datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
stack-v3-train
🥞 The Stack v3
What is it?
What is being released
How to download and use it
Dataset statistics
Dataset structure
Dataset creation
Considerations for using the data
Additional information
What is it?
The Stack v3 is the largest, most up-to-date open dataset of source code, crawled directly from GitHub and built to pre-train code LLMs with full-repository context. It is the successor to The Stack v2 and, like its predecessor, is released to make the training… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceCode/stack-v3-train.TxT360-v2
TxT360-v2
Dataset Description
Pre-training sources for the K2 Horizon training data release. This repository is part of the K2 Horizon collection.
The repository is organized into multiple subsets. Every subset has a train split backed by Parquet shards.
K2 Horizon Dataset Series
Dataset repository
Focus
Subsets
IFM/TxT360-v2
Web and question-answering text
3
IFM/Code-Reasoning
Code reasoning and task synthesis
7
IFM/Math-Reasoning… See the full description on the dataset page: https://huggingface.co/datasets/IFM/TxT360-v2.Mega-Brain-Distill
Mega-Brain-Distill
Curated merge of the top 10% highest-scoring examples from
584 community-uploaded LLM distillation/reasoning-trace datasets
on the Hub (Fable-5, Opus, GLM, Kimi, DeepSeek, GPT, MiniMax, Qwen traces,
etc.), deduplicated within and across all of them — many of these source
repos are the same underlying dump re-uploaded by different users.
Auto-generated by run.py — do not hand-edit, it will be overwritten on
the next run. Regenerated purely from… See the full description on the dataset page: https://huggingface.co/datasets/ShinMK3/Mega-Brain-Distill.kimi-k3-distillation
kimi-k3-distillation
Single-teacher slice of
r0b0tlab/qwen3.8-max-glm5.2-kimi-k3-distillation,
filtered to teacher_model == "kimi-code/k3" only. The Qwen3.8-Max-Preview and
GLM-5.2 traces are removed.
4,347 rows — 3,918 train / 212 validation / 217 test.
from datasets import load_dataset
ds = load_dataset("beyoru/kimi-k3-distillation") # sft: messages + tools
ds = load_dataset("beyoru/kimi-k3-distillation", "canonical") # + full audit columns… See the full description on the dataset page: https://huggingface.co/datasets/beyoru/kimi-k3-distillation.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-glm5.2-kimi-k3-distillation.Qwen3.8-27B-Distillation-40K
Qwen3.8-27B-Distillation (40K Traces)
Qwen3.8-27B-Distillation is a dataset containing 40,000 reasoning traces distilled from Qwen's latest model — Qwen3.8-27B. We generated this dataset locally by running the model on our own infrastructure. It covers 4 domains with prompts sourced from 12 diverse open-source datasets.
Dataset Overview
Metric
Value
Total Examples
40,000
Teacher Model
Qwen3.8-27B
Model Precision
FP8
Reasoning Effort
medium… See the full description on the dataset page: https://huggingface.co/datasets/faunix/Qwen3.8-27B-Distillation-40K.stack-v3-train
🥞 The Stack v3
What is it?
What is being released
How to download and use it
Dataset statistics
Dataset structure
Dataset creation
Considerations for using the data
Additional information
What is it?
The Stack v3 is the largest, most up-to-date open dataset of source code, crawled directly from GitHub and built to pre-train code LLMs with full-repository context. It is the successor to The Stack v2 and, like its predecessor, is released to make the training… See the full description on the dataset page: https://huggingface.co/datasets/AlanaSky/stack-v3-train.PersonaMem-v3
PersonaMem-v3: Toward Omni-Platform Personal Intelligence for Holistic User Understanding, Recommendation, and Agentic Tasks
Bowen Jiang, Yuan Yuan, Zhuoqun Hao, Yuchen Liu, Maohao Shen, Sihao Chen, Gregory Wornell,
Chris Callison-Burch, Lyle Ungar, Dan Roth, Qi Guo, Xiangjun Fan, Camillo J. Taylor, Hanchao Yu
A collaboration between:
Meta Recommendation Systems
University of Pennsylvania
MIT
Third release in the PersonaMem series:
PersonaMem-v1: [COLM… See the full description on the dataset page: https://huggingface.co/datasets/bowen-upenn/PersonaMem-v3.aopsstack-v3-train
🥞 The Stack v3
What is it?
What is being released
How to download and use it
Dataset statistics
Dataset structure
Dataset creation
Considerations for using the data
Additional information
What is it?
The Stack v3 is the largest, most up-to-date open dataset of source code, crawled directly from GitHub and built to pre-train code LLMs with full-repository context. It is the successor to The Stack v2 and, like its predecessor, is released to make the training… See the full description on the dataset page: https://huggingface.co/datasets/CathleenTico/stack-v3-train.ifm-cleaned-pretrain-30B
IFM Cleaned Pretrain — 30B target
credits to https://huggingface.co/datasets/IFM/Pretrain-Behaviors
Status: complete.
Published: 6,155,901 documents; 30,000,015,784 source-annotated tokens.
Target: 30,000,000,000 source-annotated tokens, approximately equal across all seven categories.
This repository contains text only in Parquet: earlier shards were format-cleaned; subsequent shards contain source text without the cleaner. There are no token-ID arrays or binary token shards.… See the full description on the dataset page: https://huggingface.co/datasets/domofon/ifm-cleaned-pretrain-30B.stack-v3-train
🥞 The Stack v3
What is it?
What is being released
How to download and use it
Dataset statistics
Dataset structure
Dataset creation
Considerations for using the data
Additional information
What is it?
The Stack v3 is the largest, most up-to-date open dataset of source code, crawled directly from GitHub and built to pre-train code LLMs with full-repository context. It is the successor to The Stack v2 and, like its predecessor, is released to make the training… See the full description on the dataset page: https://huggingface.co/datasets/mick260449/stack-v3-train.bagaco3
Bagaço3 🍷🇵🇹
Bagaço3 is the third version of Bagaço, the largest pretraining dataset for European Portuguese. It follows Bagaço2 and adds documents from FinePDFs and FineWiki.
Bagaço collects European Portuguese documents from upstream sources and adds an educational score and content category to each document. See Classification for details.
Methodology
Collect documents from Bagaço2, FinePDFs, and FineWiki.
Filter new FinePDFs and FineWiki documents with the… See the full description on the dataset page: https://huggingface.co/datasets/duarteocarmo/bagaco3.Magpie-Llama-3.1-Pro-300K-Filtered
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Magpie-Llama-3.1-Pro-300K-Filtered.kimi-k3-coding-and-debugging-traces
Kimi K3 Coding, Tool Use & Instruction Following Traces
582 TRAJECTORIES · 3,956 TRAINING ROWS · 3 MB PARQUET · 72 MB JSONL
Generated by moonshiner — an open harness for
distilling verified instruction-following, tool-use, and agentic coding traces.
Behavior-preserving instruction-following, tool-use, and agent trajectories
from Kimi K3 (moonshotai/kimi-k3). The category and row-share tables
below describe the actual mix seen during training rather than assuming a… See the full description on the dataset page: https://huggingface.co/datasets/greghavens/kimi-k3-coding-and-debugging-traces.sp500_earnings_transcripts
S&P 500 Earnings Transcripts Dataset
This comprehensive dataset contains earnings call transcripts for S&P 500 companies and US large-caps, spanning from 2005 to 2025. Earnings calls provide valuable insights into company performance, strategic initiatives, and management perspectives that are essential for financial analysis, natural language processing research, and market sentiment studies.
Dataset Description
This collection includes:
Complete transcripts: Full… See the full description on the dataset page: https://huggingface.co/datasets/Bose345/sp500_earnings_transcripts.Magpie-Llama-3.1-Pro-MT-300K-Filtered
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
Abstract
Click Here
High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent… See the full description on the dataset page: https://huggingface.co/datasets/Magpie-Align/Magpie-Llama-3.1-Pro-MT-300K-Filtered.PKU-SafeRLHF-30K
Dataset Card for PKU-SafeRLHF
Warning: this dataset contains data that may be offensive or harmful. The data are intended for research purposes, especially research that can make models less harmful. The views expressed in the data do not reflect the views of PKU-Alignment Team or any of its members.
Dataset Summary
The preference dataset consists of 30k+ expert comparison data. Each entry in this dataset includes two responses to a question, along with safety… See the full description on the dataset page: https://huggingface.co/datasets/PKU-Alignment/PKU-SafeRLHF-30K.stack-v3-devops
The Stack v3 DevOps Corpus
13,234,862 complete infrastructure units extracted from
The Stack v3,
grouped into seven classes and gated on content rather than popularity.
A unit is not a file, it is the thing an engineer would actually run: a Helm chart
arrives with its Chart.yaml, values.yaml and every template; a Terraform module
with all of its .tf files; an Ansible role with its tasks, defaults and handlers.
That is only possible because The Stack v3 groups rows by repository… See the full description on the dataset page: https://huggingface.co/datasets/Helmcode/stack-v3-devops.GridCorpus_9M_Sudoku_Puzzles_Enriched
╔══════════════════════════════════════════════════════════════════════╗
║ ║
║ G R I D C O R P U S ║
║ ║
║ "004300209005009001070060043..." ║
║ │ ║
║ ▼… See the full description on the dataset page: https://huggingface.co/datasets/beta3/GridCorpus_9M_Sudoku_Puzzles_Enriched.jupyter-scripts-smollm3
The Stack v2 Jupyter Notebooks as Scripts
This dataset contains script representations of the Jupyter notebooks in
The Stack v2. It was
created from the materialized Jupyter_Notebook split in
jordangong/the-stack-v2-smollm3.
The output schema follows the Jupyter-script schema used by
bigcode/starcoderdata,
but this release is not deduplicated, PII-filtered, or otherwise equivalent
to StarCoderData's filtered split.
Relationship to the SmolLM3 training mix
This… See the full description on the dataset page: https://huggingface.co/datasets/jordangong/jupyter-scripts-smollm3.stack-v3-train
🥞 The Stack v3
What is it?
What is being released
How to download and use it
Dataset statistics
Dataset structure
Dataset creation
Considerations for using the data
Additional information
What is it?
The Stack v3 is the largest, most up-to-date open dataset of source code, crawled directly from GitHub and built to pre-train code LLMs with full-repository context. It is the successor to The Stack v2 and, like its predecessor, is released to make the training… See the full description on the dataset page: https://huggingface.co/datasets/happahhap2026/stack-v3-train.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/o0Biggz0o/qwen3.8-max-glm5.2-kimi-k3-distillation.AO3-2020
⚠️ adult_18plus_100K — SAFETY / ALIGNMENT SUBSET
❌ DO NOT train a generative model on this subset as ordinary SFT data.
✅ Use it inverted — as the rejected side of preference pairs.
This subset contains the first ~100,000 tokens of content rated
Explicit or Mature on AO3 (26 chapters).
Technique
How to use this subset
DPO / RLHF
Label completions as rejected; safe rewrites as chosen.
Content-safety classifier
Negative class in… See the full description on the dataset page: https://huggingface.co/datasets/ray0rf1re/AO3-2020.lilm1-pretrain-mix-32b
LiLM Experiment 3 pretraining corpus
Private research corpus with 32,000,010,072 globally
exact-deduplicated train tokens plus 328,933,246 held-out
tokens. Data are stored as EOS-delimited little-endian uint16 binaries with
aligned Parquet provenance.
This repository combines ODC-By FinePDFs-Edu, CC-BY-4.0 DCLM, ODC-By
SmolLM/FineMath sources, Apache-2.0 UltraX subject to its upstream terms,
per-file permissively licensed Stack-Edu code subject to The Stack v2 terms,
StarCoder2… See the full description on the dataset page: https://huggingface.co/datasets/glouriousgautam/lilm1-pretrain-mix-32b.b3-agent-security-benchmark-weak[paper] [blogpost] [game]
b3 AI Security Benchmark: Breaking Agent Backbones
Highly contextalized prompt injections crowd-sourced during the Gandalf Agent Breaker Challenge.
This is a low-quality version of the data behind Breaking Agent Backbones: Evaluating the Security
of Backbone LLMs in AI Agents.
The high quality dataset was used to evaluate the security of more than 30 LLMs.
Dataset Summary
Purpose: This dataset contains crowdsourced adversarial attacks… See the full description on the dataset page: https://huggingface.co/datasets/Lakera/b3-agent-security-benchmark-weak.trajectory_data_dream_32
d3LLM Trajectory Dataset
Project Page | Paper | GitHub | Blog
This repository contains the pseudo-trajectory distillation data used for training d3LLM (pseuDo-Distilled Diffusion Large Language Model), as introduced in the paper "d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillation".
Introduction
d3LLM is a framework designed to strike a balance between accuracy and parallelism in diffusion-based large language models (dLLMs). This dataset consists of… See the full description on the dataset page: https://huggingface.co/datasets/d3LLM/trajectory_data_dream_32.qwen3.8-max-glm5.2-kimi-k3-distill
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/ansulev/qwen3.8-max-glm5.2-kimi-k3-distill.qwen3.8-max-glm5.2-kimi-k3-distillation
Multi-Teacher Distillation Dataset (57,937 traces)
A quality-filtered, deduplicated, multi-teacher SFT corpus combining traces from three frontier models across math, code, reasoning, instruction-following, tool-use, science, long-context, multilingual, and creative dialogue domains.
Teachers
Teacher
Provider
Traces
Qwen3.8-Max-Preview
Alibaba Cloud Model Studio
48,283
GLM-5.2
Z.AI Coding Plan
5,307
Kimi Code K3
Moonshot AI (Kimi)
4,347… See the full description on the dataset page: https://huggingface.co/datasets/p-research/qwen3.8-max-glm5.2-kimi-k3-distillation.qwen3.8-max-distillation-50k
Qwen3.8-Max Distillation 50K
A curated dataset of 49,772 teacher-generated traces from qwen3.8-max-preview, prepared for supervised fine-tuning and off-policy knowledge distillation.
The teacher responses are preserved as returned by the API. Where the model emitted visible <think>...</think> blocks, those blocks remain in the assistant message. Some simpler prompts received direct answers without a thinking block.
[!CAUTION]
Terms and provenance notice — not cleared for… See the full description on the dataset page: https://huggingface.co/datasets/r0b0tlab/qwen3.8-max-distillation-50k.