datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Nemotron-SFT-Science-v2
Dataset Description:
Nemotron-Science-v2 is a science reasoning dataset with synthetic (synthetic MCQ, RQA) and non-synthetic vendor problems and LLM-generated solutions. It comprises three domains (Physics, Biology, and Chemistry), two question formats (multiple-choice questions [MCQ] and open questions [OpenQ]), and three generation setups: chain-of-thought (CoT) reasoning without tools, Python tool usage, and search tools usage with the Tavily API.
The solutions were… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-SFT-Science-v2.scientific-papers
Scientific Papers - Raw Full Text
~57 million scientific papers with full text, extracted from multiple large-scale academic paper collections. This dataset provides raw full text suitable for pre-training, fine-tuning, or building search indices over scientific literature.
Subsets
Subset
Papers
Size
Source
papers-2
~18.5M
~358 GB
S2ORC papers collection (untitled subset)
papers-3
~27.4M
~198 GB
S2ORC scientific-papers collection
pes2o
~8.2M
~106 GB… See the full description on the dataset page: https://huggingface.co/datasets/scientifi-papers/scientific-papers.Sci-Fi-ZH一份 VeejaLiu 正在手工清洗的数据：https://github.com/VeejaLiu/ScienceFictionCollection
SciCode-Domain-Code
DATA1: Domain-Specific Code Dataset
Dataset Overview
DATA1 is a large-scale domain-specific code dataset focusing on code samples from interdisciplinary fields such as biology, chemistry, materials science, and related areas. The dataset is collected and organized from GitHub repositories, covering 178 different domain topics with over 1.1 billion lines of code.
Dataset Statistics
Total Datasets: 178 CSV files
Total Data Size: ~115 GB
Total Lines of Code: Over… See the full description on the dataset page: https://huggingface.co/datasets/SciCodePile/SciCode-Domain-Code.SciCode-Verified
SciCode-Verified
SciCode-Verified is the corrected, human-verified release of the
SciCode scientific-code-generation benchmark.
A problem-by-problem audit identified 263 defects in the 65-problem SciCode test split and
corrected every confirmable defect. The released evaluation set contains 64 main problems and
287 scored subproblems; one original problem is excluded because its specification does not
determine a unique, verifiable answer.
Paper: SciCode-Verified: How Benchmark… See the full description on the dataset page: https://huggingface.co/datasets/shhu2001/SciCode-Verified.AgentSearch-V1
Getting Started
The AgentSearch-V1 dataset boasts a comprehensive collection of over one billion embeddings, produced using jina-v2-base. The dataset encompasses more than 50 million high-quality documents and over 1 billion passages, covering a vast range of content from sources such as Arxiv, Wikipedia, Project Gutenberg, and includes carefully filtered Creative Commons (CC) data. Our team is dedicated to continuously expanding and enhancing this corpus to improve the search… See the full description on the dataset page: https://huggingface.co/datasets/SciPhi/AgentSearch-V1.Scientific_Research_Tokenized
NexaSci Scientific Research Tokenized
This dataset repository now holds the active NexaSci scientific pretraining reservoir, the NexaMat controller fine-tuning pack, and archived legacy reservoir builds. The current production reservoir is the 10B-token Apache Arrow release under nexasci_reservoir_v3_10b_prod_rust/.
Current Status
The active large-scale training artifact is:
nexasci_reservoir_v3_10b_prod_rust/
It was produced from the NexaSci 10B data-engineering campaign… See the full description on the dataset page: https://huggingface.co/datasets/AethronPhantom/Scientific_Research_Tokenized.scientific-systems-sft-dpo-distilled
Scientific and Systems Distillation Golden Dataset (SFT & DPO)
Bu veri seti, temel ve uygulamalı bilimler ile yüksek başarımlı hesaplama (HPC) ve Linux sistem mühendisliği alanlarında yapay zeka modellerini ince ayar (fine-tuning) ve tercih hizalama (preference alignment) süreçlerine tabi tutmak amacıyla tasarlanmış, 2.551 adet ileri düzey teknik prompt ve bunlara karşılık gelen yüksek kaliteli model çıktılarından derlenmiş zengin bir sentetik veri kümesidir.
🚀 Veri… See the full description on the dataset page: https://huggingface.co/datasets/onkanat/scientific-systems-sft-dpo-distilled.hle_material_science
HLE Material Science: A Specialized Benchmark for Materials Science
A Materials Science Subset of Humanity's Last Exam (HLE)
Overview
HLE Material Science is a carefully curated materials science subset derived from the Humanity's Last Exam (HLE) dataset, containing 106 high-quality expert-level questions covering 25+ materials science subfields, with 97% of questions rated as high confidence.
This dataset is designed to evaluate large language models'… See the full description on the dataset page: https://huggingface.co/datasets/TalentZHOU/hle_material_science.natural-science-reasoning
Natural Sciences Reasoning: the "smolest" reasoning dataset
A smol-scale open dataset for reasoning tasks using Hugging Face Inference Endpoints. While intentionally limited in scale, this resource prioritizes:
Reproducible pipeline for reasoning tasks using a variety of models (Deepseek V3, Deepsek-R1, Llama70B-Instruct, etc.)
Knowledge sharing for domains other than Math and Code reasoning
In this repo, you can find:
The prompts and the pipeline (see the config file).
The… See the full description on the dataset page: https://huggingface.co/datasets/dvilasuero/natural-science-reasoning.islamic-sciences
islamlab — The Islamic Sciences Corpus
The Islamic sciences other than Qur'an and hadith, as their authors wrote
them: 4,022 works by scholars who died between the
0st and the 14th Hijri century, cut along their own chapter
and biographical-entry boundaries into 1,864,389 units
(3.41 billion characters of Arabic), each carrying the volume and
page it sits on so a quotation can be cited rather than merely produced.
Scope is Ahl al-Sunnah wa'l-Jamāʿah, and the gate is the author… See the full description on the dataset page: https://huggingface.co/datasets/islamlab/islamic-sciences.Sci-Fi-Books-gutenberg
Gutenberg Sci-Fi Book Dataset
This dataset contains information about science fiction books. It’s designed for training AI models, research, or any other purpose related to natural language processing.
Data Format
The dataset is provided in CSV format. Each record represents a book and includes the following fields:
ID: A unique identifier for the book.
Title: The title of the book.
Author: The author(s) of the book.
Text: The text content of the book (e.g., summary… See the full description on the dataset page: https://huggingface.co/datasets/stevez80/Sci-Fi-Books-gutenberg.Nemotron-RL-Science-v1
Dataset Description:
Nemotron-RL-Science-v1 is a reinforcement learning (RL) dataset for science reasoning. Each example provides a problem, a reference answer, and a verifiable RL environment configuration (the agent prompt, the agent/verifier reference, and the answer-extraction template) so that a policy model can be trained with verifiable rewards. It covers three domains (Physics, Biology, and Chemistry), the open-question (OpenQ) format, and two generation setups:… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Science-v1.SciCode-Domain-Code
DATA1: Domain-Specific Code Dataset
Dataset Overview
DATA1 is a large-scale domain-specific code dataset focusing on code samples from interdisciplinary fields such as biology, chemistry, materials science, and related areas. The dataset is collected and organized from GitHub repositories, covering 178 different domain topics with over 1.1 billion lines of code.
Dataset Statistics
Total Datasets: 178 CSV files
Total Data Size: ~115 GB
Total Lines of Code: Over… See the full description on the dataset page: https://huggingface.co/datasets/liuhangbiao/SciCode-Domain-Code.llama-nemotron-science-reasoning-on-canonical-think-full
Llama-Nemotron science reasoning — Delphi canonical-think (COMPLETE, no length filter)
The complete reasoning:on science split of
nvidia/Llama-Nemotron-Post-Training-Dataset, converted once into the canonical
Delphi chat-template thinking format. 708,920 rows.
Unlike the cold-start warmup slice
open-athena/llama-nemotron-science-reasoning-on-le3000tok-100k
(and its -canonical-think variant), this build applies no length cap and no subsample — every
long-CoT science example is… See the full description on the dataset page: https://huggingface.co/datasets/laion/llama-nemotron-science-reasoning-on-canonical-think-full.SciCode-Programming-Problems
DATA3: Programming Problems Generation Dataset
Dataset Overview
DATA3 is a large-scale programming problems generation dataset that contains AI-generated programming problems inspired by real scientific computing code snippets. The dataset consists of 22,532 programming problems, each paired with a comprehensive solution. These problems focus on scientific computing concepts such as numerical algorithms, data analysis, mathematical modeling, and computational methods in… See the full description on the dataset page: https://huggingface.co/datasets/SciCodePile/SciCode-Programming-Problems.openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16
OpenThoughts-4 Science SDG: Qwen3-30B-A3B-Thinking-2507 (n=8, top-16 logprobs)
Synthetic generations from
Qwen/Qwen3-30B-A3B-Thinking-2507
on the Marin OpenThoughts-4 science SDG prompt
set.
Each prompt is sampled n=8 times, and for every generated token the dataset
stores the chosen-token log probability plus the top-16 log probabilities
over the vocabulary, enabling distillation, KL-style fine-tuning,
reranking, and uncertainty analysis.
Generation setup
Field… See the full description on the dataset page: https://huggingface.co/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-30b-a3B-thinking-2507-n8-flattened-logprobs-k16.Science-QnA
Science-QnA
The Science-QnA is a large-scale, high-quality science-focused dataset (~5.63M rows) curated using synthetic data generation through distillation techniques and select open-source resources. Designed to train and evaluate reasoning-capable models in science domains with emphasis on conceptual understanding, numerical problem-solving, and exam-style Q&A patterns across Physics, Chemistry, Biology, and Mathematics.
Summary
• Domain: Science, Physics… See the full description on the dataset page: https://huggingface.co/datasets/169Pi/Science-QnA.Dr.SCI
Dr. SCI Dataset (Reproduced)
[📜 Original Paper] •
[🤗 Reproduced Dataset] •
[💻 Reproduced Github]
Disclaimer: This is an unofficial reproduction of the Dr. SCI dataset introduced in"Improving Data and Reward Design for Scientific Reasoning in Large Language Models" [arXiv].A detailed implementation of the curation process is available in my GitHub Repo.This work is not affiliated with or endorsed by the original authors. Please refer to the original paper for… See the full description on the dataset page: https://huggingface.co/datasets/MiniByte-666/Dr.SCI.hle_material_science
HLE Material Science: A Specialized Benchmark for Materials Science
A Materials Science Subset of Humanity's Last Exam (HLE)
Overview
HLE Material Science is a carefully curated materials science subset derived from the Humanity's Last Exam (HLE) dataset, containing 106 high-quality expert-level questions covering 25+ materials science subfields, with 97% of questions rated as high confidence.
This dataset is designed to evaluate large language models'… See the full description on the dataset page: https://huggingface.co/datasets/stonelight/hle_material_science.SciCode-Runnable-Benchmark-Reviewedopenthoughts4-science-26041-prompts-qwen3-32b-n8-flattened-logprobs-k16
OpenThoughts-4 Science SDG: Qwen3-32B (n=8, top-16 logprobs)
Synthetic generations from
Qwen/Qwen3-32B
on the Marin OpenThoughts-4 science SDG prompt
set.
Each prompt is sampled n=8 times, and for every generated token the dataset
stores the chosen-token log probability plus the top-16 log probabilities
over the vocabulary, enabling distillation, KL-style fine-tuning,
reranking, and uncertainty analysis.
Generation setup
Field
Value
Generator model… See the full description on the dataset page: https://huggingface.co/datasets/marin-community/openthoughts4-science-26041-prompts-qwen3-32b-n8-flattened-logprobs-k16.10001-Science-Facts
10,001 Science Facts
10,000+ obscure, surprising, and verifiable science facts
The kind that make you go "wait, really?"
🔗 GitHub Repository •
📁 Download by Category
🤔 What is this?
A curated dataset of 10,003 science facts across 32 categories — from quantum physics to parasites to the history of food.
Every fact is:
Sourced — from Wikipedia, Wikidata, academic sources
Verifiable — no LLM hallucinations
Surprising — passes the "dinner party test"… See the full description on the dataset page: https://huggingface.co/datasets/Royal-lobster/10001-Science-Facts.stem-scientific-code-sample
AxiomSet Labs STEM Scientific-Code Sample
A 30-task sample of STEM reasoning and scientific-code problems across five domains.
Domains
Biology: 6 tasks
Chemistry: 6 tasks
Materials Science: 6 tasks
Mathematics: 6 tasks
Physics: 6 tasks
Each task contains two subproblems and one main problem, including prompts, scientific background, testing templates, and reference solutions.
Files
data/sample.jsonl — one task per line; used by the Hugging Face… See the full description on the dataset page: https://huggingface.co/datasets/AxiomSetLabs/stem-scientific-code-sample.science-qa-sft-100k
Science QA SFT (100K)
100,000 science Q&A examples with step-by-step explanations for SFT fine-tuning. Covers physics, chemistry, biology, astronomy, and earth science at beginner through advanced difficulty.
Motivation
Models trained on general text often give superficially plausible but mechanistically wrong answers to science questions — stating the right conclusion without understanding the underlying reasoning. This dataset trains models to explain why an… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/science-qa-sft-100k.control-sci-corpus
ControlSci Corpus
Control science structured corpus with two configs: Sci-Align benchmark (500 questions) and Sciverse SFT instruction pairs (924 ChatML entries).
License: CC-BY-4.0
Project: MorningStar0709/ControlMind
Configs
benchmark — Sci-Align Benchmark (500 questions)
4-dimension control science evaluation benchmark generated from the ControlSci structured corpus.
Split: core (500 questions)
Load:
from datasets import load_dataset
ds =… See the full description on the dataset page: https://huggingface.co/datasets/MorningStar0709/control-sci-corpus.SciPredict
SciPredict: Can LLMs Predict the Outcomes of Research Experiments?
Paper: SciPredict: Can LLMs Predict the Outcomes of Research Experiments in Natural Sciences?
Overview
SciPredict is a benchmark evaluating whether AI systems can predict experimental outcomes in physics, biology, and chemistry. The dataset comprises 405 questions derived from recently published empirical studies (post-March 2025), spanning 33 subdomains.
Dataset Structure
Total Questions: 405… See the full description on the dataset page: https://huggingface.co/datasets/ScaleAI/SciPredict.video-scissors-sessions
Coding agent session traces for kaofelix/video-scissors-sessions
This dataset contains redacted coding agent session traces collected while working on git@github.com:kaofelix/video-scissors.git. The traces were exported with pi-share-hf from a local pi workspace and filtered to keep only sessions that passed deterministic redaction and LLM review.
Data description
Each *.jsonl file is a redacted pi session. Sessions are stored as JSON Lines files where each line… See the full description on the dataset page: https://huggingface.co/datasets/kaofelix/video-scissors-sessions.scipar_parallel_docs
SciPar Parallel Documents
Dataset Description
This dataset contains parallel documents (i.e., titles & abstracts) extracted from academic theses, dissertations, and other scientific texts.
In the original paper, we've extracted 9.17M sentence pairs in 31 language pairs from 86 repositories.
This version has been created through further processing and filtering to extract parallel documents instead of parallel sentences.
To do this, we kept only the parallel titles and… See the full description on the dataset page: https://huggingface.co/datasets/ilsp/scipar_parallel_docs.Multilingual-Normalizer
Multilingual TTS text normalizer (written → spoken)
Training pairs for fine-tuning a small LLM as a text-to-speech normalizer: text is a sentence
the way people type it (digits, currency symbols, dates, phone numbers, …) and normalized is the
exact spoken form, in the same language, with nothing left that a TTS model cannot say.
52,698 rows — 16 monolingual locales and 6 Malaysian code-switched pairs. Every row is
digit-free on the spoken side.
from datasets import load_dataset… See the full description on the dataset page: https://huggingface.co/datasets/Scicom-intl/Multilingual-Normalizer.