datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ChatTS-Training-Dataset
ChatTS-Training Data
This repository contains the training data for the ChatTS project. This is the dataset for training the ChatTS-14B model.
Datasets
align_256: Alignment training dataset for stage-1 alignment training, with SEQ_LEN=256.
align_random: Alignment training dataset with random sequence lengths between 64 and 1024.
sft: SFT dataset generated with Time Series Evol-Instruct.
ift: Instruction following dataset.
dev: A small dataset for development and testing.… See the full description on the dataset page: https://huggingface.co/datasets/ChatTSRepo/ChatTS-Training-Dataset.dnd-35-training-dataset
D&D 3.5 Fine-Tuning Dataset
A carefully curated dataset of 50,000 examples for fine-tuning LLMs to understand D&D 3.5 mechanics.
Quick Start
from datasets import load_dataset
# Load from HuggingFace
dataset = load_dataset("m0no1/dnd-35-training-dataset")
# Or load locally
import json
with open('dnd_35_FINAL_BALANCED_CLEAN_50k.jsonl', 'r') as f:
data = [json.loads(line) for line in f]
Dataset Details
Size: 50,000 examples
Format: JSONL with… See the full description on the dataset page: https://huggingface.co/datasets/m0no1/dnd-35-training-dataset.SciDocBench-Training-Data
SciDocBench Training Data
Training data accompanying SciDocBench
(paper) for scientific document understanding.
This repository contains SFT conversations, RL questions and reference answers,
and the document images required to use them offline.
Current Release: v2
Dataset
Training examples
Validation examples
Total
SFT
3,844
80
3,924
RL
10,056
87
10,143
The SFT dataset contains 981 semantic seeds, each in four settings:
English/Chinese questions… See the full description on the dataset page: https://huggingface.co/datasets/HenryExcellent/SciDocBench-Training-Data.SAND-Post-Training-Dataset
SAND-Post-Training-Dataset: High-Quality Synthetic Reasoning Dataset Built with AMD GPUs
Dataset Summary
We introduce the SAND-Post-Training-Dataset, a high-quality synthetic reasoning dataset for mathematics and science built entirely using a synthetic data pipeline running on the AMD ROCm™ stack and AMD Instinct™ MI325 GPUs.
This dataset prioritizes difficulty and novelty over volume, demonstrating that high-difficulty synthetic data can elevate… See the full description on the dataset page: https://huggingface.co/datasets/amd/SAND-Post-Training-Dataset.hipaa-compliance-training
HIPAA Compliance Training Dataset
Dataset Description
The first comprehensive HIPAA compliance training dataset for LLM fine-tuning, covering the Security Rule, Privacy Rule, Breach Notification Rule, and implementation guidance from NIST and FDA.
Dataset Summary
Total Examples: 1,287 (1,029 train / 258 validation)
Source Documents: 9 federal publications (~5.6 MB extracted content)
Format: JSONL with chat-formatted messages
License: CC0-1.0 (Public Domain)… See the full description on the dataset page: https://huggingface.co/datasets/ethanolivertroy/hipaa-compliance-training.procedural-reasoning-training-pool
Procedural reasoning training pool
Reasoning questions from 101 procedural generators, each of which writes a question, computes its
own answer and ships a verifier that scores an attempt at it, plus a collection of solved Sudoku
puzzles. Every answer is short and exactly checkable, so a trained model can be marked against the
key by a program and no judge is needed. Laid out twice. Train on either layer or on both.
pool.jsonl
Every generator rewritten into one… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/procedural-reasoning-training-pool.k3-sft-cc0-flan
Dataset Card for K3 SFT CC0 FLAN
844-row Kimi K3 synthetic instruction-tuning shard built from DPI-traced CC0/public-domain
FLAN prompts in the Tülu mix. Four overlapping Hub configs expose different cohort
views; adaptive is the recommended default for quality-conscious SFT mixing.
Dataset Details
Curated by: Training Datasmith
Teacher: kimi-k3 via deltafin (local inference)
Languages: English prompts; translation pairs include German, Spanish, Czech, Igbo… See the full description on the dataset page: https://huggingface.co/datasets/Training-Datasmith/k3-sft-cc0-flan.reading-comprehension-training-pool
Reading comprehension training pool
Public reading comprehension questions from six datasets, each a question about a passage with an
answer that is a span of it, a number or a date, read at the pinned revisions named below and laid
out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 310728 rows, one JSON object per line, with these fields.
Field
What it holds
id
a row identifier unique within this file… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/reading-comprehension-training-pool.hypencoder-msmarco-training-datasetThe MSMARCO training data used to train the models from Hypencoder: Hypernetworks for Information Retrieval
.
Dataset Overview
This dataset is based on the MSMARCO Passage dataset and includes all the queries which have a positive passage in the original dataset (there are additional queries with no positive passages which we do not use). Each query has the known positive passage as well as 200 additional passages. These additional passages may be unlabeled positives or negatives.… See the full description on the dataset page: https://huggingface.co/datasets/jfkback/hypencoder-msmarco-training-dataset.gaiasky-training-dataset
Gaia Sky Expert Dataset
This dataset is designed for fine-tuning Large Language Models to become experts in the Gaia Sky ecosystem. It covers 3D astronomical visualization, Java engine architecture, Python scripting API, and GLSL shader logic.
Dataset Structure
The repository is organized into two primary configurations:
1. Distilled (Instruction-Tuned)
File: train.jsonl
Format: {"instruction": "...", "output": "...", "source_file": "..."}
Description:… See the full description on the dataset page: https://huggingface.co/datasets/Langurmonkey/gaiasky-training-dataset.code-training-il
Code-Training-IL
A 40,330-example instruction-tuning dataset for code: 20K Python (NVIDIA OpenCodeInstruct, test-filtered) + 20K TypeScript + 330 hand-written bilingual identity examples.
Overview
code-training-il is a curated, filtered instruction-tuning corpus for training small coding assistants. It is the dataset used to fine-tune code-il-E4B, a 4B on-device model.
The dataset was designed around a thesis: less data, better filtered, beats more data. The… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/code-training-il.grade-school-math-training-pool
Grade school math training pool
Public training data for grade school math word problems, gathered from 11 sources,
2,699,281 distinct problems in all. The pool ships in two layers holding the same rows, so you can
take whichever suits your pipeline.
normalised/ every source in one format, one row per distinct question, in 6 gzipped jsonl shards
sources/ every source as it was downloaded, in its own file format with its own fields
README.md this file… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/grade-school-math-training-pool.zignet-training-dataset
ZigNet Training Dataset
Curated dataset of Zig programming examples for LLM fine-tuning
This dataset was created for the ZigNet project to train language models on Zig programming language patterns, idioms, and documentation.
Dataset Structure
Files
data/training/
├── dataset-train.jsonl # 9,629 examples (70%)
├── dataset-validation.jsonl # 2,063 examples (15%)
├── dataset-test.jsonl # 2,064 examples (15%)
└── dataset-stats.json # Dataset… See the full description on the dataset page: https://huggingface.co/datasets/fulgidus/zignet-training-dataset.medical-training-il
Medical-Training-IL
A bilingual (Hebrew / English) medical instruction-tuning corpus — curated for training small, on-device medical models for Israeli residents preparing for Stage A exams.
Overview
medical-training-il is a curated, bilingual medical instruction-tuning dataset designed to fine-tune language models for Israeli clinical reasoning. It combines high-quality English medical QA (USMLE-style, basic sciences, research-grounded) with ~5,000 Hebrew-native… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/medical-training-il.Training-Ai-Islamic-Dataset
🕌 Training AI Islamic Dataset
18.7M passages from classical Islamic books spanning 1,400 years of scholarship.
Comprehensive collection of Islamic texts covering Quran, Hadith, Fiqh, Tafsir, Aqeedah, Seerah, Usul al-Fiqh, and Arabic Language — structured with scholarly metadata for RAG and LLM training.
📊 Dataset Structure
collections/: Categorized Islamic passages compressed in JSONL format.
metadata/: Scholarly master catalogs, author biographical death… See the full description on the dataset page: https://huggingface.co/datasets/hozifa1/Training-Ai-Islamic-Dataset.tcm-divination-training
TCM & Divination Training Dataset v2
Comprehensive training dataset for Bazi, Tử Vi (Zi Wei Dou Shu), TCM, and divination domains.
Dataset Summary
Metric
Value
Total Unique Samples
162,384
File Size
651 MB
Languages
Vietnamese, English, Chinese
Last Updated
2026-01-11
Data Sources
Source
Unique Samples
Description
bazi_books
74,533
Extracted from Bazi/Tử Vi books (OCR)
gpt_training_ready
48,551
GPT-generated Q&A pairs… See the full description on the dataset page: https://huggingface.co/datasets/jakeveo05/tcm-divination-training.solutions-training-v4
Parameter Golf Auxiliary Dataset V4 (20,000 examples)
This dataset is a synthetic auxiliary training corpus designed around three concrete model failure modes observed during iterative work on OpenAI Parameter Golf submissions.
The dataset is built mainly from the following recurring failure patterns:
Hallucination / premature guessing instead of calibrated uncertainty
The model answers too early from weak context.
It should ask for high-impact missing variables first.
It should… See the full description on the dataset page: https://huggingface.co/datasets/8Planetterraforming/solutions-training-v4.screenshot-training-naive-top2-hn-ablation
Chrisyichuan/screenshot-training-naive-top2-hn-ablation
Ablation variant of
Chrisyichuan/screenshot-training-natural-filtered-v2.
Same queries, same positives. Only neg_chunk_paths differ.
The filtered-v2 dataset applies a Gemini VLM judge to filter false negatives
out of the retrieved candidates. This ablation set skips that filter entirely:
for every (query, chunk_path), we hit the text-retrieval search API for the
top-10 results and keep the first two non-positive hits as… See the full description on the dataset page: https://huggingface.co/datasets/Chrisyichuan/screenshot-training-naive-top2-hn-ablation.Mid-Training_data_of_separate_domains
Breaking the Data Barrier – Building GUI Agents Through Task Generalization
This is the official dataset repository of GUIMid
1. Data Overview
AgentBoard is composed of 9 diverse tasks: 7 vision and language tasks and 4 lanuage only tasks.
The performances of different domains as mid-training data are as follows:
Domains
Observation
WebArena (PR)
WebArena (SR)
AndroidWorld (SR)
GUI Post-Training Only
Image
26.3
6.2
9.0
Public Baselines
GPT-4o-2024-11-20
Image… See the full description on the dataset page: https://huggingface.co/datasets/MidGUI/Mid-Training_data_of_separate_domains.System-Prompt-Instruction-Real-world-Implementation-Training-set
SPIRIT Dataset (System Prompt Instruction Real-world Implementation Training-set)
Dataset Summary
SPIRIT is a high-quality system prompt instruction dataset designed to enhance language models' ability to follow complex system prompts. The dataset comprises real-world system prompts collected from GitHub repositories and synthetically generated conversations, specifically curated to improve system prompt adherence in large language models.
Dataset Creation… See the full description on the dataset page: https://huggingface.co/datasets/EricLu/System-Prompt-Instruction-Real-world-Implementation-Training-set.aultra-unified-training-data
AUltra Unified Training Data
This dataset package contains the reconstructed chat-format training data used for the AUltra Unified defensive cybersecurity and code-assistant fine-tune.
The dataset was reconstructed from the original preparation scripts, deterministic seeds, local Hugging Face cache, and the same public upstream dataset. The reconstructed split sizes match the documented training run.
Transparency Notice
This dataset is an experimental, partially… See the full description on the dataset page: https://huggingface.co/datasets/Anes-03/aultra-unified-training-data.cmmc-training-data-2026-08-31
[!WARNING]
EXPIRED VERSION. This release has been superseded by
Nathan-Maine/cmmc-training-data-2026-09-16. Regulations change continuously —
do not train compliance models on this version. It remains
available for reproducibility and provenance only.
CMMC Training Data — 2026-08-31
A curated training corpus (train + validation splits) for fine-tuning small- and mid-size language models on CMMC 2.0, NIST SP 800-171/172, and related defense compliance frameworks. This is… See the full description on the dataset page: https://huggingface.co/datasets/Nathan-Maine/cmmc-training-data-2026-08-31.qwen_qa_pairs_cli_training.jsonl
Data sources
Multiple datasets from Hugging Face related to natural language to CLI pairs were gathered.
Human reviewed synthetic data from Claude Opus4.6 and ChatGPT4.5 were added.
A handful of grounding rows related to the organisation "Spicy Lemonade" were added (see details below)
Data processing
As part of the processing, data was converted to the Alpaca format with instruction (natural language), input (typically blank) and output (the CLI command) columns.
The… See the full description on the dataset page: https://huggingface.co/datasets/spicy-lemonade/qwen_qa_pairs_cli_training.jsonl.legal-training-il
Legal-Training-IL
A 17,613-example bilingual instruction-tuning corpus for Israeli legal reasoning — covering rulings, statutes, citizen-rights pages, and contract clauses.
Overview
legal-training-il is a curated, bilingual (Hebrew / English) instruction-tuning dataset designed to adapt general-purpose language models to Israeli legal work. It was built to train law-il-E2B, a 2B-parameter on-device legal assistant.
The dataset is not a scraped dump. Every example… See the full description on the dataset page: https://huggingface.co/datasets/BrainboxAI/legal-training-il.aisec-training-data
AI Security Framework Fine-Tuning Dataset
A supervised fine-tuning dataset for training language models to perform
structured cross-framework AI security and risk management analysis.
Covers four frameworks:
NIST AI RMF 1.0 — AI risk governance (Govern / Map / Measure / Manage)
MITRE ATLAS — Adversarial ML threats, TTPs, and detection
OWASP AI Exchange — Runtime attack surfaces and technical controls
Google SAIF — Secure AI Framework component responsibility model… See the full description on the dataset page: https://huggingface.co/datasets/dbristol/aisec-training-data.slipstream-training
SlipStream Training Dataset
Training corpus for the SlipStream multi-agent system, sourced from Cyclone Vector DB semantic knowledge base.
Dataset Structure
Cyclone Exports ()
File
Rows
Description
\
1,019
All conversation pairs merged
\
179
CN-CLI instruction tuning (React, TypeScript, PostgreSQL)
\
331
Cross-domain analogical reasoning (mechanism x problem)
\
147
DPO preference pairs from vortex critic scoring
\
421
Temporal… See the full description on the dataset page: https://huggingface.co/datasets/EylstrumCore/slipstream-training.crisis-response-training-v2
Crisis Response Training Dataset
A synthetic dataset of 2,000 training examples for fine-tuning language models on crisis response scenarios. Each example includes structured responses from both civilian and first responder perspectives.
Dataset Description
This dataset contains 2,000 instruction examples in Unsloth Alpaca format, generated synthetically using large language models (LLMs) for training crisis response systems. The data is designed to help models learn… See the full description on the dataset page: https://huggingface.co/datasets/ianktoo/crisis-response-training-v2.luganda-fln-training-data
Luganda FLN Training Data
Training data for foundational literacy and numeracy (FLN) models targeting Ugandan primary school teachers (P1–P3). Designed to train small language models (1B parameters) to generate pedagogically sound content in Luganda and English.
Dataset Description
This dataset contains 1,368 training examples across four complementary splits, each targeting different aspects of teacher pedagogical content knowledge for early literacy instruction.… See the full description on the dataset page: https://huggingface.co/datasets/CraneAILabs/luganda-fln-training-data.solutions-training
Parameter Golf Auxiliary Dataset (PG-CCE)
This dataset is designed to reduce entropy in language models by addressing three critical failure modes:
Hallucination / uncertainty calibration
Long-context consistency and state tracking
Exact discrete reasoning (numbers, code, structured patterns)
It is intended as a small auxiliary training dataset to improve bits-per-byte (BPB) performance in Parameter Golf.
Usage
Mix as auxiliary data:
97% main dataset (FineWeb)
3%… See the full description on the dataset page: https://huggingface.co/datasets/8Planetterraforming/solutions-training.solutions-training-v2
Parameter Golf Auxiliary Dataset — PG-CCE-200
PG-CCE-200 is a compact auxiliary training dataset designed to target three failure modes that hurt next-token modeling quality in compact language models:
Hallucination / overconfident guessing
Weak long-context state tracking
Fragility on exact discrete structure (numbers, filenames, delimiters, ordered lists, exact project state)
This dataset was derived from real failure patterns observed during iterative model training, repository… See the full description on the dataset page: https://huggingface.co/datasets/8Planetterraforming/solutions-training-v2.