datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Nemotron-Competitive-Programming-v1
Dataset Description:
Nemotron-Competitive-Programming-v1 is a large-scale synthetic coding and reasoning dataset designed to push LLM performance on challenging programming and systems tasks. It combines Python and C++ samples across unique competitive programming questions.
Beyond problem solving, the dataset includes InfiniByte, a cross-domain subset with problems derived from scientific fields.
This dataset is ready for commercial use.
Competitive Coding
The… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Competitive-Programming-v1.Nemotron-SFT-Competitive-Programming-v2
Dataset Description:
Nemotron-Competitive-Programming-v2 is a large-scale synthetic coding and reasoning dataset designed to push LLM performance on challenging programming and systems tasks. It combines Python and C++ samples across unique competitive programming questions.In addition, the dataset includes samples targetting a subset of programming language exercises from Exercism.
Beyond problem solving, the dataset includes Text-to-SQL, which aims at training LLMs to reason and… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-SFT-Competitive-Programming-v2.latent-programming-horizons-trajs
latent-programming-horizons-trajs
Agent trajectories and per-edit correctness labels from the
program-probes project, which
measures whether a language model's internal hidden states linearly predict
properties of its own agentic output (e.g. "does the code currently compile?")
before those properties are realised.
Each trajectory is a run of a coding agent (mini-SWE-agent) attempting a
SWE-bench (Verified or Pro) instance. This dataset contains the raw
transcripts and labels… See the full description on the dataset page: https://huggingface.co/datasets/ASSERT-KTH/latent-programming-horizons-trajs.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.Competitive-Programming-python-blend
Dataset Card for Competitive-Programming-python-blend
Summary
Competitive-Programming-python-blend is a mixed supervised fine-tuning dataset centered on competitive programming, code reasoning, and instruction-style problem solving. The blend is Python-first, but it also keeps a small amount of C++, agentless SWE, and reasoning-oriented chat supervision to broaden training coverage.
The current release is published as a single HF-friendly JSONL file, clean.jsonl.… See the full description on the dataset page: https://huggingface.co/datasets/Jackrong/Competitive-Programming-python-blend.Competitive-Programmingprogramming_books_engspatial-programmingColumns:
input_text — text prompt (may be null)
input_image — reference image (may be null); at least one of text / image is set
code — Python source that builds the asset
glb — the resulting GLB file, raw bytes
platform — e.g. blender
type — e.g. modeling
model — which model wrote the code (fable, fable 5.1, opus 5, astra, sol)
Rows are appended one parquet shard per upload under data/.
textbook_quality_programming
Dataset Card for "textbook_quality_programming"
Synthetic programming textbooks generated with GPT-3.5 and retrieval. Very high quality, aimed at being used in a phi replication. Currently 115M tokens. Covers many languages and technologies, with a bias towards python.
~10k of the books (65M tokens) use an older generation method, and average 6k tokens in length. ~1.5k books (50M tokens) use a newer generation method, with a more detailed outline, and average 33k tokens in… See the full description on the dataset page: https://huggingface.co/datasets/vikp/textbook_quality_programming.IndustryCorpus2_computer_programming_code
IndustryCorpus2: Programming
This repository contains the IndustryCorpus2: Programming domain subset of BAAI/IndustryCorpus2.
Refer to the parent dataset card for data construction, intended use, limitations,
and licensing details.
Citation
If you use this dataset in your work, please cite IndustryCorpus2:
@misc{shi2024industrycorpus2,
title = {IndustryCorpus2},
author = {Xiaofeng Shi and Lulu Zhao and Hua Zhou and Donglin Hao},
year = {2024}… See the full description on the dataset page: https://huggingface.co/datasets/BAAI/IndustryCorpus2_computer_programming_code.PersonaSignal-PersonalizedResponse-Programming-Expertise-gpt-5-mini
Dataset card for PersonaSignal-PersonalizedResponse-Programming-Expertise-gpt-5-mini
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "programming_expertise",
"dimension_values": [
"Novice",
"Intermediate",
"Advanced"
],
"dimension_description": "Represents the user's practical fluency in software engineering. It shapes how they decompose problems, choose abstractions, weigh… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PersonalizedResponse-Programming-Expertise-gpt-5-mini.PersonaSignal-PerceivabilityTest-Programming-Expertise-gpt-5-mini
Dataset card for PersonaSignal-PerceivabilityTest-Programming-Expertise-gpt-5-mini
This dataset was made with Curator.
Dataset details
A sample from the dataset:
{
"dimension_name": "programming_expertise",
"dimension_values": [
"Novice",
"Intermediate",
"Advanced"
],
"dimension_description": "Represents the user's practical fluency in software engineering. It shapes how they decompose problems, choose abstractions, weigh… See the full description on the dataset page: https://huggingface.co/datasets/JasonYan777/PersonaSignal-PerceivabilityTest-Programming-Expertise-gpt-5-mini.Puma-Programming-Language-Dataset
license: mit
Puma Programming Language Dataset
The Puma Programming Language Dataset is a curated collection of Puma programming examples designed for developers, learners, educators, researchers, and AI systems working with the Puma programming language.
The dataset contains practical Puma code examples covering language syntax, programming patterns, data structures, functions, iteration, backend development, HTTP services, APIs, WebSocket communication… See the full description on the dataset page: https://huggingface.co/datasets/dominexmacedon/Puma-Programming-Language-Dataset.programming_books_llama
Dataset Card for "programming_books_llama"
400M tokens of programming books generated by gpt-3.5 (70M tokens) and a finetuned codellama 34b. The gpt-3.5 data is extremely high quality. The llama data has lower quality and shorter length, but is still good. This was generated with the textbook quality repo.
programming-solutionsThe programming solutions retrieval source for code-rag-bench, comprising programming solutions for the HumanEval and MBPP datasets.
paloma_programming_languagescodeagent-pythonsystems_programming_and_administrationHunter-Alpha-Programming-160000x
Hunter-Alpha-Programming-160000x - just a filtered version of the original dataset with like 50k more programming questions, DO NOT FINETUNE ON BOTH ONLY USE ONE
160,000 programming reasoning traces distilled from Hunter Alpha on OpenRouter at high and xhigh reasoning
Distribution:
Includes: Webdev, C++, Java, JS, C, Ruby, Lua, Rust, and C#
Token Count as of 3/17/2026: a lot idk probably 1 billion
[!NOTE]
This will be the last dataset update for hunter alpha I beleive… See the full description on the dataset page: https://huggingface.co/datasets/ianncity/Hunter-Alpha-Programming-160000x.quantum-compilation-and-programming
Neura Parse — Quantum Compilation & Programming
A code-heavy vertical on the quantum software/compilation stack: turning abstract quantum circuits and unitaries into device-executable programs. Covers unitary decomposition and circuit synthesis (Euler/ZYZ, KAK/Cartan, Solovay-Kitaev, Ross-Selinger gridsynth, numerical synthesis with BQSKit), gate-set/basis transpilation to native gate sets, qubit layout/mapping and routing under connectivity constraints (SABRE, VF2, SWAP… See the full description on the dataset page: https://huggingface.co/datasets/Neura-parse/quantum-compilation-and-programming.amalia-Nemotron-SFT-Competitive-Programming-v2
AMALIA Nemotron-SFT-Competitive-Programming-v2
Version of the nvidia/Nemotron-SFT-Competitive-Programming-v2 dataset used in the AMALIA's Supervised Fine-Tuning stage.
This dataset went through a processing pipeline to remove entries that reference other LLMs or research labs;
Original Dataset: https://huggingface.co/datasets/nvidia/Nemotron-SFT-Competitive-Programming-v2
This dataset is provided as part of the AMALIA project and is included in the data mix used to… See the full description on the dataset page: https://huggingface.co/datasets/amalia-llm/amalia-Nemotron-SFT-Competitive-Programming-v2.golang-programming-style-best-practicesNote: WIP - This dataset has not yet been curated to remove duplicates and filler
Dataset trained on several popular open source Golang style guides and the effective go book.
Source training data:
https://go.dev/doc/effective_go
https://github.com/uber-go/guide/blob/master/style.md
https://google.github.io/styleguide/go/best-practices
Q&A generated using Augment Toolkit
Generation models:
yi-large (API access kindly donated by 01.ai)
hermes-2-theta-llama-3-8b
qwen2-72b-instruct
build-competitive-programming-problem-dataset-cmsokff2
Build Competitive Programming Problem Dataset
Each item is a build competitive programming problem dataset example providing Problem statement, Constraints, Reference solution, Test cases, Time limit (ms). Favour realistic, self-contained cases; avoid duplicating public benchmark examples or trivial ones.
About
This dataset was produced by the DataBounty community and published here as part of an open, karma-only program.
Accepted items: 1000
Language: Python… See the full description on the dataset page: https://huggingface.co/datasets/databounty-io/build-competitive-programming-problem-dataset-cmsokff2.lunamax-multitask-programming-1000
LunaMax Multitask Programming 1000
A 1,000-record synthetic multitask programming dataset generated with
ChatGPT LunaMax.
The recovered dataset combines code review, implementation, bug and severity
classification, and strict output-contract tasks across multiple programming
languages.
The historical source shards were reviewed with ChatGPT 5.6 Sol High according
to dataset creator confirmation. During Hugging Face publication preparation,
all 1,000 records received a new… See the full description on the dataset page: https://huggingface.co/datasets/TaskPuppyAI/lunamax-multitask-programming-1000.Vulnerable_Programming_DatasetVulnerable Programming Dataset
Overview
The Vulnerable Programming Dataset is a comprehensive collection of 550 unique code vulnerabilities across 10 programming languages: Python, JavaScript, PHP, Java, Ruby, Go, TypeScript, C++, SQL, and C. Designed for cybersecurity professionals, red teamers, pentesters, and developers, this dataset highlights unconventional vulnerabilities such as insecure interprocess communication, misconfigured rate limiting, insecure dependency pinning, and logic… See the full description on the dataset page: https://huggingface.co/datasets/darkknight25/Vulnerable_Programming_Dataset.Nemotron-SFT-Competitive-Programming-v2-prompt-only
Nemotron-SFT-Competitive-Programming-v2-prompt-only
Prompt-only extraction from nvidia/Nemotron-SFT-Competitive-Programming-v2.
Files:
prompts.csv: one prompt extraction record per source row. Records include
prompt, separated system_prompt, and structured tools when the source row
defines available tools. Nested values are JSON-encoded inside CSV cells.
summary.md: source row counts, extracted row counts, count deltas, and failed prompt counts.
null_or_empty_rows.md: row… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-SFT-Competitive-Programming-v2-prompt-only.Nemotron-Competitive-Programming-v1-prompt-only
Nemotron-Competitive-Programming-v1-prompt-only
Prompt-only extraction from nvidia/Nemotron-Competitive-Programming-v1.
Files:
prompts.csv: one prompt extraction record per source row. Records include
prompt, separated system_prompt, and structured tools when the source row
defines available tools. Nested values are JSON-encoded inside CSV cells.
summary.md: source row counts, extracted row counts, count deltas, and failed prompt counts.
null_or_empty_rows.md: row indexes where… See the full description on the dataset page: https://huggingface.co/datasets/jamesdborin/Nemotron-Competitive-Programming-v1-prompt-only.Code-290k-labels-programming_languages-NO_Chatgpt
Para etiquetar los lenguajes de programación en un conjunto de datos extenso de fragmentos de código, se aplicaron técnicas automatizadas de procesamiento de texto y patrones específicos de cada lenguaje, sin recurrir al uso de modelos de lenguaje avanzados como ChatGPT o LLMs. Se inició con la extracción y preparación de datos usando pandas, una biblioteca de análisis de datos en Python, que facilitó la manipulación y el procesamiento del conjunto de datos obtenido de Hugging Face's… See the full description on the dataset page: https://huggingface.co/datasets/NickyNicky/Code-290k-labels-programming_languages-NO_Chatgpt.python_programming_questionslinear-programming