datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
soc-ratchakitcha
Royal Gazette Thailand (Ratchakitcha) Dataset
ชุดข้อมูลราชกิจจานุเบกษา (แบบ Machine Readable)
โครงการ Open Law Data Thailand ร่วมกับคณะกรรมาธิการการพาณิชย์และการอุตสาหกรรม วุฒิสภา ได้รับความอนุเคราะห์ข้อมูลจาก สำนักเลขาธิการคณะรัฐมนตรี (สลค.) เพื่อเผยแพร่ข้อมูลกฎหมายไทยสู่สาธารณะในรูปแบบที่ประมวลผลได้ด้วยคอมพิวเตอร์ (Machine Readable) เพื่อส่งเสริมนวัตกรรม Legal Tech และ AI ของประเทศไทย
Dataset Description
ชุดข้อมูลนี้รวบรวมรายการประกาศในราชกิจจานุเบกษา… See the full description on the dataset page: https://huggingface.co/datasets/open-law-data-thailand/soc-ratchakitcha.medical-o1-reasoning-SFT
News
[2025/04/22] We split the data and kept only the medical SFT dataset (medical_o1_sft.json). The file medical_o1_sft_mix.json contains a mix of medical and general instruction data.
[2025/02/22] We released the distilled dataset from Deepseek-R1 based on medical verifiable problems. You can use it to initialize your models with the reasoning chain from Deepseek-R1.
[2024/12/25] We open-sourced the medical reasoning dataset for SFT, built on medical verifiable problems and an… See the full description on the dataset page: https://huggingface.co/datasets/FreedomIntelligence/medical-o1-reasoning-SFT.RoadmapBench
RoadmapBench
A benchmark for evaluating AI coding agents on multi-target, long-horizon software development tasks derived from open-source project version upgrades.
Overview
RoadmapBench contains 115 tasks spanning 17 open-source repositories across 5 programming languages (Python, TypeScript, Go, Rust, C++). Each task requires an agent to implement multiple interdependent features that correspond to a real version upgrade of the target project.
Quick Start… See the full description on the dataset page: https://huggingface.co/datasets/UnipatAI/RoadmapBench.RegexEval
Dataset Card for RegexEval
Re(gEx|DoS)Eval is a framework that includes a dataset of 762 regex descriptions (prompts) from real users, refined prompts with examples, and a robust set of tests.
Dataset Details
Dataset Sources
Repository: https://github.com/s2e-lab/RegexEval
Paper: https://s2e-lab.github.io/preprints/icse_nier24-preprint.pdf
Dataset Structure
dataset.jsonl: dataset file in jsonl format. Every line contains a JSON object with… See the full description on the dataset page: https://huggingface.co/datasets/s2e-lab/RegexEval.RWKU
Dataset Card for Real-World Knowledge Unlearning Benchmark (RWKU)
Dataset Summary
RWKU is a real-world knowledge unlearning benchmark specifically designed for large language models (LLMs).
This benchmark contains 200 real-world unlearning targets and 13,131 multi-level forget probes, including 3,268 fill-in-the-blank probes, 2,879 question-answer probes, and 6,984 adversarial-attack probes.
RWKU is designed based on the following three key factors:
For the task setting… See the full description on the dataset page: https://huggingface.co/datasets/jinzhuoran/RWKU.soc-builder-rtl-v1
SoC Builder RTL Dataset — v1 (Experiment Release)
A reproducible, machine-generated corpus of synthesizable System-on-Chip (SoC) RTL designs for machine learning on hardware: RTL representation learning today, and — as the corpus grows — netlist, timing, and placement prediction. Every design is a complete, hierarchical, lint-clean Verilog SoC assembled from real open-source IP — RISC-V CPU cores, network-on-chip (NoC) interconnects, accelerators, peripherals, memories and… See the full description on the dataset page: https://huggingface.co/datasets/hasankursun/soc-builder-rtl-v1.rlvr-reward-hacking-scale-no-conftest-20260909-completion
Matched no-conftest RLVR study 20260909-completion
Lossless research records, grouped by model and trajectory type. Only the listed
configurations have published records. Canary diagnostics are excluded from study
estimates; run status in provenance distinguishes retired diagnostics from active
or completed training. Valid failures, refusals and truncations are retained.
The train split name is a dataset-loader convention; record_type identifies
whether a record is training… See the full description on the dataset page: https://huggingface.co/datasets/lucabaroni/rlvr-reward-hacking-scale-no-conftest-20260909-completion.alpaca-data-gpt4-chineseAudio2Tool
Audio2Tool: Speak, Call, Act — A Dataset for Benchmarking Speech Tool Use
Authors: Ramit Pahwa1,∗,∗∗, Apoorva Beedu1,∗, Parivesh Priye1, Rutu Gandhi†1, Saloni Takawale†1, Aruna Baijal1, Zengli Yang1
1 Rivian & Volkswagen Technologies · ∗ equal contribution · ∗∗ corresponding author · † equal contribution
📄 Project page / demo: https://audio2tool.github.io/
📦 Dataset: https://huggingface.co/datasets/RVtech/Audio2Tool
✉️ Contact (corresponding… See the full description on the dataset page: https://huggingface.co/datasets/RVtech/Audio2Tool.SlimPajama-Meta-rater
Annotated SlimPajama Dataset
Dataset Description
This dataset contains the first fully annotated SlimPajama dataset with comprehensive quality metrics for data-centric large language model research. The dataset includes approximately 580 billion tokens from the training set of the original SlimPajama dataset, annotated across 25 different quality dimensions.
Note: This dataset contains only the training set portion of the original SlimPajama dataset, which is why the… See the full description on the dataset page: https://huggingface.co/datasets/opendatalab/SlimPajama-Meta-rater.RoadmapBench
RoadmapBench
A benchmark for evaluating AI coding agents on multi-target, long-horizon software development tasks derived from open-source project version upgrades.
Overview
RoadmapBench contains 115 tasks spanning 17 open-source repositories across 5 programming languages (Python, TypeScript, Go, Rust, C++). Each task requires an agent to implement multiple interdependent features that correspond to a real version upgrade of the target project.
Task Structure… See the full description on the dataset page: https://huggingface.co/datasets/benchmark-anon-2026/RoadmapBench.newswire
Dataset Card for NewsWire
Dataset Summary
NewsWire contains 2.7 million unique public domain U.S. news wire articles, written between 1878 and 1977. Locations in these articles are georeferenced, topics are tagged using customized neural topic classification, named entities are recognized, and individuals are disambiguated to Wikipedia using a novel entity disambiguation model.
Languages
English (en)
Dataset Structure
Each year in the dataset is… See the full description on the dataset page: https://huggingface.co/datasets/dell-research-harvard/newswire.Lego-RL-2699
SWE-Lego-RL-2699
2,699 executable, difficulty-filtered SWE tasks for agentic RL, shipped in two
parallel views of the same instances:
View
Path
What it is
Official OpenSWE records
openswe_official_2699/
The original upstream GAIR/OpenSWE rows for exactly these 2,699 instances
Harbor RL environments
openswe_harbor_2699/
The same instances converted into ready-to-run task directories (+ the training index)
Both views cover the identical 2,699 instance_ids. The… See the full description on the dataset page: https://huggingface.co/datasets/Lego-X/Lego-RL-2699.bankertoolbench
BankerToolBench
BankerToolBench is a benchmark of 100 end-to-end investment banking tasks for
evaluating AI agents. Each task mirrors real junior-banker work — building
financial models, preparing pitch decks, writing memos — and produces multi-file
deliverables (Excel, PowerPoint, Word) that are scored against expert-authored
rubrics.
The benchmark was developed with 502 investment bankers from firms including
Goldman Sachs, JPMorgan, Evercore, and others. Human completion time… See the full description on the dataset page: https://huggingface.co/datasets/handshake-ai-research/bankertoolbench.Fable-5.1-Max-Reasoning-Filtered-10000x
Dataset Description
This dataset contains 10,000 agentic coding and reasoning multi-turn high-quality traces generated by the new Fable 5.1 model using max reasoning effort.
It holds almost 500,000,000 tokens of step-by-step chain-of-thought programming across multiple complex domains.
It has also been deduplicated and heavily filtered to remove low-quality traces, keeping only high-quality traces.
Dataset Statistics
Metric
Value
Total Examples
10,000… See the full description on the dataset page: https://huggingface.co/datasets/MoreThought/Fable-5.1-Max-Reasoning-Filtered-10000x.natural_reasoningNaturalReasoning is a large-scale dataset for general reasoning tasks. It consists of high-quality challenging reasoning questions backtranslated from pretraining corpora DCLM and FineMath. The questions have been deduplicated and decontaminated from popular reasoning benchmarks including MATH, GPQA, MMLU-Pro, MMLU-STEM. For each question, we extract the reference final answer from the original document from the pretraining corpora if possible. We also provide a model-generated response from… See the full description on the dataset page: https://huggingface.co/datasets/facebook/natural_reasoning.kimi-cyber-reasoning
Kimi Cyber Reasoning
997 chain-of-thought records covering 13 cybersecurity disciplines and 4 systems engineering domains, distilled from the Kimi K3 reasoning model via API. Every record provides an explicit step-by-step <think> reasoning trace followed by a technical resolution, unified code diff fix, or structured tool invocation.
The dataset was curated as an anchor set for training, healing, and specializing compact reasoning models on systems security and tool calling… See the full description on the dataset page: https://huggingface.co/datasets/echel0nn1881/kimi-cyber-reasoning.Light-R1-SFTData
Light-R1: Surpassing R1-Distill from Scratch* with $1000 through Curriculum SFT & DPO
*from models without long COT
technical report
GitHub page
Here are the two-stage SFT data we used to train Light-R1-32B.
Simply refer to stage1-76k.json and stage2-3k.json
Model
Trained From
Release Date
AIME24
AIME25
DeepSeek-R1-Distill-Llama-70B
Llama-3.3-70B-Instruct
25.1.20
70.0
54.1
DeepSeek-R1-Distill-Qwen-32B
Qwen2.5-32B
25.1.20
72.6
54.9
LIMO (32B)
Qwen2.5-32B-Instruct
25.2.4… See the full description on the dataset page: https://huggingface.co/datasets/qihoo360/Light-R1-SFTData.nemotron-post-training-v2-qwen-3.5-9b-regen
Dataset Card for Nemotron Post Training v2 Qwen 3.5 9B Regen
Regenerated responses from nvidia/Nemotron-Post-Training-Dataset-v2 dataset using Qwen3.5 9B model.
Parameter
Value
Max Tokens
4096
Temperature
1.0
Top-k
20
Top-p
0.95
Repetition Penalty
1.5
Dataset consists only the english samples from the Nemotron Post Training Dataset. 85% of the chat prompts have reasoning enabled, every other category has reasoning disabled.
Category
Value
math… See the full description on the dataset page: https://huggingface.co/datasets/Dogacel/nemotron-post-training-v2-qwen-3.5-9b-regen.swe-marathon
SWE Marathon: Ultra Long-Horizon Software Engineering Tasks
20 ultra long-horizon software-engineering tasks designed to challenge frontier coding agents. Each task ships with a containerized environment, a precise instruction, comprehensive tests, and a reference oracle solution. All tasks pass NOP-baseline / Oracle-fix validation.
Homepage: https://github.com/abundant-ai/swe-marathon
License: Apache 2.0
Format: Harbor task format (task.toml + instruction.md + environment/ +… See the full description on the dataset page: https://huggingface.co/datasets/rdesai2/swe-marathon.Nemotron-RL-Agentic-Terminal-Pivot-v1
Dataset Description
The Nemotron-RL-Agentic-Terminal-Pivot-v1 dataset provides training samples for reinforcement learning of command-line ("terminal use") LLM agents with the terminus_judge environment in NeMo Gym.
Each record is a single agent decision point extracted from a successful agent trajectory on a terminal task:
responses_create_params.input — the prompt: the task instruction plus the terminal interaction history (prior agent actions and terminal outputs) up to the… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Terminal-Pivot-v1.rejected-nepali-law-v2
Nepali Source-Grounded Instruction Dataset — REJECTED
Synthetic Nepali instruction-tuning data generated with NVIDIA NeMo Data
Designer from authoritative Nepali documents (agriculture manuals, legal
texts). Answers are grounded strictly in the source; unanswerable questions
get an explicit refusal. Records use chat messages format plus metadata
and per-record quality_scores (grounding / correctness / naturalness, 1-5,
LLM-as-judge). One data/train-<shard>.jsonl per source… See the full description on the dataset page: https://huggingface.co/datasets/aarajbhattarai/rejected-nepali-law-v2.Superior-Reasoning-SFT-gpt-oss-120b
Superior-Reasoning-SFT-gpt-oss-120b
📣 News
Our dataset ranked #1 on the Hugging Face Datasets Trending leaderboard from January 20 to January 30.
🚀 Overview
The Superior-Reasoning-SFT-gpt-oss-120b dataset is a high-quality, open-source collection containing 435K samples designed to democratize the training of high-performance Long Chain-of-Thought (Long-CoT) models. Unlike standard distilled datasets that rely on random sampling or… See the full description on the dataset page: https://huggingface.co/datasets/Alibaba-Apsara/Superior-Reasoning-SFT-gpt-oss-120b.Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1
Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1
Dataset Description:
Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1 is an RL dataset for training and evaluating a tool-using agent's ability to resist Indirect Prompt Injection (IPI) attacks hidden inside tool-returned environment data. In each record, the agent receives a benign user request that requires calling a read tool whose output contains an adversarial instruction disguised as legitimate domain content… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1.Nemotron-RL-Ultra-Training-Blends
Dataset Description:
This dataset provides Reinforcement Learning (RL) and Multi-teacher On-Policy Distillation (MOPD) training-data blends used by the public Nemotron-3-Ultra post-training recipe. The blends are consumed by the NeMo RL training recipes through the NeMo Gym agent framework, in which each prompt is paired with an agent/environment that returns a verifiable or judge-based reward. Each subset is a separate blend; see the recipe for how the blends are used.
The… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Ultra-Training-Blends.refute
Can AI read new science honestly?
Models can sound convincing while misreading a result or expressing more confidence than the evidence deserves. That matters when people use them to summarize papers, compare studies, or decide what to investigate next.
REFUTE tests whether a model knows the finding, spots quiet flaws, names what would overturn a claim, and matches its confidence to the evidence.
Truth Score is the main result. It combines factual accuracy, flaw… See the full description on the dataset page: https://huggingface.co/datasets/BGPT-OFFICIAL/refute.Cybersecurity_Reasoning_Dataset
Cybersecurity Reasoning Dataset (Model-Agnostic)
A model-agnostic re-architecture of the Cybersecurity Reasoning Dataset. The original
corpus was format-bound to the Mistral/Llama ### Instruction: / ### Response: template;
this dataset losslessly separates reasoning content from format, providing one
neutral canonical corpus plus four per-family rendered training variants
(Mistral/Llama, DeepSeek, ChatML, Gemma).
Why this exists. Identical content scored 88.1 on a… See the full description on the dataset page: https://huggingface.co/datasets/dpevzner/Cybersecurity_Reasoning_Dataset.sat-vl-sft-training-ready-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-training-ready-v1.researchscope-papers
ResearchScope Papers
Open CS research paper dataset maintained by ResearchScope.
Updated automatically via GitHub Actions.
Quick start
from datasets import load_dataset
ds = load_dataset("kishormorol/researchscope-papers", "papers", split="train")
print(ds[0])
See Usage below for per-source splits, instruction-tuning, and the per-section fine-tuning data.
Stats
34,946 papers (raw metadata) — 9,946 arXiv · 20,000 conference · 5,000 journal
174,282… See the full description on the dataset page: https://huggingface.co/datasets/kishormorol/researchscope-papers.DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/ronaldcmz/DeepSeek-v4-Pro-Agent.