Lerobot SO-101 Manipulations Dataset

Materials Platform for Data Science Dataset (Project Pauling Files) is a highly curated inorganic materials dataset with a 30-year track record, based on about half a million peer-reviewed scientific publications and powering several successful commercial products from the world's leading publishers. It integrates materials data from 405,100 publications, linking 139,005 phase diagrams, 409,771 crystalline nanostructures, and 1,075,676 physical property sets into 189,682 materials phases within a large-scale materials graph of ~5 million nodes and ~150 million edges, making it a comprehensive materials platform for data science and a cornerstone among modern material science datasets and materials project database resources.

This droid dataset comprises 1,000 hours of multimodal recordings of cleaning, laundry folding, and dishwashing activities, combining head-mounted video with detailed robotic data streams from seven 9-axis IMU units positioned on the chest, upper arms, forearms, and wrists. The robot dataset captures complete robotic data through 3-axis accelerometer (linear acceleration), 3-axis gyroscope (angular velocity), and 3-axis magnetometer (magnetic field orientation) sensors, using onboard sensor-fusion to generate orientation quaternions rather than raw motion signals, enabling precise motion tracking and real-world robotic learning.

SWE-Bench Coding Tasks Dataset  is an extended programming languages dataset that builds on the original SWE-Bench benchmark with broader language coverage, golden/test patches, and real-world coding tasks like bug fixing, code completion, and automated code review. It supports coding agents, language models, and developer tools with verified benchmark scores and multi-language test sets.

Sound Effects Dataset contains 500,000 professionally recorded and curated sound effect audio files, designed for a wide range of creative, research, and machine learning applications. This high-quality audio dataset includes real-world and synthetic audio events, rich metadata, and stereo recordings, enabling audio classification, speech recognition, and scene analysis across diverse acoustic environments.

This music genres dataset contains 500,000 studio-grade music tracks in lossless FLAC format, designed for music genre classification and detection tasks. It provides rich music metadata, including detailed genre labels, instruments, and artist information, making it ideal training data for machine learning and deep learning models in audio analysis.