a bilingual (Bulgarian/English) dataset containing 10,035 conversations specifically designed for training language models in function calling and tool use capabilities. This dataset enables the development of AI agents capable of determining when to use… See the full description on the dataset page: https://huggingface.co/datasets/llm-bg/Tucan-BG-v1.0.han-instruct-dataset-v1.0
Dataset Card for "han-instruct-dataset-v1.0"
The newest dataset version is https://huggingface.co/datasets/pythainlp/han-instruction-dataset.
Dataset Summary
🪿 Han (ห่าน or goose) Instruct Dataset is a Thai instruction dataset by PyThaiNLP. It collect the instruction following in Thai from many source.
Many question are collect from Reference desk at Thai wikipedia.
Data sources:
Reference desk at Thai wikipedia.
Law from justicechannel.org… See the full description on the dataset page: https://huggingface.co/datasets/pythainlp/han-instruct-dataset-v1.0.OpenTopics-1.0-20K
OpenTopics-1.0-20K
What is this dataset?
OpenTopics-1.0-20K is a collection of 20,003 topic names spanning a wide variety of subjects, including physics, medicine, history, law, engineering, and the arts.
AtomixLabs built this dataset to help developers, researchers, and AI builders who need a large, organized list of topics. It works great for creating synthetic prompts, testing search systems, and training models to classify text.
What is inside… See the full description on the dataset page: https://huggingface.co/datasets/AtomixLabs/OpenTopics-1.0-20K.midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512
midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512
Pre-tokenized MIDI pieces for IsoFLOP scaling-law runs. Each row is one full
piece (no time-windowing); training crops sequences from packed token bins.
The source column is the original piece metadata as JSON so a row can be
traced back to Maestro, GiantMIDI, ATEPP, or MusicNet.
Based on MIDI datasets gathered by EPR Labs.
Codec
name: dyadic
tokenizer vocab size: 512
max_time_step: 1.0
n_velocity_bins: 32… See the full description on the dataset page: https://huggingface.co/datasets/wmatejuk/midi-tokens-dyadic-tu0.01-vb32-mts1.0-vocab512.OmniRoute-SFT-1.0-32K
OmniRoute-SFT-1.0-32K
AS OF 7/31/2026
The first large-scale supervised dataset for training LLM routing models.
Every day, production AI systems face a deceptively simple question: which model should handle this query? A one-line cooking question doesn't need a 400-billion-parameter reasoning engine. A graduate-level proof doesn't belong on a chat-optimized 8B model. Routing gets this right — and saves orders of magnitude in compute — but until now, there has been no… See the full description on the dataset page: https://huggingface.co/datasets/AtomixLabs/OmniRoute-SFT-1.0-32K.thai_food_v1.0
Thai Food Recipe dataset v1.0
The Thai Food Recipe dataset is a collection of Thai recipes from old Thai books and social networks.
List Book
ตำรับอาหาร - เตื้อง สนิทวงศ์, ม.ร.ว., 2426-2510 - Work in process (ยังไม่ครบ) in v2.0
ผัดกะเพรา - “ทีมครัวเนื้อหอม” จ.ลำปาง
สูตร "เกี๊ยวกุ้ง"
License: cc0-1.0
ATANTV1.0-corpus
ATANT Narrative Test Corpus
Automated Test for Acceptance of Narrative Truth, v1.0
The first open evaluation corpus for measuring continuity in AI systems: the ability to persist, update, disambiguate, and reconstruct meaningful context across time.
Paper: ATANT: An Evaluation Framework for AI Continuity (arXiv:2604.06710)
Standard repository: github.com/Kenotic-Labs/ATANT
Author: Samuel Sameer Tanguturi
Affiliation: Kenotic Labs
Published: April 2026
Why this corpus… See the full description on the dataset page: https://huggingface.co/datasets/Kenotic-Labs/ATANTV1.0-corpus.NemoSlides-DPO-mix-v1.0
Slide-DPO
Direct Preference Optimization dataset for training LLMs to generate slide
presentations in Slidev markdown format, derived from the
Slides-Align human
preference rankings over the
SlidesGen-Bench benchmark.
Each row is a preference pair: a brief plus an available image pool as the
prompt, and two Slidev-markdown responses (with <think> reasoning traces)
that were generated by differently-ranked AI slide-generation products for
the same brief.
Row schema… See the full description on the dataset page: https://huggingface.co/datasets/trillionlabs/NemoSlides-DPO-mix-v1.0.