Datasets
Conjuntos de datos de entrenamiento de IA de código abierto y alta calidad
Colección de Conjuntos de Datos de Rostros Abrazados
El mayor hub de datasets open source de la web, alojado por Hugging Face. Abarca tareas de NLP, visión por computador, audio y multimodales, con cientos de miles de datasets contribuidos por la comunidad y utilidades eficientes de streaming y carga. Ya sea que necesites benchmarks de clasificación de texto, pares imagen-texto o corpus de voz a texto, esta es la infraestructura de referencia para conseguir datos de calidad.
Plataforma de Conjuntos de Datos Kaggle
Kaggle aloja decenas de miles de datasets públicos que abarcan competiciones de machine learning, problemas empresariales reales e investigación académica. Todos están disponibles para análisis en línea mediante notebooks integrados y pueden descargarse para experimentación local y entrenamiento de modelos.
Artículos con conjuntos de datos en código
Papers with Code enlaza datasets de referencia de papers académicos con sus modelos de última generación y implementaciones de código abierto, facilitando comparar métodos, reproducir resultados y encontrar el dataset y baseline de código adecuados para tu investigación.
Búsqueda de Conjuntos de Datos en Google
Google Dataset Search es un buscador especializado que indexa datasets públicos de miles de fuentes, permitiendo a investigadores y desarrolladores localizar rápidamente los datos que necesitan para entrenamiento, análisis o validación, sin importar dónde estén alojados.
Corpus de Rastreo Común
Common Crawl es un corpus abierto de rastreo web a escala de petabytes que sirve como fuente de datos fundamental para el preentrenamiento de muchos modelos de lenguaje grandes. El corpus está disponible para descarga gratuita y se usa ampliamente en investigación NLP, minería de datos y experimentos de ML a gran escala.
Conjuntos de datos multimodales abiertos LAION
LAION (Large-Scale Artificial Intelligence Open Network) ofrece enormes datasets abiertos de pares imagen-texto utilizados ampliamente para entrenar modelos multimodales como CLIP y Stable Diffusion. Los datasets son de acceso libre y constituyen infraestructura clave para la investigación en IA generativa abierta.
Repositorio de Aprendizaje Automático UCI
El UCI Machine Learning Repository es una colección clásica de cientos de datasets estructurados mantenida desde finales de los años 90, ampliamente usada para práctica, benchmarking de algoritmos y enseñanza de conceptos fundamentales de machine learning en cursos académicos de todo el mundo.
El conjunto de datos de código fuente de la pila
The Stack es un gran dataset de código fuente con licencias permisivas del proyecto BigCode, que cubre cientos de lenguajes de programación. Se usa ampliamente para preentrenar y evaluar modelos de generación de código, e incluye mecanismos de opt-out y herramientas de deduplicación para prácticas de datos responsables.
ImageNet
ImageNet es un dataset de imágenes histórico a gran escala con millones de imágenes etiquetadas en miles de categorías. Impulsó la revolución del deep learning en visión por computador mediante la competición ILSVRC y sigue siendo fundamental para preentrenamiento, transfer learning y evaluación de modelos de clasificación.
alpaca cleaned
yahma/alpaca-cleaned is a community dataset on Hugging Face with 876 likes and 24,234 recent downloads.
Alpaca CoT
QingyiSi/Alpaca-CoT is a community dataset on Hugging Face with 786 likes and 9,743 recent downloads.
alpaca
tatsu-lab/alpaca is a community dataset on Hugging Face with 1104 likes and 119,254 recent downloads.
bad prompt
Nerfgun3/bad_prompt is a community dataset on Hugging Face with 963 likes and 1,957 recent downloads.
c4
allenai/c4 is a community dataset on Hugging Face with 643 likes and 1,282,773 recent downloads.
Chinese DeepSeek R1 Distill data 110k
Congliu/Chinese-DeepSeek-R1-Distill-data-110k is a community dataset on Hugging Face with 781 likes and 1,087 recent downloads.
COIG CQIA
m-a-p/COIG-CQIA is a community dataset on Hugging Face with 768 likes and 27,976 recent downloads.
cosmopedia
HuggingFaceTB/cosmopedia is a community dataset on Hugging Face with 742 likes and 26,991 recent downloads.
CulturaX
uonlp/CulturaX is a community dataset on Hugging Face with 674 likes and 18,353 recent downloads.
databricks dolly 15k
databricks/databricks-dolly-15k is a community dataset on Hugging Face with 1084 likes and 63,630 recent downloads.
diffusiondb
poloclub/diffusiondb is a community dataset on Hugging Face with 653 likes and 13,093 recent downloads.
dolma
allenai/dolma is a community dataset on Hugging Face with 1074 likes and 2,784 recent downloads.
EasyNegative
gsdf/EasyNegative is a community dataset on Hugging Face with 1212 likes and 27,637 recent downloads.
falcon refinedweb
tiiuae/falcon-refinedweb is a community dataset on Hugging Face with 953 likes and 82,020 recent downloads.
finepdfs
HuggingFaceFW/finepdfs is a community dataset on Hugging Face with 928 likes and 52,686 recent downloads.
fineweb
HuggingFaceFW/fineweb is a community dataset on Hugging Face with 3287 likes and 429,372 recent downloads.
fineweb 2
HuggingFaceFW/fineweb-2 is a community dataset on Hugging Face with 864 likes and 92,367 recent downloads.
fineweb edu
HuggingFaceFW/fineweb-edu is a community dataset on Hugging Face with 1272 likes and 405,910 recent downloads.
GAIA
gaia-benchmark/GAIA is a community dataset on Hugging Face with 783 likes and 10,260 recent downloads.
glue
nyu-mll/glue is a community dataset on Hugging Face with 730 likes and 784,661 recent downloads.
gsm8k
openai/gsm8k is a community dataset on Hugging Face with 1593 likes and 1,235,950 recent downloads.
hh rlhf
Anthropic/hh-rlhf is a community dataset on Hugging Face with 2010 likes and 39,057 recent downloads.
hle
cais/hle is a community dataset on Hugging Face with 948 likes and 40,101 recent downloads.
imagenet 1k
ILSVRC/imagenet-1k is a community dataset on Hugging Face with 896 likes and 114,854 recent downloads.
imdb
stanfordnlp/imdb is a community dataset on Hugging Face with 699 likes and 189,340 recent downloads.
Infinity Instruct
BAAI/Infinity-Instruct is a community dataset on Hugging Face with 756 likes and 3,673 recent downloads.
internship warehouse
FlyRank/internship-warehouse is a community dataset on Hugging Face with 588 likes and 18,716 recent downloads.
Llama Nemotron Post Training Dataset
nvidia/Llama-Nemotron-Post-Training-Dataset is a community dataset on Hugging Face with 700 likes and 7,276 recent downloads.
LLaVA Instruct 150K
liuhaotian/LLaVA-Instruct-150K is a community dataset on Hugging Face with 635 likes and 5,676 recent downloads.
lmsys chat 1m
lmsys/lmsys-chat-1m is a community dataset on Hugging Face with 984 likes and 6,496 recent downloads.
medical o1 reasoning SFT
FreedomIntelligence/medical-o1-reasoning-SFT is a community dataset on Hugging Face with 1176 likes and 17,355 recent downloads.
medical prescription english audio
HumynLabs/medical-prescription-english-audio is a community dataset on Hugging Face with 716 likes and 152 recent downloads.
mmlu
cais/mmlu is a community dataset on Hugging Face with 826 likes and 714,868 recent downloads.
MNBVC
liwu/MNBVC is a community dataset on Hugging Face with 652 likes and 99,676 recent downloads.
NuminaMath CoT
AI-MO/NuminaMath-CoT is a community dataset on Hugging Face with 599 likes and 161,288 recent downloads.
oasst1
OpenAssistant/oasst1 is a community dataset on Hugging Face with 1567 likes and 30,179 recent downloads.
OpenHermes 2.5
teknium/OpenHermes-2.5 is a community dataset on Hugging Face with 896 likes and 32,526 recent downloads.
OpenOrca
Open-Orca/OpenOrca is a community dataset on Hugging Face with 1589 likes and 25,012 recent downloads.
OpenR1 Math 220k
open-r1/OpenR1-Math-220k is a community dataset on Hugging Face with 787 likes and 126,715 recent downloads.
OpenThoughts 114k
open-thoughts/OpenThoughts-114k is a community dataset on Hugging Face with 907 likes and 110,694 recent downloads.
Opus 4.6 Reasoning 3000x filtered
nohurry/Opus-4.6-Reasoning-3000x-filtered is a community dataset on Hugging Face with 648 likes and 586 recent downloads.
PersonaHub
proj-persona/PersonaHub is a community dataset on Hugging Face with 797 likes and 11,476 recent downloads.
PhysicalAI Autonomous Vehicles
nvidia/PhysicalAI-Autonomous-Vehicles is a community dataset on Hugging Face with 1003 likes and 208,205 recent downloads.
prompts.chat
fka/prompts.chat is a community dataset on Hugging Face with 9815 likes and 25,078 recent downloads.
RedPajama Data 1T
togethercomputer/RedPajama-Data-1T is a community dataset on Hugging Face with 1190 likes and 2,094 recent downloads.
ShareGPT Vicuna unfiltered
anon8231489123/ShareGPT_Vicuna_unfiltered is a community dataset on Hugging Face with 915 likes and 419,677 recent downloads.
squad
rajpurkar/squad is a community dataset on Hugging Face with 709 likes and 265,395 recent downloads.
synthetic text to sql
gretelai/synthetic_text_to_sql is a community dataset on Hugging Face with 691 likes and 2,938 recent downloads.
the stack v2
bigcode/the-stack-v2 is a community dataset on Hugging Face with 624 likes and 5,758 recent downloads.
TinyStories
roneneldan/TinyStories is a community dataset on Hugging Face with 1140 likes and 84,980 recent downloads.
ultrachat 200k
HuggingFaceH4/ultrachat_200k is a community dataset on Hugging Face with 885 likes and 112,449 recent downloads.
+ 948 more — VER TODO →