datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
git-commits-merged
Themis-Git-Commits-Merged
Overview
Themis-Git-Commits-Merged is a large-scale dataset of ~3.98M single-file code commits from permissively licensed GitHub repositories that have been cross-referenced with GHTorrent pull request data to retain only commits that are part of successfully merged, non-reverted pull requests. This provides implicit human validation of each code change — a merge decision by project maintainers confirms the intent and quality of… See the full description on the dataset page: https://huggingface.co/datasets/project-themis/git-commits-merged.sat-vl-sft-postprocessed-merged-v1
Dataset Summary
NuTonic/sat-bbox-metadata-sft-v1 is a metadata-first, procedural VLM SFT dataset built from an existing “sat-bbox” style dataset tree (Sentinel‑2 chips + per-tile JSON metadata sidecars, optionally paired Mapbox stills).
The goal is to create high-signal, production-shaped supervision for multimodal chat models:
Captioning for satellite chips
Grounding (bounding boxes in normalized coordinates) for land-cover regions
Class-focused captions and absence checks for… See the full description on the dataset page: https://huggingface.co/datasets/NuTonic/sat-vl-sft-postprocessed-merged-v1.LONGCOT-merged-1Mthis is PowerInfer/QWQ-LONGCOT-500K + PowerInfer/LONGCOT-Refine-500K shuffled together with the following changes:
pointless starting phrases at the beginning (in english) such as "Okay, ..." "So, ..." etc are removed
config en has been filtered to include only rows detected as lang en in both prompt and response columns via fasttext
PersianCorpus_merged
Persian Corpus (Merged)
Dataset Summary
Persian Corpus (Merged) is a large-scale, Persian corpus meticulously aggregated from multiple high-quality Persian datasets available on the Hugging Face Hub. Designed to advance Persian NLP research and applications, this corpus consolidates diverse textual sources into a single resource, providing researchers and developers with a robust foundation for training and evaluating language models.
Why Use This Corpus?
By… See the full description on the dataset page: https://huggingface.co/datasets/mshojaei77/PersianCorpus_merged.Fineweb-Edu-Chinese-V2.1-merged-score4_5Fineweb-Edu-Chinese-V2.1 的评分为 4~5 的数据的子集。原数据集切片很细，大约每 10MB 一个切片，本数据集做了集合，每 400 个切片合并为 1 个切片，新的切片每个大小约为 4GB。
为了便于加载，按照切片分割了子集，子集命名来源于原数据集的切片范围，可以指定加载其中的一个子集：
from datasets import load_dataset
# 加载其中的一个子集
ds = load_dataset("Mxode/Fineweb-Edu-Chinese-V2.1-merged-score4_5", "0-399")
也可以加载多个子集或者全部子集，可以通过如下方式获取子集名称：
from datasets import get_dataset_config_names
configs = get_dataset_config_names("Mxode/Fineweb-Edu-Chinese-V2.1-merged-score4_5")
print(configs)
>>> ['0-399', '1200-1599'… See the full description on the dataset page: https://huggingface.co/datasets/Mxode/Fineweb-Edu-Chinese-V2.1-merged-score4_5.jupyter-code-text-pairs-merged
Merged Jupyter Notebooks Dataset
Introduction
This dataset is a transformed version of the Jupyter Code-Text Pairs dataset. The original dataset contains markdown, code, and output pairs extracted from Jupyter notebooks. This transformation merges these components into a single, cohesive format that resembles a Jupyter notebook, making it easier to analyze and understand the flow of information.
Dataset Details
Source
The original dataset is… See the full description on the dataset page: https://huggingface.co/datasets/taher30/jupyter-code-text-pairs-merged.claudesidian-behaviors-merged
Claudesidian Merged Behavioral Dataset
Dataset Description
This dataset contains 1,852 synthetic training examples demonstrating 8 different behavioral patterns for training language models to use the Claudesidian-MCP toolset effectively with Obsidian vaults.
The dataset is specifically formatted for KTO (Kahneman-Tversky Optimization) preference learning with properly interleaved positive and negative examples.
Behavioral Categories
This dataset includes… See the full description on the dataset page: https://huggingface.co/datasets/professorsynapse/claudesidian-behaviors-merged.merged-tool-use
merged-tool-use
High-quality, multi-source dataset normalized to a single, OpenAI-style tool-calling schema. Built by unifying multiple public datasets into one consolidated corpus ready for training and evaluation.
Total examples: 220,247
Formats: Parquet and JSONL
Schema: messages: list[message] where each message has role, optional content, and optional tool_calls/function fields.
Contents
This dataset merges and normalizes the following sources:… See the full description on the dataset page: https://huggingface.co/datasets/Akicou/merged-tool-use.dark_thoughts_case_study_merged
Dark Thoughts 案例研究推理数据集
数据集描述
概述
Dark Thoughts 案例研究推理数据集是一个全面的多语言商业案例研究及相关推理响应集合。它通过先进的语言模型处理 Cablegate 电报，生成中英文商业案例研究，并进一步丰富了利益相关者特定的推理视角。对于对商业分析、多语言内容生成和推理能力感兴趣的研究人员和从业人员来说，该数据集是宝贵的资源。
支持的任务
该数据集支持以下任务：
文本生成
推理与分析
双语案例研究生成
跨语言内容分析
商业战略制定
利益相关者视角建模
语言
该数据集为双语数据集：
英语 (en)
中文 (zh)
数据集结构
数据字段
{
'id': 'int32', # 条目的唯一标识符
'response': 'string', # 生成的推理响应
'query': 'string', # 原始查询或案例研究内容
'source_data': 'string', #… See the full description on the dataset page: https://huggingface.co/datasets/DataTonic/dark_thoughts_case_study_merged.african-history-knowledge-merged-sft-cleaned
African History Knowledge Merged SFT — Cleaned
A reproducible, format-cleaned version of MaatAI/african-history-knowledge-merged-sft, pinned to source commit 0a40eb041d85d59b86219641de0fd87786ee0f77.
Split
Rows
train
28,585
validation
1,589
test
1,589
Total
31,763
Cleaning performed
Quarantined 13 training records: 12 have no final answer after a closing thinking tag, and one has ambiguous repeated closing tags. Their original text and… See the full description on the dataset page: https://huggingface.co/datasets/MaatAI/african-history-knowledge-merged-sft-cleaned.writingPrompts-merged
WritingPrompts Merged Dataset
This dataset contains the writing prompts and corresponding stories from euclaise/writingprompts with all original train/validation/test splits merged into a single unified dataset.
This version is based on the dataset used in the paper "Hierarchical Neural Story Generation" by Fan et al. (2018).
Sumtablets_Merged
SumTablets Merged Multimodal Dataset
A merged, deduplicated, quality-filtered training corpus combining cuneiform tablet text records with tablet photography and lineart, structured for vision-language model (VLM) fine-tuning.
Target model: Qwen3-VL-8B-Instruct via Unsloth StudioCombined license: CC-BY-4.0 (most restrictive of the two source licenses applies)
Quick Start
Load the dataset
from datasets import load_dataset
ds =… See the full description on the dataset page: https://huggingface.co/datasets/TRACCERR/Sumtablets_Merged.african-history-merged-v1
African History Chat (Merged v1)
Unified supervised fine-tuning (SFT) dataset of multi-turn chat examples focused on African history, colonial history, and related Q&A. Six public Hub chat datasets were normalized, concatenated, and deduplicated by conversation content into a single train split.
Rows: 26,897 · Split: train only · Format: chat (messages with role / content)
Artifacts (Parquet / JSONL / manifest): Svngoku/jobs-artifacts
What this dataset is for… See the full description on the dataset page: https://huggingface.co/datasets/Svngoku/african-history-merged-v1.merged-cot
MergeCoT Dataset
A large-scale dataset for training models to resolve git merge conflicts using chain-of-thought (CoT) reasoning. This dataset contains 87,690 examples across multiple programming languages, with detailed reasoning traces for merge conflict resolution.
Dataset Summary
MergeCoT provides paired examples of:
Base versions and two conflicting changes (a and b)
Merged results that correctly combine both changes
Chain-of-thought reasoning explaining the merge… See the full description on the dataset page: https://huggingface.co/datasets/chunyoupeng/merged-cot.openthoughts_merged_think_39k
openthoughts_merged_think_39k
Merged think-format SFT dataset (ShareGPT-style: system + conversations with
from/value), 39,874 examples, for OLMo SFT.
Composition (concatenation of two decontaminated think-format sources):
open-thoughts114k_math_20k_decontam_think — 20,000 examples sampled from
OpenThoughts-114k math, decontaminated against the OpenThoughts3 set below.
openthoughts3_math_decontam_resp_lt8192_think — 19,874 examples from… See the full description on the dataset page: https://huggingface.co/datasets/pre-to-post-olmo/openthoughts_merged_think_39k.claude-tools-sft-merged
claude-tools-sft-merged
Merged SFT dataset in ChatML format (<|im_start|> / <|im_end|>),
deduplicated and filtered, ready for instruction fine-tuning.
Covers general instruction following, reasoning (<think> traces),
function calling, coding, and multi-turn conversation.
Statistics
Metric
Value
Total examples
298,979
Duplicates removed
44,928
Min length (chars)
142
Median length (chars)
3,033
Mean length (chars)
4,231
P90 length (chars)
11… See the full description on the dataset page: https://huggingface.co/datasets/someoneatemylastsliceofpizza/claude-tools-sft-merged.dpo-merged-binarizedfinance-legal-mrc_merged-table
데이터셋 설명
shchoice/finance-legal-mrc 데이터 중 병합된 테이블만 추출한 뒤 이미지와 함께 저장한 데이터입니다.
math_merged_deduped_OR1_dapo
Math subset for training L1 using RL
This dataset is inspired by LLM360/Reasoning360(GURU92K-math), but reproduced from DAPO-Math-17K and Skywork-OR1-Math. DeepScaleR was not used for source duplications.
Dataset Details
Dataset Description
Curated by: Leon (Me)
Funded by [optional]: AIGCode/Koting Intelligence
Language(s) (NLP): Mostly in English with a few in Chinese
License: MIT (following GURU-92K)
Dataset Sources [optional]… See the full description on the dataset page: https://huggingface.co/datasets/Leon-Leee/math_merged_deduped_OR1_dapo.cybersec-reasoning-merged
Cybersecurity Reasoning Dataset (Merged)
Dataset Description
This dataset combines two high-quality cybersecurity reasoning datasets to create a comprehensive resource for training language models on security-related tasks with chain-of-thought reasoning.
Dataset Summary
Total Samples: 23,146
Languages: English
Format: Instruction-following with explicit reasoning chains
Domain: Cybersecurity (vulnerabilities, CVE/CWE mapping, security analysis)… See the full description on the dataset page: https://huggingface.co/datasets/Mohannadcse/cybersec-reasoning-merged.dpo-mergedPyThagoreans-Merged
PyThagoreans Dataset
Overview
The PyThagoreans dataset is a comprehensive collection of math problems and their solutions, designed to assist in learning and practicing mathematical problem-solving. This dataset includes a variety of problems, expected answers, and predicted answers, making it a valuable resource for students, educators, and researchers.
Dataset Details
Modalities
Text: The dataset primarily contains text data, including math… See the full description on the dataset page: https://huggingface.co/datasets/prithivMLmods/PyThagoreans-Merged.Cleaned-sharegpt_Merged-Opus-33159-ShareGPTmerged-corpus
Merged Corpus
Welcome to this repository.This dataset is japanese corpus that includes wiki, wikibooks, wikiversity, cc100, and oscar2109.
Getting Started
If you want to use this, please run as follows.This process takes about 3 hours.
mkdir -p pretrain/input/
cd pretrain/input/
GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/datasets/ce-lery/merged-corpus.git
cd merged-corpus
git lfs pull
bash merge_train.sh
landing-pages-styling-css-only-v2v3-merged
merged_v2_v3_dedup
This dataset is the deduplicated merged training set built from the repository's landing_page_v2 and landing_page_v3 pipelines.
It contains chat-format rows for CSS generation on landing pages:
messages + metadata
Each sample keeps:
messages: the training conversation, usually a user prompt plus the assistant CSS response.
metadata: generation and provenance fields such as source pipeline, style phrase, recipe, and other analysis attributes.… See the full description on the dataset page: https://huggingface.co/datasets/kogai/landing-pages-styling-css-only-v2v3-merged.rocstories-merged
ROCStories Merged
A convenience variant of the ROCStories corpus where the five individual sentences (sentence1–sentence5) are concatenated into a single text column.
Two splits are available, matching the original ROCStories splits:
spring2016 — 45,496 stories
winter2017 — 52,665 stories
Each row contains storyid, storytitle, and text (the full story with sentences joined by a space).
Skywork-DeepScaleR-Merged-Verifiable-Dedup
Skywork + DeepScaleR — Verifiable, Cross-Deduplicated
98,941 questions. Only parser-compatible references are included. Here, “verifiable” means every reference component parses with Math-Verify 0.8.0, with string fallback disabled. It does not mean that the answer has been independently proved correct or that grading model outputs is error-free.
The two independently cleaned pools were merged, cross-source duplicate questions were collapsed, unresolved cross-source answer… See the full description on the dataset page: https://huggingface.co/datasets/zbeeb/Skywork-DeepScaleR-Merged-Verifiable-Dedup.merged-data-v2
Info
This dataset is a merge of the following datasets:
flpelerin/openorca-alpaca-50k
sam-liu-lmi/databricks-dolly-15k-alpaca-style
TokenBender/roleplay_alpaca
vicgalle/alpaca-gpt4
CreitinGameplays/chat-assistant
CreitinGameplays/filter
structured_data_merged_v2v5_0222
Dataset Card for structured_data_merged_v2v5_0222
Dataset Details
Dataset Description
structured_data_merged_v2v5_0222 is a dataset for Supervised Fine-Tuning (SFT) focused on structured data format conversion tasks — specifically, interconversion among JSON, XML, YAML, TOML, and CSV.
It was created by deduplicating and merging the following two existing datasets:
u-10bei/structured_data_with_cot_dataset_512_v2 (train split only)… See the full description on the dataset page: https://huggingface.co/datasets/takami2022/structured_data_merged_v2v5_0222.merged_dataset_final_clean_v41
merged_dataset_final_clean_v41
English
Rule-based cleaned SFT dataset for structured output generation (JSON / YAML / XML / TOML / CSV).
Data Source
This dataset was built from competition-provided datasets only.
The cleaning pipeline loads the following source groups:
u-10bei (6 datasets: source ids 1-1 to 1-6)
daichira (3 datasets: source ids 2-1 to 2-3)
After strict filtering and sampling for v4.1, the final retained rows are from u-10bei sources (1-1 to… See the full description on the dataset page: https://huggingface.co/datasets/yamaTK/merged_dataset_final_clean_v41.