Datasets

xperience-10m

⚠️ Important: If you have already submitted an access request but have not completed the required DocuSign agreement, your request will remain pending. Please complete signing and we will grant access once verified. Interactive Intelligence from Human Xperience Xperience-10M Dataset Summary Xperience-10M is a large-scale egocentric multimodal dataset of human experience for embodied AI, robotics, world models, and spatial… See the full description on the dataset page: https://huggingface.co/datasets/ropedia-ai/xperience-10m.

PhysicalAI-Autonomous-Vehicles

PHYSICAL AI AUTONOMOUS VEHICLES The PhysicalAI-Autonomous-Vehicles dataset provides one of the largest, geographically diverse collections of multi-sensor data empowering AV researchers to build the next generation of Physical AI based end-to-end driving systems. This dataset is ready for commercial/non-commercial AV use per the license agreement. Data Collection Method Automatic/Sensor Labeling Method Automatic/Sensor This dataset has a total of 1700 hours of driving… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles.

Egocentric-100K

Egocentric-100K is the largest dataset of manual labor. You can visualize the dataset here. Egocentric-100K is state-of-the-art in hand visibility and active manipulation density compared to previous in-the-wild egocentric datasets. The complete 30,000 frame evaluation set is available at Egocentric-100K-Evaluation. Dataset Statistics Attribute Value Total Hours 100,405 Total Frames 10.8 billion Video Clips 2,010,759 Median Clip Length 180.0 seconds Mean Hours… See the full description on the dataset page: https://huggingface.co/datasets/builddotai/Egocentric-100K.

PhysicalAI-Robotics-GR00T-X-Embodiment-Sim

PhysicalAI-Robotics-GR00T-X-Embodiment-Sim Github Repo: Isaac GR00T N1 We provide a set of datasets used for post-training of GR00T N1. Each dataset is a collection of trajectories from different robot embodiments and tasks. Cross-embodied bimanual manipulation: 9k trajectories Dataset Name #trajectories bimanual_panda_gripper.Threading 1000 bimanual_panda_hand.LiftTray 1000 bimanual_panda_gripper.ThreePieceAssembly 1000… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-GR00T-X-Embodiment-Sim.

jat-dataset

JAT Dataset Dataset Description The Jack of All Trades (JAT) dataset combines a wide range of individual datasets. It includes expert demonstrations by expert RL agents, image and caption pairs, textual data and more. The JAT dataset is part of the JAT project, which aims to build a multimodal generalist agent. Paper: https://huggingface.co/papers/2402.09844 Usage >>> from datasets import load_dataset >>> dataset = load_dataset("jat-project/jat-dataset"… See the full description on the dataset page: https://huggingface.co/datasets/jat-project/jat-dataset.

dronescapes2

Dronescapes Experts dataset This dataset is an extension of the original dronescapes dataset with new modalities generated using VRE 100% from scratch (aka pretrained experts). The only data that is not generable by VRE is the Ground Truth: semantic (human annotated), depth & normals (SfM) that is inherited from the original dataset for evaluation purposes only. 1. Downloading the data Option 1. Download the pre-processed dataset from HuggingFace… See the full description on the dataset page: https://huggingface.co/datasets/Meehai/dronescapes2.

droid_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "franka", "total_episodes": 92233, "total_frames": 27044326, "total_tasks": 31308, "total_videos": 276699, "total_chunks": 93, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:92233" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/droid_lerobot.

language_table_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "xarm", "total_episodes": 442226, "total_frames": 7045476, "total_tasks": 127605, "total_videos": 442226, "total_chunks": 443, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:442226" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/language_table_lerobot.

robotwin-clean-and-aug-lerobot

Robotwin Dataset Robotwin dataset in Lerobot format, with video latents already extracted in WAN 2.2 format, ready for use in Lingbot-VA post-training. License Agreement This project is licensed under the CC BY-NC-SA 4.0.

PhysicalAI-Robotics-Open-H-Embodiment

As of 3/16 the last components of the Open-H-Embodiment dataset are still being uploaded Dataset Description: Open-H-Embodiment is a community‑driven dataset initiative building the open, shared foundation needed to train and evaluate AI autonomy models for surgical robotics and ultrasound. This dataset is a multi-embodiment collection of LeRobot datasets of paired kinematics and video, across tasks such as tabletop exercises, clinical procedures, as well as simulations… See the full description on the dataset page: https://huggingface.co/datasets/introvoyz041/PhysicalAI-Robotics-Open-H-Embodiment.

course-images

Dataset from huggingface-deep-rl-course. Tags: region:us.

PhysicalAI-Robotics-GR00T-X-Embodiment-Sim

PhysicalAI-Robotics-GR00T-X-Embodiment-Sim Github Repo: Isaac GR00T N1 We provide a set of datasets used for post-training of GR00T N1. Each dataset is a collection of trajectories from different robot embodiments and tasks. Cross-embodied bimanual manipulation: 9k trajectories Dataset Name #trajectories bimanual_panda_gripper.Threading 1000 bimanual_panda_hand.LiftTray 1000 bimanual_panda_gripper.ThreePieceAssembly 1000… See the full description on the dataset page: https://huggingface.co/datasets/sini-21/PhysicalAI-Robotics-GR00T-X-Embodiment-Sim.

bridge_orig_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "widowx", "total_episodes": 53192, "total_frames": 1893026, "total_tasks": 19974, "total_videos": 212768, "total_chunks": 54, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:53192" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/bridge_orig_lerobot.

dronescapes

Dronescapes dataset Visit the official website for more information: link. This dataset was introduced in our ICCV 2023 workshop paper: link. For citing, see at the end of the page. Note: An fully-automated extended variant of this dataset (generating new modalities as inputs) is available at this repository: link. 1. Downloading the data git lfs install # Make sure you have git-lfs installed (https://git-lfs.com) git clone… See the full description on the dataset page: https://huggingface.co/datasets/Meehai/dronescapes.

RoboTwin2.0

Dataset from TianxingChen. Tags: license:mit, region:us.

PhysicalAI-Robotics-Open-H-Embodiment

Dataset Description: Open-H-Embodiment is a community‑driven dataset initiative building the open, shared foundation needed to train and evaluate AI autonomy models for surgical robotics and ultrasound. This dataset is a multi-embodiment collection of LeRobot datasets of paired kinematics and video, across tasks such as tabletop exercises, clinical procedures, as well as simulations of healthcare robotics applications. Maintainer / Hosting Organization: NVIDIA… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-Open-H-Embodiment.

bridge_orig_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "widowx", "total_episodes": 53192, "total_frames": 1893026, "total_tasks": 19974, "total_videos": 212768, "total_chunks": 54, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:53192" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/arekborucki/bridge_orig_lerobot.

fractal20220817_data_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "google_robot", "total_episodes": 87212, "total_frames": 3786400, "total_tasks": 599, "total_videos": 87212, "total_chunks": 88, "chunks_size": 1000, "fps": 3, "splits": { "train": "0:87212" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/fractal20220817_data_lerobot.

PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams

PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams Paper | Paper Website | GitHub Download We provide a download script to download our dataset. If you have enough space, you can use git to download a dataset from huggingface. usage: download.py [-h] --odir ODIR [--file_types {hdmap,lidar,synthetic}[,…]] [--workers N] [--clean_cache] required arguments: --odir ODIR Output… See the full description on the dataset page: https://huggingface.co/datasets/kze193825/PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams.

ai2thor-hab

AI2THOR-Hab AI2THOR scene datasets include iTHOR, RoboTHOR, ProcTHOR-10K, ArchitecTHOR. Many of the assets of the interactable objects are shared across these datasets. iTHOR: includes 120 single room scenes, 30 scenes for each bedroom, bathroom, kitchen, and living room. In our extracted dataset, there are additional 30 foyers scenes. RoboTHOR: includes 89 apartments in maze style, where the rooms are subdivided by wall panels. Same of the scenes share the same room layout, but… See the full description on the dataset page: https://huggingface.co/datasets/hssd/ai2thor-hab.

mesa-all-train-lerobot

Dataset from albertwilcox. Tags: license:cc-by-4.0, region:us.

PhysicalAI-Robotics-Manipulation-Kitchen-Demos

PhysicalAI-Robotics-Manipulation-Kitchen-Demos We provide a 600 hours of human-teleoperated demonstrations across 316 different tasks, totalling 55k trajectories. The datasets are collected using Franka Panda robot with an Omron mobile base. The datasets follow the LeRobot format. Here is an overview of important elements of each dataset: Click to expand dataset structure lerobot/ ├── meta/ # Metadata files describing the dataset │ ├── info.json… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-Manipulation-Kitchen-Demos.

mdm_depth

LingBot-Depth Dataset Self-curated RGB-D dataset for training LingBot-Depth, a masked depth modeling approach (arxiv:2601.17895). Each sample contains an RGB image, raw sensor depth, and ground truth depth. Total size: 2.71 TBDepth scale: millimeters (mm), stored as 16-bit PNGLicense: CC BY-NC-SA 4.0 Sub-datasets Name Description Samples RobbyReal Real-world indoor scenes captured with multiple RGB-D cameras 1,400,000 RobbyVla Real-world data collected… See the full description on the dataset page: https://huggingface.co/datasets/robbyant/mdm_depth.

FastUMI_100k_lerobot

Dataset from IPEC-COMMUNITY. Tags: region:us.

mikasa_robo_all_lerobot

Original dataset here: https://github.com/CognitiveAISystems/MIKASA-Robo The original dataset is published in .npz files. The original action is delta joint. I changed the action to be delta x,y,z, yaw, pitch, roll of the end effector. The delta action is calculated by subtracting the state of end effector from successive frames.

humanoid-everyday

Humanoid Everyday A Comprehensive Robotic Dataset for Open-World Humanoid Manipulation Overview Humanoid Everyday is a large-scale, diverse humanoid manipulation dataset designed for open-world robotic learning and embodied intelligence. It contains over 260 tasks across 7 major categories, covering dexterous manipulation, human–humanoid interaction, and locomotion-integrated activities.All data were collected through a human-supervised teleoperation pipeline, recording… See the full description on the dataset page: https://huggingface.co/datasets/USC-PSI-Lab/humanoid-everyday.

kuka_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "kuka_iiwa", "total_episodes": 209880, "total_frames": 2455879, "total_tasks": 1, "total_videos": 209880, "total_chunks": 210, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:209880" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/kuka_lerobot.

TartanGround

TartanGround: A Large-Scale Dataset for Ground Robot Perception and Navigation Dataset Description TartanGround is a large-scale, multi-modal dataset designed to advance the perception and autonomy of ground robots operating in diverse environments. Collected across 63 photorealistic simulation environments, it provides comprehensive data streams for various robotic tasks. Key Features Environments: 63 diverse simulation environments categorized into: Indoor… See the full description on the dataset page: https://huggingface.co/datasets/theairlabcmu/TartanGround.

pi05-metaworld-activations-v1-15env

Dataset from brandonyang. Tags: region:us.

control-arena-runs

Dataset from idacy. Tags: region:us.

hh-rlhf

Dataset Card for HH-RLHF Dataset Summary This repository provides access to two different kinds of data: Human preference data about helpfulness and harmlessness from Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback. These data are meant to train preference (or reward) models for subsequent RLHF training. These data are not meant for supervised training of dialogue agents. Training dialogue agents on these data is likely to lead… See the full description on the dataset page: https://huggingface.co/datasets/Anthropic/hh-rlhf.

ManiTwin-100K

ManiTwin-100K: Manipulation-Ready Digital Object Twins Project Page | Paper ManiTwin-100K is a large-scale dataset of manipulation-ready digital object twins designed for robotic manipulation research. Each object includes simulation-ready 3D meshes, physical properties, functional point annotations, grasp configurations, and rich language descriptions—all validated through physics-based simulation. Note: We are currently releasing approximately 1K sample objects with a… See the full description on the dataset page: https://huggingface.co/datasets/ManiTwin/ManiTwin-100K.

msi-drone-crop-surveys

Dataset Metadata Identification Information Citation Title:Aerial surveys of a sunflower crop’s lifecycle from April to September 2023 Originator: Sofia University - Faculty of Mathematics and Informatics, SAP LABS Bulgaria Publication Date: 2023.11.08 Abstract Efficient food production is shaping up to be one of the new frontiers for new technologies and solutions. One such prominent domain is the remote sensing ecosystem, and more precicely… See the full description on the dataset page: https://huggingface.co/datasets/su-fmi/msi-drone-crop-surveys.

RealSource-World

RealSource World RealSource World is a large-scale real-world robotics manipulation dataset collected using RS-02 dual-arm humanoid robot. This dataset contains diverse long-horizon manipulation tasks performed in real-world environments, with detailed annotations for atomic skills and quality assessments. Key Features 14+ million frames of real-world dual-arm manipulation demonstrations. 11,428+ episodes across 35 distinct manipulation tasks. 71-dimensional… See the full description on the dataset page: https://huggingface.co/datasets/RealSourceData/RealSource-World.

PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams

PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams Paper | Paper Website | GitHub Download We provide a download script to download our dataset. If you have enough space, you can use git to download a dataset from huggingface. usage: download.py [-h] --odir ODIR [--file_types {hdmap,lidar,synthetic}[,…]] [--workers N] [--clean_cache] required arguments: --odir ODIR Output… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicle-Cosmos-Drive-Dreams.

robotwin2_lerobot_v21

Dataset from IrvingJackson. Tags: size_categories:10K<n<100K, modality:video, library:datasets, library:mlcroissant, region:us.

OmniAction

RoboOmni: Proactive Robot Manipulation in Omni-modal Context 📖 arXiv Paper (Accepted to ICLR 2026 🎉) | 🌐 Website | 🤗 Model | 🤗 Dataset | 🛠️ Github | Recent advances in Multimodal Large Language Models (MLLMs) have driven rapid progress in Vision–Language–Action (VLA) models for robotic manipulation. Although effective in many scenarios, current approaches largely rely on explicit instructions, whereas in real-world interactions, humans rarely issue… See the full description on the dataset page: https://huggingface.co/datasets/rad1d1m123/OmniAction.

osworld_office_RL

Dataset from Bin12345. Tags: size_categories:10K<n<100K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

DataDump_02-03-2026

RoboArena Dataset Snapshot — 2026‑02‑03 This dataset contains autonomous policy rollouts, task success scores, policy preference annotations, and related metadata collected by the RoboArena benchmark from its inception through February 3, 2026. Data layout DataDump_02-03-2026/ ├── global_metadata.yaml └── evaluation_sessions/ ├── <eval_session_id_0>/ │ └── … ├── <eval_session_id_1>/ │ └── … └── <eval_session_id_n>/ ├── metadata.yaml… See the full description on the dataset page: https://huggingface.co/datasets/RoboArena/DataDump_02-03-2026.

Galaxea-Open-World-Dataset

🚀 Galaxea Open-World Dataset Key features 500+ hours of real-world mobile manipulation data. All data collected using one uniform robotic embodiment for consistency. Fine-grained subtask language annotations. Covers residential, kitchen, retail, and office settings. Dataset in RLDS and LeRobot format. Dataset Structure For convenience, we divided the 500 hours of data into four equal parts by time. We also provide a small sample dataset for quick… See the full description on the dataset page: https://huggingface.co/datasets/OpenGalaxea/Galaxea-Open-World-Dataset.

grand_tour_dataset

The GrandTour Dataset A project brought to you by RSL - ETH Zurich. References • Contributing • Citation References Official dataset webpage: grand-tour.leggedrobotics.com Getting started & examples: github.com/leggedrobotics/grand_tour_dataset Boxi used to collect the data: github.com/leggedrobotics/grand_tour_box Contributing We warmly welcome contributions to improve and expand this project. Whether it's new examples, enhancements, or… See the full description on the dataset page: https://huggingface.co/datasets/leggedrobotics/grand_tour_dataset.

OmniAction

RoboOmni: Proactive Robot Manipulation in Omni-modal Context 📖 arXiv Paper (Accepted to ICLR 2026 🎉) | 🌐 Website | 🤗 Model | 🤗 Dataset | 🛠️ Github | Recent advances in Multimodal Large Language Models (MLLMs) have driven rapid progress in Vision–Language–Action (VLA) models for robotic manipulation. Although effective in many scenarios, current approaches largely rely on explicit instructions, whereas in real-world interactions, humans rarely issue… See the full description on the dataset page: https://huggingface.co/datasets/OpenMOSS-Team/OmniAction.

visgym_data

VisGym Dataset Project Page | Paper | GitHub VisGym consists of 17 diverse, long-horizon environments designed to systematically evaluate, diagnose, and train Vision-Language Models (VLMs) on visually interactive tasks. In these environments, agents must select actions conditioned on both their past actions and observation history, challenging their ability to handle complex, multimodal sequences. Dataset Summary This dataset contains trajectories and interaction data… See the full description on the dataset page: https://huggingface.co/datasets/VisGym/visgym_data.

mesa-70-lerobot

Dataset from albertwilcox. Tags: license:cc-by-4.0, size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant.

bridge_v2_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "widowx", "total_episodes": 53192, "total_frames": 1999410, "total_tasks": 19974, "total_videos": 212768, "total_chunks": 54, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:53192" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/jesbu1/bridge_v2_lerobot.

libero-microwave-grm-rollouts

LIBERO Microwave GRM Rollouts Dense-reward-annotated rollout dataset from GRPO training of OpenVLA-OFT on LIBERO-10 Task 9 ("put the yellow and white mug in the microwave and close it"). Dataset Stat Value Episodes ~2,100 (T >= 5 steps) Format LeRobot (parquet + images) Task put the yellow and white mug in the microwave and close it Size 41 GB Reward model Robo-Dopamine GRM-3B Policy OpenVLA-OFT (LoRA, GRPO-trained) Simulator LIBERO (robosuite)… See the full description on the dataset page: https://huggingface.co/datasets/Auryal/libero-microwave-grm-rollouts.

control-arena-32217692-valid-runs-20260312

control-arena 32217692 valid runs Source host: 32217692 Created: 2026-03-12T00:53:34.882011+00:00 Contents: valid and potentially useful run directories uploaded separately under valid_runs/. Includes the partially successful pck-chain-sonnet45-pos-r1x5-cached batch, retained because most outputs are valid. Manifest: 32217692_valid_hf_manifest.csv.

humanoid-everyday

Humanoid Everyday A Comprehensive Robotic Dataset for Open-World Humanoid Manipulation Overview Humanoid Everyday is a large-scale, diverse humanoid manipulation dataset designed for open-world robotic learning and embodied intelligence. It contains over 260 tasks across 7 major categories, covering dexterous manipulation, human–humanoid interaction, and locomotion-integrated activities.All data were collected through a human-supervised teleoperation pipeline, recording… See the full description on the dataset page: https://huggingface.co/datasets/SmallSmartPiggy/humanoid-everyday.

RoboMIND

RoboMIND: Benchmark on Multi-embodiment Intelligence Normative Data for Robot Manipulation Accepted by Robotics: Science and Systems (RSS) 2025. 📢 ANNOUNCEMENT | RoboMIND V2.0 Release We're excited to announce the release of RoboMIND V2.0! 🔗 Access the complete RoboMIND V2.0 collection on ModelScope 💾 Overview of RoboMIND 💾 🤖 Composition of RoboMIND 🤖 We present RoboMIND (Multi-embodiment Intelligence Normative Dataset and Benchmark… See the full description on the dataset page: https://huggingface.co/datasets/x-humanoid-robomind/RoboMIND.

PhysicalAI-Robotics-GR00T-Teleop-GR1

Introduction TL;DR: DreamDojo is a generalist robot world model pretrained on 44k hours of human egocentric data, showing unprecedented generalization to diverse objects and environments. Project page: https://dreamdojo-world.github.io/ Paper: https://arxiv.org/abs/2602.06949 Code: https://github.com/NVIDIA/DreamDojo How to Use Check out https://github.com/NVIDIA/DreamDojo Citation @article{gao2026dreamdojo, title={DreamDojo: A Generalist Robot… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-GR00T-Teleop-GR1.

unieqa

Dataset from EmbodiedArena. Tags: modality:image, region:us.

patch-of-behavior-1k

BEHAVIOR-1K Asset Database (Patch) Complete metadata and visualizations for all 8662 BEHAVIOR-1K object assets. Contents Path Description assets.db SQLite database — descriptions + DoF for all 8662 assets objects/{category}/{model}/description.json VLM-generated visual description (5676 assets) objects/{category}/{model}/dof.json Articulation / DoF metadata objects/{category}/{model}/front.png Front-view visualization objects/{category}/{model}/back.png… See the full description on the dataset page: https://huggingface.co/datasets/EastDong0704/patch-of-behavior-1k.

PhysicalAI-Autonomous-Vehicles-NuRec

PHYSICAL AI AUTONOMOUS VEHICLES NUREC The PhysicalA-AutonomousVehicle-NuRec dataset contains 918 scenes of dynamic neural reconstructions. The scenes are about 20 seconds long and stored in the form of usdz files, along with respective surface meshes. The reconstructions were generated using 6 camera views (front-wide 120 deg, front-tele 30 deg, cross right/left 120 deg and rear right/left 70 deg). Users can use these 3D reconstructed driving scenes for training and testing their… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles-NuRec.

pusht

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 206, "total_frames": 25650, "total_tasks":1, "total_videos": 206, "total_chunks": 1, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:206" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/pusht.

Evo1_MetaWorld_Dataset

Dataset from limiqsqsqdq. Tags: region:us.

vlabench_primitive_ft_groot_lerobot

Dataset from happyhappy-jun. Tags: size_categories:10K<n<100K, modality:video, library:datasets, library:mlcroissant, region:us.

control-arena-32709064-backup-20260327

ControlArena ControlArena is a Python library for running AI Control experiments, created by the UK AI Security Institute and Redwood Research. AI Control aims to design and evaluate control protocols. Control protocols are plans designed to prevent unsafe actions by AI systems, even if those AIs were misaligned and intended to subvert the safeguards. 📚 Documentation: Visit our documentation site for detailed guides and API reference. 💬 Community & Support: Join us in the… See the full description on the dataset page: https://huggingface.co/datasets/idacy/control-arena-32709064-backup-20260327.

robotwin-assets

Dataset from sssssssssssssssy. Tags: license:mit, size_categories:1K<n<10K, format:imagefolder, modality:image, modality:video.

rlbench-raw-test

Dataset from KEVIN04087. Tags: size_categories:10K<n<100K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

ArtVIP

ArtVIP dataset card 🎉🎉🎉 ArtVIP is accepted by ICLR 2026 Key Features ✅ 206 high-quality digital-twin articulated objects. ✅ 6 pre-configured digital-twin scenes, 6 user-defined scenes ✅ Reusable Modular interaction ✅ Physics fidelity ✅ Pixel-level affordance annotations **** Version History Click here to view… See the full description on the dataset page: https://huggingface.co/datasets/X-Humanoid/ArtVIP.

control-arena-rescores

Dataset from idacy. Tags: region:us.

anima-thermal-synthetic

ANIMA Thermal Synthetic Dataset See Through Fog, Rain, Smoke, and Total Darkness 27GB | 30,781 paired RGB-to-Thermal images | Production-ready Built by Robot Flow Labs / AIFLOW LABS LIMITED Part of the ANIMA Perception Suite — 96 AI modules for robotics The Problem Thermal cameras see what RGB cameras cannot: through fog, rain, smoke, and complete darkness. But thermal training data is scarce and expensive: Radiometric thermal cameras cost $5,000+… See the full description on the dataset page: https://huggingface.co/datasets/robotflowlabs/anima-thermal-synthetic.

robocasa_mg_lerobot

Dataset from daixianjie. Tags: region:us.

Retargeted_AMASS_for_robotics

Retargeted AMASS for Robotics Project Overview This project aims to retarget motion data from the AMASS dataset to various robot models and open-source the retargeted data to facilitate research and applications in robotics and human-robot interaction. AMASS (Archive of Motion Capture as Surface Shapes) is a high-quality human motion capture dataset, and the SMPL-X model is a powerful tool for generating realistic human motion data. By adapting the motion data from AMASS… See the full description on the dataset page: https://huggingface.co/datasets/fleaven/Retargeted_AMASS_for_robotics.

RoboTwin-Randomized

Dataset from StarVLA. Tags: size_categories:10K<n<100K, modality:video, library:datasets, library:mlcroissant, region:us.

BRIGHT

BRIGHT benchmark BRIGHT is the first text retrieval benchmark that requires intensive reasoning to retrieve relevant documents. The queries are collected from diverse domains (StackExchange, LeetCode, and math competitions), all sourced from realistic human data. Experiments show that existing retrieval models perform poorly on BRIGHT, where the highest score is only 22.1 measured by nDCG@10. BRIGHT provides a good testbed for future retrieval research in more realistic and… See the full description on the dataset page: https://huggingface.co/datasets/xlangai/BRIGHT.

SynthRender_Robotics

Dataset from moiaraya. Tags: license:cc-by-nc-3.0, size_categories:100K<n<1M, format:text, modality:image, modality:text.

EO-Data1.5M

🤖 EO-Data-1.5M A Large-Scale Interleaved Vision-Text-Action Dataset for Embodied AI The first large-scale interleaved embodied dataset emphasizing temporal dynamics and causal dependencies among vision, language, and action modalities. 📊 Dataset Overview EO-Data-1.5M is a massive, high-quality multimodal embodied reasoning dataset designed for training generalist robot… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/EO-Data1.5M.

pi05-metaworld-activations-v1-2env

Dataset from brandonyang. Tags: region:us.

droid_1.0.1

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "Franka", "total_episodes": 95617, "total_frames": 27618651, "total_tasks": 49611, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 500, "fps": 15, "splits": { "train": "0:95617"}, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/droid_1.0.1.

DECO-50

DECO: Decoupled Multimodal Diffusion Transformer for Bimanual Dexterous Manipulation with a Plugin Tactile Adapter DECO-50 is a bimanual dexterous manipulation dataset with tactile sensing, comprising 50 hours of teleoperated data across 4 scenarios and 28 subtasks, totaling over 5 million frames collected on real dual-arm robots. Dataset Structure DECO-50/ ├── task1/ │ ├── sub_task_1/ │ │ ├── episode_000000/ │ │ │ ├──… See the full description on the dataset page: https://huggingface.co/datasets/BAAI-Humanoid/DECO-50.

dsqa

Dataset from rl-research. Tags: size_categories:n<1K, format:parquet, modality:text, library:datasets, library:pandas.

modified_libero_rlds

Modified LIBERO RLDS Datasets This repository contains the four modified LIBERO datasets used in the OpenVLA fine-tuning experiments, stored in RLDS data format. See Appendix E in the OpenVLA paper for details about the fine-tuning experiments and specific dataset modifications, and see the OpenVLA GitHub README for instructions on how to run OpenVLA in LIBERO environments. Citation BibTeX: @article{kim24openvla, title={OpenVLA: An Open-Source… See the full description on the dataset page: https://huggingface.co/datasets/openvla/modified_libero_rlds.

OpenX-Embodiment

Open X-Embodiment Dataset (unofficial) This is an unofficial Dataset Repo. This Repo is set up to make Open X-Embodiment Dataset (55 in 1) more accessible for people who love huggingface🤗. Open X-Embodiment Dataset is the largest open-source real robot dataset to date. It contains 1M+ real robot trajectories spanning 22 robot embodiments, from single robot arms to bi-manual robots and quadrupeds. More information is located on RT-X website… See the full description on the dataset page: https://huggingface.co/datasets/jxu124/OpenX-Embodiment.

metaworld_mt50

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "metaworld", "total_episodes": 2500, "total_frames": 204806, "total_tasks": 49, "chunks_size": 1000, "fps": 80, "splits": { "train": "0:2500" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/metaworld_mt50.

robotwin-prefvla

Dataset from kaiwen2. Tags: modality:video, region:us.

RoboTwin-LeRobot-v3.0

Dataset from hxma. Tags: license:apache-2.0, size_categories:n<1K, modality:video, library:datasets, library:mlcroissant.

aloha_sim_transfer_cube_human

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 50, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_transfer_cube_human.

SandGO

SandGO Dataset (v1.0) SandGO 是一个专为具身智能 (Embodied AI) 和机器人学习设计的 long-horizon 多模态数据集。该数据集由原始 MarsMind_data 经过精炼、去冗余和结构化处理后得到，旨在支持 长序列决策、指令跟随 和 多模态模仿学习 等任务。数据集包含完整的状态‑动作序列、多视角 RGB 图像、LiDAR 点云、深度点云 以及对应的 自然语言指令，为训练 Vision-Language-Action (VLA) 模型提供了高质量的训练材料。 📊 数据集概览 (Dataset Summary) 项目 描述 总 episodes 约 168 条长序列轨迹 总时间步数 约 38,011 步 任务类型 巡航、避障、抓取、放置、搜索移动容器等 传感器模态 主相机 RGB、腕部相机 RGB、LiDAR 点云、深度点云、机器人状态、机器人动作 指令形式 原始自然语言文本（每元动作一条） 数据格式 NumPy 数组、JPEG 图像、JSON… See the full description on the dataset page: https://huggingface.co/datasets/XDUImageLab/SandGO.

libero_plus_lerobot

Dataset from Sylvest. Tags: license:mit, size_categories:1M<n<10M, format:parquet, modality:tabular, modality:timeseries.

loco-mujoco-datasets

Dataset from robfiras. Tags: license:cc-by-nc-nd-4.0, region:us.

aloha_sim_insertion_human

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 25000, "total_tasks": 1, "total_videos": 50, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_insertion_human.

Open-RL

Open-RL Dataset Summary This dataset contains self-contained, verifiable, and unambiguous STEM reasoning problems across Physics, Mathematics, Biology, and Chemistry. Each problem: Requires multi-step reasoning Involves symbolic manipulation and/or numerical computation Has a deterministic, objectively verifiable final answer The problems were evaluated against contemporary large language models. Observed pass rates indicate that the tasks are non-trivial yet… See the full description on the dataset page: https://huggingface.co/datasets/TuringEnterprises/Open-RL.

bridge-rlds

Dataset Structure These datasets are used for MemoryVLA training. This is the standard setting and can be directly used for other models as well.All data follow the RLDS format from the Bridge dataset. bridge_orig — 60k+ episodes, widowx robot

RoboCerebra_lerobot_v2.1

RoboCerebra (LeRobot v2.1 Format) Dataset Description This dataset is a converted version of RoboCerebra, now formatted for compatibility with the LeRobot ecosystem (v2.1). While the official RoboCerebra dataset only provides HDF5 files and RLDS conversion methods, this version enables seamless training with VLA models within the LeRobot framework. Original Dataset: RoboCerebra Conversion Tool: any4lerobot Format: LeRobot v2.1 Robot: Panda Robot (7-DOF Arm + Gripper)… See the full description on the dataset page: https://huggingface.co/datasets/CollisionCode/RoboCerebra_lerobot_v2.1.

bc_z_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "google_robot", "total_episodes": 39350, "total_frames": 5471693, "total_tasks": 104, "total_videos": 39350, "total_chunks": 40, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:39350" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/bc_z_lerobot.

SimWorld

SimWorld: An Open-ended Realistic Simulator for Autonomous Agents in Physical and Social Worlds SimWorld is a simulation platform for developing and evaluating LLM/VLM AI agents in complex physical and social environments. 🎬 Demonstration ▶ See all our demo videos on YouTube 🔥 News 2026.1 SimWorld now supports importing customized environments and agents! 2025.11 The white paper of SimWorld is… See the full description on the dataset page: https://huggingface.co/datasets/SimWorld-AI/SimWorld.

no_robots

Dataset Card for No Robots 🙅‍♂️🤖 Look Ma, an instruction dataset that wasn't generated by GPTs! Dataset Summary No Robots is a high-quality dataset of 10,000 instructions and demonstrations created by skilled human annotators. This data can be used for supervised fine-tuning (SFT) to make language models follow instructions better. No Robots was modelled after the instruction dataset described in OpenAI's InstructGPT paper, and is comprised mostly of single-turn… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceH4/no_robots.

example_data_fastumi_pro_raw

FastUMI Pro – Multimodal Sample Dataset Small-Scale Demonstration Data from the FastUMI Pro Multimodal Sensing System (Only Hundreds of Trajectories — Full Dataset Available Upon Request) Project Homepage 📦 FastUMI Data Market 🔥FastUMI Data Market is online!(click here).🔥 Link: www.fastumi.com/data-market 📖 Overview The FastUMI Pro Sample Dataset provides a public preview of the multimodal sensing capabilities of the FastUMI Pro… See the full description on the dataset page: https://huggingface.co/datasets/LumosRobotics-FastUMIPro/example_data_fastumi_pro_raw.

ManiTwin-100K

ManiTwin-100K: Manipulation-Ready Digital Object Twins Project Page | Paper ManiTwin-100K is a large-scale dataset of manipulation-ready digital object twins designed for robotic manipulation research. Each object includes simulation-ready 3D meshes, physical properties, functional point annotations, grasp configurations, and rich language descriptions—all validated through physics-based simulation. Note: We are currently releasing approximately 1K sample objects with a… See the full description on the dataset page: https://huggingface.co/datasets/Dhanush944/ManiTwin-100K.

molmobot-data

MolmoBot-data Training episode data (actions, visual inputs, and other sensor data) for 8 tasks on 2 robotic platforms: DoorOpeningDataGenConfig RBY1OpenDataGenConfig RBY1PickDataGenConfig FrankaPickOmniCamConfig RBY1PickAndPlaceDataGenConfig FrankaPickAndPlaceOmniCamConfig FrankaPickAndPlaceColorOmniCamConfig FrankaPickAndPlaceNextToOmniCamConfig Please note that every package indexed by the parquet files can contan several instances of episode data. We also provide an extension… See the full description on the dataset page: https://huggingface.co/datasets/allenai/molmobot-data.

calvin-task-ABCD-D-lerobot

Dataset from fywang. Tags: task_categories:robotics, license:mit, size_categories:1M<n<10M, format:parquet, modality:image.

3d_optical_flow_droid

3D Optical Flow DROID Dataset Processed DROID robotics dataset with optical flow and scene flow annotations. Dataset Structure Organized by lab, each trajectory in separate tar.gz archive: IPRL/IPRL+2023-06-19+Mon_Jun_19_23:27:48_2023.tar.gz CLVR/CLVR+2023-...tar.gz ... (15 labs, ~33K trajectories) Each trajectory contains: metadata.json - Trajectory metadata trajectory.h5 - Robot state and actions camera_left/, camera_right/ - Camera data rgb/ - RGB images depth/ -… See the full description on the dataset page: https://huggingface.co/datasets/Salesforce/3d_optical_flow_droid.

fmb_dataset_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "franka", "total_episodes": 8612, "total_frames": 1137459, "total_tasks": 24, "total_videos": 34448, "total_chunks": 9, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:8612" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/fmb_dataset_lerobot.

PhysicalAI-Robotics-GR00T-Teleop-Sim

Simulation GR1 Tabletop Task 1K Dataset Dataset Description: The PhysicalAI-Robotics-GR00T-Teleop-GR1 dataset consists of 1000 teleoperation trajectories in simulation using the GR1 robot with upper body control. The simulation setup mimics tabletop manipulation tasks and uses RGB observations with a virtual camera. The robot is equipped with simulated Fourier hands. This dataset is ready for non-commercial use. Dataset Owner(s): NVIDIA GEAR Dataset… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-GR00T-Teleop-Sim.

CarlaOcc

Database_structure CarlaOcc/ ├── CarlaOccV1/ │ ├── calib/ │ │ └── calib.yaml │ ├── splits/ │ │ ├── test.txt │ │ ├── train.txt │ │ └── val.txt │ ├── SceneMeshes/ │ │ ├── fg_actors/ │ │ ├── fg_actor_occ/ │ │ └── TownXX_Opt/ │ │ ├── bg_actors/ │ │ └── bg_actor_occ/ │ ├── TownXX_Opt_SeqXX/ │ │ ├── poses/ │ │ │ ├── cam_00.txt │ │ │ └── lidar.txt │ │ ├── rgb/ │ │ │ ├── image_00/ │ │ │ │ ├── 0000.png │… See the full description on the dataset page: https://huggingface.co/datasets/fengyi233/CarlaOcc.

community_dataset_v3

Lerobot Community Datasets v3 - A Cross-Embodiment Pretraining Dataset for Vision Language Action Models A large-scale robotics dataset for vision-language-action learning, featuring 791 datasets across 46 robot types, enabling cross-embodiment pretraining for generalist robot policies. Overview This is a crowdsourced, open-source dataset compiled from 235 community contributors worldwide. Building upon the pretraining datasets used for SmolVLA, Community Datasets v1… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceVLA/community_dataset_v3.

MultiAgent-Content

MultiAgent UE5 Content MultiAgent-Unreal 项目的配套 UE5 资产库。 快速克隆 #### 安装 Git LFS #### ## Mac 系统 brew install git-lfs git lfs install ## Linux 系统 sudo apt-get install git-lfs git lfs install #### 克隆到项目目录 #### cd unreal_project ## 有🪜 git clone https://huggingface.co/datasets/WindyLab/MultiAgent-Content Content ## 中国用户 git clone https://hf-mirror.com/datasets/WindyLab/MultiAgent-Content Content 📂 目录结构 Content/ ├── Agent/ # 智能体配置 ├── Characters/… See the full description on the dataset page: https://huggingface.co/datasets/WindyLab/MultiAgent-Content.

NuminaMath-1.5-RL-Verifiable

Dataset Card for NuminaMath-1.5-RL-Verifiable Dataset Summary NuminaMath-1.5-RL-Verifiable is a curated subset of the NuminaMath-1.5 dataset, specifically filtered to support reinforcement learning applications requiring verifiable outcomes. This collection consists of 131,063 math word problems from the original dataset that meet strict filtering criteria: all problems have definitive numerical answers, validated problem statements and solutions, and come from… See the full description on the dataset page: https://huggingface.co/datasets/nlile/NuminaMath-1.5-RL-Verifiable.

xarm_lift_medium

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 800, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_lift_medium.

PhysicalAI-Robotics-Manipulation-SingleArm

Dataset Description: PhysicalAI-Robotics-Manipulation-SingeArm is a collection of datasets of automatic generated motions of a Franka Panda robot performing operations such as block stacking, opening cabinets and drawers. The dataset was generated in IsaacSim leveraging task and motion planning algorithms to find solutions to the tasks automatically [1, 3]. The environments are table-top scenes where the object layouts and asset textures are procedurally generated [2].This dataset… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-Manipulation-SingleArm.

jat-dataset

JAT Dataset Dataset Description The Jack of All Trades (JAT) dataset combines a wide range of individual datasets. It includes expert demonstrations by expert RL agents, image and caption pairs, textual data and more. The JAT dataset is part of the JAT project, which aims to build a multimodal generalist agent. Paper: https://huggingface.co/papers/2402.09844 Usage >>> from datasets import load_dataset >>> dataset = load_dataset("jat-project/jat-dataset"… See the full description on the dataset page: https://huggingface.co/datasets/jobs-git/jat-dataset.

libero-assets

Dataset from lerobot. Tags: size_categories:n<1K, format:imagefolder, modality:3d, modality:image, library:datasets.

climbing-holds-v2

Dataset from rlogh. Tags: region:us.

aloha_sim_insertion_scripted

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 50, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_insertion_scripted.

rbo_oxe_base_language_table_lerobot

Language Table (LeRobot) — Task-Pruned, Reindexed Subset This release is a task-pruned subset of the original IPEC-COMMUNITY/language_table_lerobot. We subsampled by task text and rebuilt the package so it remains internally consistent (indices, splits, stats, paths). Robot: xArm Modality: RGB video + states + actions FPS / Resolution: 10 FPS, 360×640, AV1 License: apache-2.0 (inherits from source) What’s different in this subset Kept ~0.85% of unique tasks… See the full description on the dataset page: https://huggingface.co/datasets/saaduddinM/rbo_oxe_base_language_table_lerobot.

AAWR-DROID

Active Perception AAWR Dataset in GRASP Lab Mock Kitchen Paper | Project page | Code Here is our full dataset (13.5GB) in DROID format: https://huggingface.co/datasets/Everloom/AAWR-DROID/ It includes 4 scene as below, each are cleaned with a white list (filter the idle frames) Sample Usage To get started with AAWR, follow these steps for a toy simulated task, as described in the code repository: Setup Install Python dependencies: You can refer to the… See the full description on the dataset page: https://huggingface.co/datasets/Everloom/AAWR-DROID.

gm100-cobotmagic-lerobot

Dataset Card for GM-100 Cobot Magic Part (Lerobot2.1 Format) This is a part of the Great March 100 (GM-100) Project. Raw teleoperation data collected using the Agilex Cobot Magic robot. This dataset is stored in Lerobot2.1 format. Notice This is a preview version of this dataset. A few task prompts may contain minor issues, which we will address in the upcoming full release. We welcome any feedback or suggestions you may have. Dataset Description This… See the full description on the dataset page: https://huggingface.co/datasets/rhos-ai/gm100-cobotmagic-lerobot.

Madison-Driving-Dataset-R1

!!Unfinished Project!! Will upload more data in the future Route-1 Driving Dataset Real-world vehicle driving dataset collected on 2025-09-30, containing front-camera frame-level driving signals and 3-second future trajectory labels. Dataset Description Each record corresponds to a single front-camera frame and includes: Driving signals: speed, throttle, brake pressure, steering wheel angle Future trajectory: 6 future poses over a 3-second horizon (sampled at 0.5 s… See the full description on the dataset page: https://huggingface.co/datasets/Young1219/Madison-Driving-Dataset-R1.

ZwZ-RL-VQA-2k

Dataset from muyuho. Tags: size_categories:n<1K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

RoboGene

Dataset Card for RoboGene-Dataset Figure 1: Overview of the RoboGene dataset. RoboGene is a large-scale robotic task dataset generated by a diversity-driven agentic framework. Specifically designed for Vision-Language-Action (VLA) model pre-training, it addresses the critical bottlenecks of limited scene variety and insufficient physical grounding in existing datasets, significantly enhancing the generalization capabilities of humanoid robots in complex real-world… See the full description on the dataset page: https://huggingface.co/datasets/X-Humanoid/RoboGene.

humanoid-everyday

Humanoid Everyday A Comprehensive Robotic Dataset for Open-World Humanoid Manipulation Overview Humanoid Everyday is a large-scale, diverse humanoid manipulation dataset designed for open-world robotic learning and embodied intelligence. It contains over 260 tasks across 7 major categories, covering dexterous manipulation, human–humanoid interaction, and locomotion-integrated activities.All data were collected through a human-supervised teleoperation pipeline, recording… See the full description on the dataset page: https://huggingface.co/datasets/FedorX8/humanoid-everyday.

RoboChallenge-lerobot

RoboChallenge-lerobot Note: The original dataset comes from RoboChallenge/Table30. This is an unofficial conversion to LeRobot v3.0 format. Dataset Description RoboChallenge benchmark dataset in LeRobot v3.0 format. Contains 26 manipulation tasks with 22,111 episodes and 29,162,844 frames in total. Each subtask is a separate config (subset). To load a specific subtask: from lerobot.common.datasets.lerobot_dataset import LeRobotDataset dataset =… See the full description on the dataset page: https://huggingface.co/datasets/Traly/RoboChallenge-lerobot.

android-control

Dataset from smolagents. Tags: size_categories:10K<n<100K, format:parquet, modality:text, library:datasets, library:dask.

pusht_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 206, "total_frames": 25650, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:206" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/pusht_image.

jsi_franka_peg_in_hole_v12

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "robotblockset_robot", "total_episodes": 2308, "total_frames": 106364, "total_tasks": 30, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 15, "splits": { "train": "0:2308" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/bkuster/jsi_franka_peg_in_hole_v12.

Reagent-RL-709K

Official Repo of Reagent Agent RL training dataset (Reagent-RL-709K). Paper: https://arxiv.org/abs/2601.22154 Abstract: Agentic Reinforcement Learning (Agentic RL) has achieved notable success in enabling agents to perform complex reasoning and tool use. However, most methods still relies on sparse outcome-based reward for training. Such feedback fails to differentiate intermediate reasoning quality, leading to suboptimal training results. In this paper, we introduce… See the full description on the dataset page: https://huggingface.co/datasets/bunny127/Reagent-RL-709K.

fluidgym-experiments

FluidGym Experiments Paper | GitHub | Documentation FluidGym is a standalone, fully differentiable benchmark suite for reinforcement learning (RL) in active flow control (AFC). Built entirely in PyTorch on top of the GPU-accelerated PICT solver, it provides standardized evaluation protocols and diverse environments for systematic comparison of control methods. This repository contains the training and test datasets with results for all experimental runs presented in the paper.… See the full description on the dataset page: https://huggingface.co/datasets/safe-autonomous-systems/fluidgym-experiments.

Reverse-Text-RL

Reverse-Text-RL A small, scrappy RL dataset used in prime-rl's CI to debug RL training asking a model to reverse small sentences character-by-character. Follows the general format of PrimeIntellect/Reverse-Text-SFT The following script was used to generate the dataset. from datasets import Dataset, load_dataset dataset = load_dataset("willcb/R1-reverse-wikipedia-paragraphs-v1-1000", split="train") prompt = "Reverse the text character-by-character. Put your answer in <reversed_text>… See the full description on the dataset page: https://huggingface.co/datasets/PrimeIntellect/Reverse-Text-RL.

molmospaces

MolmoSpaces This respository contains asset data for MolmoSpaces, including Objects Robots Scenes Grasps Benchmarks Updates [2026/02/16] - Isaac-compatible USD objects and scenes now also available! Downloading We recommend using the download.py script to list available data sources and extract data from one or all sources to a local cache directory. To use the script you'll need a few dependecies in your Python environment that can be installed, e.g., by:… See the full description on the dataset page: https://huggingface.co/datasets/allenai/molmospaces.

submissions

Dataset from LeRobot-worldwide-hackathon. Tags: license:cc-by-sa-4.0, size_categories:n<1K, modality:video, library:datasets, library:mlcroissant.

aloha_sim_transfer_cube_scripted

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4", "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_transfer_cube_scripted.

calvin-task-ABC-D-lerobot

Dataset from fywang. Tags: size_categories:1M<n<10M, modality:image, modality:timeseries, region:us.

syn_realDataset

Capstone VLA Datasets LeRobot v2.1 datasets for a liquid pouring task on a KuavoV4Pro humanoid robot (34-DOF, bimanual). Used to benchmark GR00T N1.6, pi0.5, and Diffusion Policy under varying real-to-synthetic data ratios. Mixture Configurations Dataset Type Episodes sim_mixed_40_60 unknown 500 Dataset Structure (LeRobot v2.1) <dataset>/ ├── meta/ │ ├── info.json │ ├── tasks.jsonl │ ├── episodes.jsonl │ └── stats.safetensors ├── data/ │… See the full description on the dataset page: https://huggingface.co/datasets/Lusmse/syn_realDataset.

spacetools-sft

SpaceTools SFT Dataset Supervised fine-tuning data for spatial reasoning with vision tool calling. Dataset Summary Train: 7,907 samples (V2 pipeline: 11 reasoning tools + 5 robot tools) Format: JSON with relative image paths Images: Stored in images/a/ and images/b/ directories Note: The HuggingFace viewer may show fewer rows due to preview limits on JSON files. The full dataset contains 7,907 samples. Use load_dataset() or download directly to access all rows.… See the full description on the dataset page: https://huggingface.co/datasets/siyich/spacetools-sft.

RLCo-Example-Mix-Data

Dataset from RLinf. Tags: license:artistic-2.0, size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant.

level12_rac_2_2026-02-07_and_mir

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 10770, "total_frames": 26748966, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:10770" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot-data-collection/level12_rac_2_2026-02-07_and_mir.

HelpSteer3

HelpSteer3 HelpSteer3 is an open-source dataset (CC-BY-4.0) that supports aligning models to become more helpful in responding to user prompts. HelpSteer3-Preference can be used to train Llama 3.3 Nemotron Super 49B v1 (for Generative RMs) and Llama 3.3 70B Instruct Models (for Bradley-Terry RMs) to produce Reward Models that score as high as 85.5% on RM-Bench and 78.6% on JudgeBench, which substantially surpass existing Reward Models on these benchmarks. HelpSteer3-Feedback and… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/HelpSteer3.

calvin_lerobot

Dataset from JensenYuan. Tags: size_categories:1M<n<10M, format:parquet, modality:image, modality:timeseries, library:datasets.

Big-Math-RL-Verified

Big-Math: A Large-Scale, High-Quality Math Dataset for Reinforcement Learning in Language Models Big-Math is the largest open-source dataset of high-quality mathematical problems, curated specifically for reinforcement learning (RL) training in language models. With over 250,000 rigorously filtered and verified problems, Big-Math bridges the gap between quality and quantity, establishing a robust foundation for advancing reasoning in LLMs. Request Early Access to Private… See the full description on the dataset page: https://huggingface.co/datasets/SynthLabsAI/Big-Math-RL-Verified.

visual_distracting_metaworld

Empty Dataset

trace_robotwin_rand

Dataset from quincyu. Tags: size_categories:1M<n<10M, format:parquet, format:optimized-parquet, modality:image, modality:timeseries.

SimulationImage

Dataset from TaxonomyProject. Tags: region:us.

peract_rlbench_640x480

Dataset from vrlfdvla. Tags: region:us.

libero_spatial_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "panda", "total_episodes": 432, "total_frames": 52970, "total_tasks": 10, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:432" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/libero_spatial_image.

tatvton-tattoo-raw

Dataset from rlaope. Tags: size_categories:1K<n<10K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

libero_90_no_noops_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "franka", "total_episodes": 3921, "total_frames": 569249, "total_tasks": 73, "total_videos": 7842, "total_chunks": 3, "chunks_size": 1000, "fps": 20, "splits": { "train": "0:3921" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/libero_90_no_noops_lerobot.

libero

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "panda", "total_episodes": 1693, "total_frames": 273465, "total_tasks":40, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 500, "fps": 10.0, "splits": { "train": "0:1693" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/libero.

aloha_sim_transfer_cube_human_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_transfer_cube_human_image.

D4RL

Dataset from farama-minari. Tags: region:us.

aloha_sim_insertion_human_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 25000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_insertion_human_image.

MOSAIC_Dataset

MOSAIC Dataset This repository releases the built-in MOSAIC multi-source motion dataset in the following paper: MOSAIC: Bridging the Sim-to-Real Gap in Generalist Humanoid Motion Tracking and Teleoperation with Rapid Residual Adaptation The dataset is organized into: Human motions stored in an AMASS-style format Unitree G1 motions retargeted from human motions and converted to NPZ for training/visualization It includes motions from: Optical MoCap (Vicon) Inertial MoCap (IO-AI)… See the full description on the dataset page: https://huggingface.co/datasets/chuhonglin/MOSAIC_Dataset.

aloha_sim_insertion_scripted_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_insertion_scripted_image.

merged_LIBERO_lerobot_v2_IPEC_COMMUNITY

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "franka", "total_episodes": 1693, "total_frames": 273465, "total_tasks": 40, "total_videos": 3386, "total_chunks": 2, "chunks_size": 1000, "fps": 20, "splits": { "train": "0:1693"}, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/binhng/merged_LIBERO_lerobot_v2_IPEC_COMMUNITY.

calvin_abc_d_lerobot_v2.1

Calvin ABC D (LeRobot v2.1 Format) Dataset Description This dataset is a converted version of clavin_abc_d.zip, now formatted for compatibility with the starVLA framework. While the official Calvin dataset only provides zip files and HDF5 conversion methods, this version enables seamless training with modern imitation learning pipelines like Diffusion Policy, ACT, and VLA models within the LeRobot framework. Original Dataset: Calvin Conversion Tool: RoboTron-Mani… See the full description on the dataset page: https://huggingface.co/datasets/CollisionCode/calvin_abc_d_lerobot_v2.1.

robomme_data_lerobot

RoboMME Training Data (LeRobot Format) Arxiv Paper | HF Paper | Website | Benchmark Code | Policy Learning Code This repo contains Lerobot format of RoboMME training data. We use this in our Diffusion Policy training code

vlabench_primitive_ft_lerobot

Dataset from VLABench. Tags: size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries, library:datasets.

control-arena-logs-20260315-sonnet45-maint-he-pos-download

logs/2026-03-15 This dataset is a direct upload of the local directory logs/2026-03-15 from the control-arena-persistent-state-eval workspace. It contains raw run outputs and batch-level summaries for a March 2026 set of maintenance-task persistent-state evaluations using Claude Sonnet 4.5 as the untrusted agent and GPT-4o as the trusted monitor model. What Is Included The upload contains two experiment bundles: config_telemetry_maint_sonnet45_he_pos_download/… See the full description on the dataset page: https://huggingface.co/datasets/idacy/control-arena-logs-20260315-sonnet45-maint-he-pos-download.

labelled_robotwin_aloha-agilex_clean_50

Dataset from JimmyH16. Tags: region:us.

aloha_sim_transfer_cube_scripted_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_sim_transfer_cube_scripted_image.

InternData-A1-LeRobot-v3.0-by-embodiment

InternData-A1 dataset taken from InternRobotics/InternData-A1, with the tarballs extracted and directory structure "transposed" so that the top-level subdirectories are the four embodiments. Two franka dirs For the franka embodiment, there are two different feature spaces, so we split it into the franka-1 and franka-2 directories. The feature spaces differ in image shape and gripper value range. Minor fixes Some subsets such as… See the full description on the dataset page: https://huggingface.co/datasets/yixuan-tan/InternData-A1-LeRobot-v3.0-by-embodiment.

DrivingGen

DrivingGen: A Comprehensive Benchmark for Generative Video World Models in Autonomous Driving DrivingGen: A Comprehensive Benchmark for Generative Video World Models in Autonomous Driving Yang Zhou*, Hao Shao*, Letian Wang*, Zhuofan Zong, Hongsheng Li, Steven Waslander ("*" denotes equal contribution) @misc{zhou2026drivinggencomprehensivebenchmarkgenerative, title={DrivingGen: A Comprehensive Benchmark for Generative Video World… See the full description on the dataset page: https://huggingface.co/datasets/yangzhou99/DrivingGen.

cmu-fbx

CMU Motion Capture Library — FBX Format A large-scale library of 2,548 human motion capture animations in .fbx format, organized by subject/motion number. This dataset is derived from the CMU Graphics Lab Motion Capture Database and is intended for use in robotics, computer animation, game development, and machine learning research involving human motion. Source page: https://rancidmilk.itch.io/free-character-animations Dataset Summary This dataset contains over 2… See the full description on the dataset page: https://huggingface.co/datasets/gbionics/cmu-fbx.

robotwin_eval_results

Dataset from RogersPyke. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

uav-drone-detections

Dataset from rushilara. Tags: region:us.

libero_10_no_noops_1.0.0_lerobot

Dataset from IPEC-COMMUNITY. Tags: size_categories:100K<n<1M, format:parquet, modality:tabular, modality:timeseries, modality:video.

vlabench_composite_ft_lerobot_video

Dataset from VLABench. Tags: license:apache-2.0, size_categories:1M<n<10M, format:parquet, modality:tabular, modality:text.

level12_rac_2_2026-02-07

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 5551, "total_frames": 13805156, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:5551" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot-data-collection/level12_rac_2_2026-02-07.

vidar-robotwin-data

Dataset from VincentNi. Tags: region:us.

scenic_driving_scenarios

Dataset from Divyanshudiv. Tags: modality:image, region:us.

mujoco

Dataset from farama-minari. Tags: region:us.

atari

Dataset from farama-minari. Tags: region:us.

metaworld_ml45-v2

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "metaworld", "total_episodes": 4391, "total_frames": 358997, "total_tasks": 44, "total_videos": 0, "total_chunks": 5, "chunks_size": 1000, "fps": 80, "splits": { "train": "0:4391" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/brandonyang/metaworld_ml45-v2.

vlabench_primitive_ft_lerobot_video

Dataset from VLABench. Tags: license:apache-2.0, size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant.

drone_to_table

Dataset from Celina717. Tags: size_categories:1K<n<10K, modality:text, modality:video, library:datasets, library:mlcroissant.

libero_spatial_no_noops_1.0.0_lerobot

Dataset from IPEC-COMMUNITY. Tags: size_categories:10K<n<100K, format:parquet, modality:tabular, modality:timeseries, modality:video.

RoboChallenge-lerobot-merged

RoboChallenge-lerobot-merged Note: The original dataset comes from RoboChallenge/Table30. This is an unofficial conversion to LeRobot v3.0 format. Dataset Description RoboChallenge benchmark dataset (merged version) in LeRobot v3.0 format. All 26 tasks are merged into a single dataset with 22,111 episodes and 29,162,844 frames. For the per-task (multi-subset) version, see Traly/RoboChallenge-lerobot. Dataset Structure ├── data/ │ └── chunk-{000..N}/ │… See the full description on the dataset page: https://huggingface.co/datasets/Traly/RoboChallenge-lerobot-merged.

houdini-lego-sdg

Houdini Lego SDG Dataset Synthetic datasets generated by a procedural Houdini-based data generation pipeline for Lego brick instance segmentation and object detection. All images were rendered using Karma XPU in Houdini Solaris. Annotations were extracted automatically using Cryptomatte AOVs with no manual labeling on training data. GitHub: nathankimnguyen412/houdini-sdg-pipeline Dataset Versions Version Images Classes Description v1 1,000 1 Baseline: no… See the full description on the dataset page: https://huggingface.co/datasets/nathankimnguyen412/houdini-lego-sdg.

metaworld_ml45

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "metaworld", "total_episodes": 24, "total_frames": 1707, "total_tasks": 6, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 80, "splits": { "train": "0:24" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/brandonyang/metaworld_ml45.

interactive-world-sim-mujoco-data

Dataset from yixuan1999. Tags: size_categories:1M<n<10M, modality:timeseries, library:datasets, library:mlcroissant, region:us.

HPO-RL-Tiny

HPO-RL-Bench Minimal Subset (100 configs per pair of algorithm and environment) ⚠️ This is a derivative subset of the official HPO-RL-Bench benchmark. It contains 100 randomly sampled hyperparameter configurations for each of the three core algorithms (DQN, PPO, SAC) to enable rapid prototyping and testing of hyperparameter optimization methods. For the full benchmark , see the original dataset. 📌 Attribution & License This dataset is a derivative work of:… See the full description on the dataset page: https://huggingface.co/datasets/Nicknam/HPO-RL-Tiny.

BiPlay

BiPlay contains 9.7 hours of bimanual data collected with an aloha robot at the RAIL lab @ UC Berkeley, USA. It contains 7023 clips, 2000 language annotations and 326 unique scenes. Paper: https://huggingface.co/papers/2410.10088 Code: https://github.com/sudeepdasari/dit-policy If you use the dataset please cite: @inproceedings{dasari2025ingredients, title={The Ingredients for Robotic Diffusion Transformers}, author={Sudeep Dasari and Oier Mees and Sebastian Zhao and Mohan Kumar Srirama… See the full description on the dataset page: https://huggingface.co/datasets/oier-mees/BiPlay.

workshop_franka_gr00t_1000

Dataset from umegan. Tags: license:apache-2.0, size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant.

metaworld

Dataset from farama-minari. Tags: region:us.

vidar_robotwin50

Dataset from VincentNi. Tags: region:us.

Nemotron-RL-bixbench_hypothesis

Dataset Description: This dataset, Nemotron-RL-bixbench_hypothesis also referred to as BixBench-Hypothesis-Train (BBH-Train) is a training dataset paired to the BixBench-Hypothesis benchmark created by Edison Scientific. The goal of BBH-train is to improve an agentic system’s ability to accurately evaluate biological hypotheses by means of performing bioinformatics data analysis. Each instance in the dataset contains: A self-contained biological hypothesis A biological dataset with… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-bixbench_hypothesis.

calvin_abc_rlds

Dataset from zhouhongyi. Tags: region:us.

AgiBotWorld-Alpha-lerobot

Note This is an unofficial implementation. This dataset was converted from AgiBotWorld-Alpha into the LeRobot format using https://github.com/Tavish9/any4lerobot/tree/main/agibot2lerobot Configurations Each sub-task is available as a separate configuration. You can load them using load_dataset: from datasets import load_dataset # Load a specific task ds = load_dataset("Traly/AgiBotWorld-Alpha-lerobot", "task_327") Dataset Structure Each task folder… See the full description on the dataset page: https://huggingface.co/datasets/Traly/AgiBotWorld-Alpha-lerobot.

techno_dataset

Dataset from GAIA-Grand-defi-robotique-agricole. Tags: size_categories:n<1K, format:json, modality:text, library:datasets, library:dask.

OmniVitac

Dataset from tars-robotics. Tags: license:cc-by-nc-4.0, region:us.

Japan-Open-Driving-Dataset-Sample

Japan Open Driving Dataset Sample Overview This repository contains a sample subset of the Japan Open Driving Dataset, a large-scale autonomous driving dataset comprising over 100 hours of driving data collected in Tokyo, Japan. The data is stored in nuScenes format and can be loaded with the nuscenes-devkit. In addition to sensor data and 3D annotations, this dataset includes virtual captioned data for training Vision-Language-Model (VLM) and Vision-Language-Action (VLA)… See the full description on the dataset page: https://huggingface.co/datasets/turing-motors/Japan-Open-Driving-Dataset-Sample.

bridge_v2_lerobot_pathmask

PEEK VLM-Labeled BRIDGE_v2 dataset This dataset contains the LeRobot-format BRIDGE-v2 dataset with paths and masks from the PEEK VLM drawn onto the image: PEEK: Guiding and Minimal Image Representations for Zero-Shot Generalization of Robot Manipulation Policies. PEEK fine-tunes Vision-Language Models (VLMs) to predict a unified point-based intermediate representation for robot manipulation. This representation consists of: End-effector paths: specifying what actions to take.… See the full description on the dataset page: https://huggingface.co/datasets/jesbu1/bridge_v2_lerobot_pathmask.

RoboTwin

robosuite [Homepage]   [White Paper]   [Documentations]   [ARISE Initiative] Latest Updates [10/28/2024] v1.5: Added support for diverse robot embodiments (including humanoids), custom robot composition, composite controllers (including whole body controllers), more teleoperation devices, photo-realistic rendering. [release notes] [documentation] [11/15/2022] v1.4: Backend migration to DeepMind's official MuJoCo Python binding, robot textures, and bug fixes… See the full description on the dataset page: https://huggingface.co/datasets/Fxxkrobotics/RoboTwin.

benchgen_lerobot_dataset_filtered

Dataset from Genesis-Intelligence. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

xarm_push_medium_replay

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 800, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_push_medium_replay.

lerobotv2_bridge_ecot

Dataset from matCercola18. Tags: size_categories:100K<n<1M, format:parquet, modality:tabular, modality:text, modality:timeseries.

dota

DOTA dataset collection. Image Source and Usage License The DOTA images are collected from the Google Earth, GF-2 and JL-1 satellite provided by the China Centre for Resources Satellite Data and Application, and aerial images provided by CycloMedia B.V. DOTA consists of RGB images and grayscale images. The RGB images are from Google Earth and CycloMedia, while the grayscale images are from the panchromatic band of GF-2 and JL-1 satellite images. All the images are stored in 'png'… See the full description on the dataset page: https://huggingface.co/datasets/isaaccorley/dota.

libero_goal_no_noops_1.0.0_lerobot

Dataset from IPEC-COMMUNITY. Tags: size_categories:10K<n<100K, modality:tabular, modality:timeseries, modality:video, region:us.

libero_object_no_noops_1.0.0_lerobot

Dataset from IPEC-COMMUNITY. Tags: size_categories:10K<n<100K, format:parquet, modality:tabular, modality:timeseries, modality:video.

robotwin

Dataset from Joe-HPN. Tags: size_categories:1M<n<10M, format:parquet, modality:image, modality:text, modality:timeseries.

robotwin2.0-10task-motus_converted

Dataset from jiajun0613. Tags: size_categories:n<1K, modality:text, modality:video, library:datasets, library:mlcroissant.

robotwin-icl-paired-v2

Dataset from yeeeiii111. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

full_folding

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 5688, "total_frames": 14129038, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:5688"}, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/full_folding.

Humanoid-Everyday-H1

Dataset from USC-PSI-Lab. Tags: size_categories:1M<n<10M, format:parquet, format:optimized-parquet, modality:tabular, modality:timeseries.

VLA_Arena_L0_L_lerobot_openpi

VLA-Arena Dataset (L0 - Large Variant) About VLA-Arena VLA-Arena is an open-source benchmark designed for the systematic evaluation of Vision-Language-Action (VLA) models. It provides a complete and unified toolchain covering scene modeling, demonstration collection, model training, and evaluation. Featuring 150+ tasks across 11 specialized suites, VLA-Arena assesses models through hierarchical difficulty levels (L0-L2) to ensure comprehensive metrics for safety… See the full description on the dataset page: https://huggingface.co/datasets/VLA-Arena/VLA_Arena_L0_L_lerobot_openpi.

level12_rac_2_2026-02-05_ag

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 5413, "total_frames": 13497374, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:5413" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot-data-collection/level12_rac_2_2026-02-05_ag.

racing_trajectory_prediction

Dataset from deepracing. Tags: license:cc-by-4.0, region:us.

agro_dataset

Dataset from GAIA-Grand-defi-robotique-agricole. Tags: region:us.

G1_Dex1_Clean_Table

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Clean_Table.

Qwen-RL

Dataset from Amshaker. Tags: size_categories:n<1K, modality:tabular, modality:text, region:us.

lingbot_robotool_knife_spread_tomatosauce_mano_all_lerobot

Dataset from kkkkhanl. Tags: size_categories:n<1K, modality:video, library:datasets, library:mlcroissant, region:us.

xarm_lift_medium_replay

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 800, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_lift_medium_replay.

maniskill_assets

Asset Download The assets need to be placed into RLinf's ManiSkill environment folder with the name assets. # uv pip install huggingface_hub if you don't have it cd <path_to_RLinf>/rlinf/envs/maniskill hf download --repo-type dataset RLinf/maniskill_assets --local-dir ./assets You can also use git to clone the repository: cd <path_to_RLinf>/rlinf/envs/maniskill git clone https://hf.co/datasets/RLinf/maniskill_assets ./assets License Our assets are attributed to… See the full description on the dataset page: https://huggingface.co/datasets/RLinf/maniskill_assets.

Humanoid-Everyday-G1

Dataset from USC-PSI-Lab. Tags: size_categories:1M<n<10M, format:parquet, format:optimized-parquet, modality:tabular, modality:timeseries.

leju_robot_moving_parts_r

leju_robot_moving_parts_r 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 495… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_r.

lunarsim-shoemaker-traverse-100m-v1

LunarSim Rover Traverse Synthetic Dataset V1 Overview Synthetic lunar surface dataset generated using LunarSim, an NVIDIA Isaac Sim-based lunar environment simulator. Images simulate a forward-facing rover camera traversing terrain near the Shoemaker crater at the lunar south pole. The dataset includes RGB images, depth maps, semantic/instance segmentation masks, shadow masks, and COCO-format rock detection annotations. Generation Simulator: LunarSim (NVIDIA… See the full description on the dataset page: https://huggingface.co/datasets/elementrobotics/lunarsim-shoemaker-traverse-100m-v1.

CoT_length_RL

Dataset from lucferon. Tags: region:us.

rocochallenge2025

Dataset Description This dataset serves for the roco challenge @AAAI26, focusing on collaborative gearbox assembly scenarios in human-centric manufacturing. It contains demonstrations for the gearbox assembly task, conducted with Galaxea R1 robot in IsaacSim environment. The code for data collection is given here. Homepage: https://rocochallenge.github.io/RoCo2026/ Dataset Structure The dataset contains RGBD observation from 3 views (head, left hand, right hand)… See the full description on the dataset page: https://huggingface.co/datasets/rocochallenge2025/rocochallenge2025.

level12_rac_2_2026-02-04-mirror

Dataset from lerobot-data-collection. Tags: region:us.

VLA_Arena_L1_L_lerobot_openpi

VLA-Arena Dataset (L1 - Large Variant) About VLA-Arena VLA-Arena is an open-source benchmark designed for the systematic evaluation of Vision-Language-Action (VLA) models. It provides a complete and unified toolchain covering scene modeling, demonstration collection, model training, and evaluation. Featuring 150+ tasks across 11 specialized suites, VLA-Arena assesses models through hierarchical difficulty levels (L0-L2) to ensure comprehensive metrics for safety… See the full description on the dataset page: https://huggingface.co/datasets/VLA-Arena/VLA_Arena_L1_L_lerobot_openpi.

egodex_rlds

Dataset from CometsFeiyu. Tags: region:us.

dronescapes2_annotated_train_set

Dataset Card for DroneScapes2 (annotated train set) This is a FiftyOne dataset with 218 samples. It's a subset of this split from the original repo. Installation If you haven't already, install FiftyOne: pip install -U fiftyone Usage import fiftyone as fo from fiftyone.utils.huggingface import load_from_hub # Load the dataset # Note: other available arguments include 'max_samples', etc dataset =… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/dronescapes2_annotated_train_set.

EnvShip-Bench_An_Environment-Enhanced_Benchmark_for_Short-Term_Vessel_Trajectory_Prediction

EnvShip-Bench EnvShip-Bench is a benchmark for short-term vessel trajectory prediction built from raw AIS data released by the Danish Maritime Authority (DMA). This Hugging Face release currently provides two source branches under a shared layout: DMA/ NOAA/ The DMA branch is organized under: DMA/benchmark/core/ DMA/benchmark/full/ DMA/mini_benchmark/ship_core_lite/ DMA/mini_benchmark/clean_ship_core_lite_v1/ The NOAA branch is organized under: NOAA/benchmark/core/… See the full description on the dataset page: https://huggingface.co/datasets/mark000071/EnvShip-Bench_An_Environment-Enhanced_Benchmark_for_Short-Term_Vessel_Trajectory_Prediction.

AirScape-Dataset

[ACM MM'25] AirScape: An Aerial Generative World Model with Motion Controllability This repository contains the dataset introduced in the paper, consisting of two parts: 11k+ motion intention prompts and corresponding video clips. Arxiv: https://arxiv.org/pdf/2507.08885 Project: https://embodiedcity.github.io/AirScape/ Code: https://github.com/EmbodiedCity/AirScape.code Dataset Description This dataset is proposed for training and testing of aerial world models.… See the full description on the dataset page: https://huggingface.co/datasets/EmbodiedCity/AirScape-Dataset.

droid_v3

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "Franka", "total_episodes": 95658, "total_frames": 27630375, "total_tasks": 49630, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 15, "splits": { "train": "0:95658"}, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/DAVIAN-Robotics/droid_v3.

xarm_push_medium_replay_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_push_medium_replay_image.

level12_rac_2_2026-02-04

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 5357, "total_frames": 13251592, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:5357" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot-data-collection/level12_rac_2_2026-02-04.

inventory_control_benchmark_instances

Dataset from Yaopeng12138. Tags: license:mit, region:us.

community_dataset_v2

Community Dataset v2 A large-scale community-contributed robotics dataset for vision-language-action learning, featuring 340 datasets from 117 contributors worldwide. This dataset represents the second major release of community-contributed robotics data, expanding upon the Community Dataset v1. 🌟 Overview This dataset represents a collaborative effort from the robotics and AI community to build comprehensive training data for embodied AI systems. Each contribution… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceVLA/community_dataset_v2.

UAV-IndoorCL

UAV-IndoorCL This repository contains the dataset presented in the paper Learning on the Fly: Replay-Based Continual Object Perception for Indoor Drones. Project Page | GitHub Dataset Summary UAV-IndoorCL is an indoor video dataset consisting of 14,400 frames capturing inter-drone and ground vehicle footage. It was specifically designed to support and benchmark Class-Incremental Learning (CIL) research for resource-constrained aerial platforms. The frames were… See the full description on the dataset page: https://huggingface.co/datasets/kaochuang/UAV-IndoorCL.

open-australian-legal-corpus

Open Australian Legal Corpus ‍⚖️ The Open Australian Legal Corpus by Isaacus, a foundational legal AI research company, is the first and only multijurisdictional open corpus of Australian legislative and judicial documents. Comprised of 229,122 texts totalling over 60 million lines and 1.4 billion tokens, the Corpus includes every in force statute and regulation in the Commonwealth, New South Wales, Queensland, Western Australia, South Australia, Tasmania and Norfolk Island, in… See the full description on the dataset page: https://huggingface.co/datasets/isaacus/open-australian-legal-corpus.

xarm_push_medium

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 800, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_push_medium.

memorybench-lerobot

Dataset from memory-vla. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

INTELLECT-3-RL

Dataset from PrimeIntellect. Tags: size_categories:10K<n<100K, format:parquet, format:optimized-parquet, modality:tabular, modality:text.

Downstream_Physics_Simulation

GeoPT Project Page | Paper | GitHub This repository contains the physics simulation data for the paper GeoPT: Scaling Physics Simulation via Lifted Geometric Pre-Training. GeoPT is a unified model pre-trained on large-scale geometric data for general physics simulation, unlocking a scalable path for neural simulation. Overview GeoPT is evaluated on the following five simulation tasks. Dataset Mesh Size Variable Training Test Total Size Source DrivAerML… See the full description on the dataset page: https://huggingface.co/datasets/GeoPT/Downstream_Physics_Simulation.

Assignment3_Drone_Detections

Dataset from sohaibchachar. Tags: size_categories:1K<n<10K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

dronedata_new

Dataset from Celina717. Tags: region:us.

code-verl-unified

Unified Code VERL Dataset Overview This dataset aggregates seven code-reasoning collections into a single VERL-formatted repository containing approximately 958,539 unique problems. The compilation prioritizes consistent extra_info structure across all source materials for seamless compatibility with VERL training frameworks. Dataset Composition Seven distinct splits comprise the collection: Split Problems Percentage Format kodcode_v1_verl 434,876… See the full description on the dataset page: https://huggingface.co/datasets/sungyub/code-verl-unified.

pi05-metaworld-activations-v2-15env

Dataset from brandonyang. Tags: region:us.

G1_Dex1_DualRobot_Clean_Table

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_DualRobot_Clean_Table.

leju_robot_moving_parts_q

leju_robot_moving_parts_q 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 494… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_q.

AV-Deepfake1M-PlusPlus

AV-Deepfake1M++ The dataset used for the 2025 1M-Deepfakes Detection Challenge. Task 1 Video-Level Deepfake Detection: Given an audio-visual sample containing a single speaker, the task is to identify if the video is a deepfake or real. Task 2 Deepfake Temporal Localization: Given an audio-visual sample containing a single speaker, the task is to find out the timestamps [start, end] in which the manipulation is done. The assumption here is that from the perspective of spreading… See the full description on the dataset page: https://huggingface.co/datasets/ControlNet/AV-Deepfake1M-PlusPlus.

xarm_push_medium_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_push_medium_image.

berkeley_cable_routing

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "unknown", "total_episodes": 1647, "total_frames": 42328, "total_tasks": 1, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:1647" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/berkeley_cable_routing.

merged_LIBERO_lerobot_v2_IPEC_COMMUNITY_v2

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "franka", "total_episodes": 1693, "total_frames": 273465, "total_tasks": 40, "total_videos": 3386, "total_chunks": 2, "chunks_size": 1000, "fps": 20, "splits": { "train": "0:1693"}, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/binhng/merged_LIBERO_lerobot_v2_IPEC_COMMUNITY_v2.

How_RL

Dataset from Fengzhuo. Tags: region:us.

UrbanVideo-Bench

[ACL'25 Oral] UrbanVideo-Bench: Benchmarking Vision-Language Models on Embodied Intelligence with Video Data in Urban Spaces This repository contains the dataset introduced in the paper, consisting of two parts: 5k+ multiple-choice question-answering (MCQ) data and 1k+ video clips. Arxiv: https://arxiv.org/pdf/2503.06157 Project: https://embodiedcity.github.io/UrbanVideo-Bench/ Code: https://github.com/EmbodiedCity/UrbanVideo-Bench.code Dataset Description The… See the full description on the dataset page: https://huggingface.co/datasets/EmbodiedCity/UrbanVideo-Bench.

xarm_lift_medium_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_lift_medium_image.

fmb

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "unknown", "total_episodes": 1804, "total_frames": 338188, "total_tasks": 24, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:1804" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/fmb.

Vision-RL-Dataset

Dataset from haiderzm. Tags: size_categories:n<1K, format:parquet, modality:image, modality:text, library:datasets.

robotwin-data

Dataset from RogersPyke. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

KodCode-Light-RL-10K

🐱 KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding KodCode is the largest fully-synthetic open-source dataset providing verifiable solutions and tests for coding tasks. It contains 12 distinct subsets spanning various domains (from algorithmic to package-specific knowledge) and difficulty levels (from basic coding exercises to interview and competitive programming challenges). KodCode is designed for both supervised fine-tuning (SFT) and RL tuning. 🕸️… See the full description on the dataset page: https://huggingface.co/datasets/KodCode/KodCode-Light-RL-10K.

leju_robot_moving_parts_m

leju_robot_moving_parts_m 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 490… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_m.

AV-Deepfake1M

AV-Deepfake1M This is the official repository for the paper AV-Deepfake1M: A Large-Scale LLM-Driven Audio-Visual Deepfake Dataset. Abstract The detection and localization of highly realistic deepfake audio-visual content are challenging even for the most advanced state-of-the-art methods. While most of the research efforts in this domain are focused on detecting high-quality deepfake images and videos, only a few works address the problem of the localization of small… See the full description on the dataset page: https://huggingface.co/datasets/ControlNet/AV-Deepfake1M.

pusht-subtask

Dataset from lerobot. Tags: size_categories:10K<n<100K, format:parquet, modality:tabular, modality:timeseries, modality:video.

robotwin2-mt50-5k

Dataset from Oooer8. Tags: size_categories:10K<n<100K, modality:video, library:datasets, library:mlcroissant, region:us.

wildos

WildOS Frontiers Dataset Dataset Description This dataset provides visual frontier annotations for outdoor long-range navigation, created for WildOS: Open-Vocabulary Object Search in the Wild. The annotations are built on top of images from the GrandTour Dataset. Visual Frontiers denote regions in the image that correspond to candidate locations for further exploration — such as the end of a trail, an opening between trees, or a road turning at a curve. This dataset… See the full description on the dataset page: https://huggingface.co/datasets/leggedrobotics/wildos.

SSP

Search Self-Play (SSP) Dataset Paper | arXiv | Code Search Self-Play (SSP) is a reinforcement learning framework designed for training adversarial self-play agents with integrated search capabilities—enabling both proposer and solver agents to conduct multi-turn search engine calling and reasoning in a coordinated manner. Through RL training with rule-based outcome rewards, SSP enables two roles to co-evolve in an adversarial competition: the proposer learns to generate increasingly… See the full description on the dataset page: https://huggingface.co/datasets/Quark-LLM/SSP.

OpenFly-rlds

Dataset from IPEC-COMMUNITY. Tags: license:apache-2.0, region:us.

ucsd_pick_and_place_dataset

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 1355, "total_frames": 67750, "total_tasks": 3, "total_videos": 1355, "total_chunks": 2, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:1355" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/ucsd_pick_and_place_dataset.

humand_isaac_assets

Dataset from a2012203. Tags: region:us.

calvin-task-D-D-lerobot

Dataset from fywang. Tags: size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries, library:datasets.

Grap_Hook_Dataset_0324_lerobot

dataset-robotis-action-data-hub-dataset-lerobot-3

leju_robot_box_storage_parcel

leju_robot_box_storage_parcel 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 443… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel.

JEPA_Navigation

Dataset from AlterraLaniakea. Tags: license:mit, region:us.

xarm_lift_medium_replay_image

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 800, "total_frames": 20000, "total_tasks": 1, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:800" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": null, "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/xarm_lift_medium_replay_image.

aloha_mobile_cabinet

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "aloha", "total_episodes": 85, "total_frames": 127500, "total_tasks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:85" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4", "features": {… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_mobile_cabinet.

Fidelity_Driving_Bench

Dataset from AMao12138. Tags: region:us.

XHRBench

RSHR: A Benchmark for MLLMs on Ultra-High-Resolution Remote Sensing Data If our project helps you, please give us a star ⭐ on GitHub to support us💕 🔥 News 2025-11-14 🎉 We released the paper : RSHR: A Benchmark for MLLMs on Ultra-High-Resolution Remote Sensing Data. 😼RSHR Overview Large-scale ultra-high-resolution benchmark: RSHR is designed to evaluate fine-grained perception and complex reasoning of multimodal large language models in remote… See the full description on the dataset page: https://huggingface.co/datasets/RL-MIND/XHRBench.

financial-news-multisource

Multi-Source Financial & General News 🚀 57.1 MILLION ROWS OF NEWS CONTENT — one unified corpus for market-aware AI/ML I combined 24 public news datasets (many small on their own) into one consistent, ready-to-use layer so you don’t have to wrangle them yourself. Everything is normalized to a minimal schema (date, text, extra_fields) and shipped as Parquet shards per subset—streamable, DuckDB-friendly, and built with a trading date policy (this can be edited if folks see other use… See the full description on the dataset page: https://huggingface.co/datasets/Brianferrell787/financial-news-multisource.

lead_carla

LEAD: Minimizing Learner–Expert Asymmetry in End-to-End Driving Project Page | Paper | Code Official CARLA dataset accompanies our paper LEAD: Minimizing Learner–Expert Asymmetry in End-to-End Driving. We release the complete pipeline required to achieve state-of-the-art closed-loop performance on the Bench2Drive benchmark. Built around the CARLA simulator, the stack features a data-centric design with: Extensive visualization suite and runtime type validation for easier… See the full description on the dataset page: https://huggingface.co/datasets/ln2697/lead_carla.

furniture_bench_dataset_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "franka", "total_episodes": 5100, "total_frames": 3948057, "total_tasks": 9, "total_videos": 10200, "total_chunks": 6, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:5100"}, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/furniture_bench_dataset_lerobot.

pi0.5-datasets-lerobot-2026-03-09

Dataset from dalen-shi. Tags: license:apache-2.0, size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries.

carla-dataset

CARLA Dataset (Target) A large-scale driving dataset captured from CARLA simulator, containing RGB images and depth maps with camera parameters for autonomous driving research. Tar File Structure This dataset is stored in WebDataset format for efficient streaming and loading. Sharding Strategy 3 scenes per shard: Each .tar file contains exactly 3 complete scenes 93 frames per scene: Each scene is a video sequence of 93 consecutive frames ~279 frames per tar: 3… See the full description on the dataset page: https://huggingface.co/datasets/mkxdxd/carla-dataset.

D4RL

D4RL Dataset on HuggingFace This repository hosts the pre-downloaded D4RL dataset on HuggingFace. It is designed to provide accelerated data downloading for users, eliminating the need to download the dataset from scratch. Installation To use this dataset, you need to clone it into your local .d4rl directory. Here are the steps to do so: Navigate to your .d4rl directory: cd ~/.d4rl Clone the dataset repository from HuggingFace: git clone… See the full description on the dataset page: https://huggingface.co/datasets/imone/D4RL.

spacetools-rlfulltools

SpaceTools RL Full-Tools Dataset RL training data for spatial reasoning with the full suite of vision tools. Dataset Summary Train: 5,500 samples Format: Parquet with relative image paths Images: Stored in images/ directory (5,421 files) Note: The HuggingFace viewer may show fewer rows due to image path format. The full dataset contains 5,500 samples. Use load_dataset() or download directly to access all rows. Usage from datasets import load_dataset ds… See the full description on the dataset page: https://huggingface.co/datasets/siyich/spacetools-rlfulltools.

austin_sirius_dataset

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 559, "total_frames": 279939, "total_tasks": 2, "total_videos": 1118, "total_chunks": 1, "chunks_size": 1000, "fps": 20, "splits": { "train": "0:559" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/austin_sirius_dataset.

lerobot_rlbench_rvt_virtual_variation0

Dataset from vrlfdvla. Tags: size_categories:n<1K, modality:tabular, modality:timeseries, modality:video, region:us.

slime-opt-rl

Dataset from JLiangHe. Tags: region:us.

EgoVLA-Humanoid-Sim

Dataset from EgoVLA. Tags: license:mit, region:us.

VLA_Arena_L0_M_lerobot_openpi

VLA-Arena Dataset (L0 - Medium Variant) About VLA-Arena VLA-Arena is an open-source benchmark designed for the systematic evaluation of Vision-Language-Action (VLA) models. It provides a complete and unified toolchain covering scene modeling, demonstration collection, model training, and evaluation. Featuring 150+ tasks across 11 specialized suites, VLA-Arena assesses models through hierarchical difficulty levels (L0-L2) to ensure comprehensive metrics for safety… See the full description on the dataset page: https://huggingface.co/datasets/VLA-Arena/VLA_Arena_L0_M_lerobot_openpi.

genai-manipulation-detection-interior

GenAI Manipulation Detection Dataset - Interior Design Images 📋 Dataset Description This dataset contains 1000 paired images (real + manipulated) for training and evaluating GenAI manipulation detection models. Created for the MenaML Winter School 2026 Hackathon. Dataset Summary Total Images: 1000 pairs (2000 total images) Image Size: 512x512 Format: JPEG Source: Pinterest Interior Design Images (Kaggle) License: MIT 🎯 Challenge Context This… See the full description on the dataset page: https://huggingface.co/datasets/FatimahEmadEldin/genai-manipulation-detection-interior.

rdt-ft-data

Dataset Card This is the fine-tuning dataset used in the paper RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation. Source Project Page: https://rdt-robotics.github.io/rdt-robotics/ Paper: https://arxiv.org/pdf/2410.07864 Code: https://github.com/thu-ml/RoboticsDiffusionTransformer Model: https://huggingface.co/robotics-diffusion-transformer/rdt-1b Uses Download all archive files and use the following command to extract: cat rdt_data.tar.gz.* |… See the full description on the dataset page: https://huggingface.co/datasets/robotics-diffusion-transformer/rdt-ft-data.

OmniRAG-Agent

OmniRAG-Agent Dataset Training data for OmniRAG-Agent: Agentic Omnimodal Reasoning for Low-Resource Long Audio-Video Question Answering [paper] Dataset Preview Split VideoOmniBench WorldSense Daily-Omni Total train 504 504 504 1,512 Dataset Structure OmniRAG-Agent/ ├── VideoOmniBench/ │ └── train.parquet # 504 samples — visual perception & reasoning ├── WorldSense/ │ └── train.parquet # 504 samples — audio-visual understanding… See the full description on the dataset page: https://huggingface.co/datasets/JackMuX3Y/OmniRAG-Agent.

so101-segmentation

SO-101 Instance Segmentation Dataset Instance segmentation dataset for the SO-101 robot arm, combining MuJoCo domain-randomised synthetic renders with hand-labelled real frames. This dataset can be used to create a segmentation model for tracking the movements of the SO-101 Categories 8 classes, all prefixed so101_: ID Name Notes 1 so101_base 2 so101_shoulder 3 so101_upper_arm 4 so101_lower_arm 5 so101_wrist 6 so101_gripper includes moving… See the full description on the dataset page: https://huggingface.co/datasets/riversnow/so101-segmentation.

level1_single_rac4_2026-01-16

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "openarms_follower", "total_episodes": 3740, "total_frames": 7317758, "total_tasks": 2, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:3740" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot-data-collection/level1_single_rac4_2026-01-16.

lerobot_fr3_pickonly_dataset4_2000eps

Dataset from Ganaizzan. Tags: size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries, library:datasets.

LEROBOT_COLOR

Dataset from ZZR42. Tags: size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries, library:datasets.

bc_z_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "google_robot", "total_episodes": 39350, "total_frames": 5471693, "total_tasks": 104, "total_videos": 39350, "total_chunks": 40, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:39350" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/ygtxr1997/bc_z_lerobot.

berkeley_autolab_ur5

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 1000, "total_frames": 97939, "total_tasks": 5, "total_videos": 3000, "total_chunks": 1, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:1000" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/berkeley_autolab_ur5.

roboturk_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "sawyer", "total_episodes": 1796, "total_frames": 168423, "total_tasks": 3, "total_videos": 1796, "total_chunks": 2, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:1796" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/roboturk_lerobot.

Liaohe-RoboTwin-LiftPot

Dataset from Liaohe. Tags: task_categories:robotics, language:en, license:cc-by-4.0, modality:video, region:us.

robotwin_rollout

Dataset from Heisen0928. Tags: size_categories:100K<n<1M, modality:tabular, modality:text, modality:timeseries, library:datasets.

merged_LIBERO_lerobot_v2_IPEC_COMMUNITY

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "franka", "total_episodes": 1693, "total_frames": 273465, "total_tasks": 40, "total_videos": 3386, "total_chunks": 2, "chunks_size": 1000, "fps": 20, "splits": { "train": "0:1693"}, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/crislmfroes/merged_LIBERO_lerobot_v2_IPEC_COMMUNITY.

mango_data

Project Page: https://www.jeremiahcoholich.com/mango This dataset contains 150 human-teleoperated demonstrations, 150 synthetic MANGO demonstrations, and 150 sim demonstrations (for reproducibility) for each the four tasks: Stack blocks Close laptop Stack cups Pick coke Each .h5 file contains the following keys: arm_action (end-effector delta xyz and euler angles, used for training) eef_pos eef_pose eef_quat gripper_action gripper_state rgb_frames (camera observations) Additionally, data… See the full description on the dataset page: https://huggingface.co/datasets/jcoholich/mango_data.

berkeley_autolab_ur5_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "ur5", "total_episodes": 896, "total_frames": 87783, "total_tasks": 5, "total_videos": 1792, "total_chunks": 1, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:896" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/IPEC-COMMUNITY/berkeley_autolab_ur5_lerobot.

aloha_static_cups_open

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "aloha", "total_episodes": 50, "total_frames": 20000, "total_tasks": 1, "total_videos": 200, "total_chunks": 1, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:50" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/aloha_static_cups_open.

vr-folding

Dataset Card for VR-Folding Dataset Dataset Summary This is the VR-Folding dataset created by the CVPR 2023 paper GarmentTracking: Category-Level Garment Pose Tracking. This dataset is recorded with a system called VR-Garment, which is a garment-hand interaction environment based on Unity. To download the dataset, use the following shell snippet: git lfs install git clone https://huggingface.co/datasets/robotflow/garment-tracking # if you want to clone without large… See the full description on the dataset page: https://huggingface.co/datasets/robotflow/vr-folding.

maniskill_pusht_sim_train_rfm_old_reward_form

Dataset from ACriv. Tags: size_categories:1K<n<10K, format:parquet, format:optimized-parquet, modality:text, modality:timeseries.

G1_Dex1_Fold_Towel

Real Unitree G1 teleoperated dataset of the humanoid folding a towel, recorded with the Dex1 gripper. Includes full G1 state/action: left_arm (7 DoF), right_arm (7 DoF), body, Dex1 gripper, and end-effector poses. Published by Unitree Robotics.

libero_lerobot

Dataset from YANYAOYUAN. Tags: license:apache-2.0, size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries.

controlling-reasoning-models-privacy-outputs

Model Outputs Dataset Card Go to **Files and versions** tab to access the data. Dataset Description This dataset contains the raw model generations (reasoning traces and final answers) produced in the experiments described in our paper Controllable Reasoning Models are Private Thinkers. It aggregates outputs for: two model families: Qwen 3 and Phi 4, multiple model sizes (1.7B–14B), five variants per model (baseline, RT-IF–optimized, FA-IF–optimized… See the full description on the dataset page: https://huggingface.co/datasets/haritzpuerto/controlling-reasoning-models-privacy-outputs.

reachy-mini-emotions-library

Dataset from pollen-robotics. Tags: size_categories:n<1K, format:json, modality:audio, modality:text, library:datasets.

FuSe

The FuSe dataset contains 26,866 trajectories collected on a WidowX robot at the RAIL lab @ UC Berkeley, USA. It contains visual, tactile, sound and action data collected across several environments, annotated with natural language. Paper: https://huggingface.co/papers/2501.04693 Code: https://github.com/oier-mees/FuSe If you use the dataset please cite: @inproceedings{jones25fuse, title={Beyond Sight: Finetuning Generalist Robot Policies with Heterogeneous Sensors via Language Grounding}… See the full description on the dataset page: https://huggingface.co/datasets/oier-mees/FuSe.

G1_WBT_Brainco_Collect_Plates_Into_Dishwasher

Data Structure Observations observation.state.ee_state (12) End-effector states of the robot. Computed via forward kinematics (FK) from the root link to the left and right end-effectors. Includes the contribution of the waist. Represented as concatenated poses of both end-effectors. observation.state.hand_state (12 or 2) Finger states for both hands. The dimensionality depends on the hand type. Inspire Hand (range: 0.0 – 1.0, open → close)… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_WBT_Brainco_Collect_Plates_Into_Dishwasher.

Liaohe-RoboTwin-AdjustBottle

Dataset from Liaohe. Tags: task_categories:robotics, language:en, license:cc-by-4.0, modality:video, region:us.

guru-RL-92k

Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective Dataset Description Guru is a curated six-domain dataset for training large language models (LLM) for complex reasoning with reinforcement learning (RL). The dataset contains 91.9K high-quality samples spanning six diverse reasoning-intensive domains, processed through a comprehensive five-stage curation pipeline to ensure both domain diversity and reward verifiability. Dataset… See the full description on the dataset page: https://huggingface.co/datasets/LLM360/guru-RL-92k.

VLA_Arena_L1_M_lerobot_openpi

VLA-Arena Dataset (L1 - Medium Variant) About VLA-Arena VLA-Arena is an open-source benchmark designed for the systematic evaluation of Vision-Language-Action (VLA) models. It provides a complete and unified toolchain covering scene modeling, demonstration collection, model training, and evaluation. Featuring 150+ tasks across 11 specialized suites, VLA-Arena assesses models through hierarchical difficulty levels (L0-L2) to ensure comprehensive metrics for safety… See the full description on the dataset page: https://huggingface.co/datasets/VLA-Arena/VLA_Arena_L1_M_lerobot_openpi.

MetaWorld_Dataset

Dataset from handavid1. Tags: size_categories:100K<n<1M, format:parquet, modality:tabular, modality:timeseries, modality:video.

Liaohe-RoboTwin-BeatBlockHammer

Dataset from Liaohe. Tags: task_categories:robotics, language:en, license:cc-by-4.0, modality:video, region:us.

RL

Dataset from tCorp-Startup. Tags: size_categories:1K<n<10K, modality:tabular, modality:text, region:us.

leju_robot_moving_parts_a

leju_robot_moving_parts_a 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 380… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_a.

Liaohe-RoboTwin-PlaceBurgerFries

Dataset from Liaohe. Tags: task_categories:robotics, language:en, license:cc-by-4.0, modality:video, region:us.

LegalQAEval

LegalQAEval 🏋 LegalQAEval by Isaacus is the first extractive question answering benchmark for the legal domain. LegalQAEval consists of 2,410 legal texts, questions and answers split across a validation set (containing 1,204 examples) and a test set (containing 1,206 examples). As an extractive question answering benchmark, LegalQAEval is suitable for evaluating the performance of both extractive and generative question answering models as well as rerankers and relevance scorers… See the full description on the dataset page: https://huggingface.co/datasets/isaacus/LegalQAEval.

robocasa-PnPCounterToSink-MG_100_F_472

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "robomimic", "total_episodes": 572, "total_frames": 375092, "total_tasks": 55, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 20, "splits": { "train": "0:572" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/DAVIAN-Robotics/robocasa-PnPCounterToSink-MG_100_F_472.

G1_Dex1_Prepare_Fruit

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Prepare_Fruit.

sentiment-labelled-sentences

Dataset from IsaacDev. Tags: size_categories:1K<n<10K, format:text, modality:text, library:datasets, library:mlcroissant.

ManiSkill-Memory-dependence

A comprehensive memory dependence robot benchmark across 4 manipulation tasks from different memory dimension

community_dataset_v1

Community Dataset v1 A large-scale community-contributed robotics dataset for vision-language-action learning, featuring 128 datasets from 55 contributors worldwide. We used this dataset to pretrain SmolVLA. However, this is not a complete set, but the dataset that we selected using specific filters, like fps, min num of episodes, and some qualitative assessment of video qualities, using the https://huggingface.co/spaces/Beegbrain/FilterLeRobotData tool. We also manually curated the… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceVLA/community_dataset_v1.

Evo1_MetaWorld_Dataset

Dataset from MINT-SJTU. Tags: size_categories:100K<n<1M, format:parquet, modality:tabular, modality:timeseries, modality:video.

VIDIT-Depth-ControlNet

VIDIT Dataset This is a version of the VIDIT dataset equipped for training ControlNet using depth maps conditioning. VIDIT includes 390 different Unreal Engine scenes, each captured with 40 illumination settings, resulting in 15,600 images. The illumination settings are all the combinations of 5 color temperatures (2500K, 3500K, 4500K, 5500K and 6500K) and 8 light directions (N, NE, E, SE, S, SW, W, NW). Original image resolution is 1024x1024. We include in this version only the… See the full description on the dataset page: https://huggingface.co/datasets/Nahrawy/VIDIT-Depth-ControlNet.

harbor-rl-data

harbor-rl-data Harbor task datasets for online RL training with rllm-harbor. Each task is a self-contained Harbor task directory containing: task.toml — task metadata (resources, timeouts) instruction.md — problem statement shown to the agent tests/test.sh — evaluation script (runs agent patch, grades with swebench) tests/config.json — install config (python version, packages, test command) Reward is 1 if all FAIL_TO_PASS tests pass and all PASS_TO_PASS tests remain passing, else… See the full description on the dataset page: https://huggingface.co/datasets/ElvinDu518/harbor-rl-data.

RSVLM_SFT

MF-RSVLM FUSE-RSVLM: Feature Fusion Vision-Language Model for Remote Sensing Project Page | Paper | Model | Dataset If this project helps you, please give us a star on GitHub. Overview MF-RSVLM is a remote sensing vision-language model (VLM). It combines a CLIP vision encoder, a two-layer MLP projector, and a Vicuna-7B LLM, and is trained in two stages for modality alignment and instruction following. Visual Encoder: CLIP… See the full description on the dataset page: https://huggingface.co/datasets/RL-MIND/RSVLM_SFT.

Multi-SWE-RL

Multi-SWE-RL is an open-source community focused on building high-quality RL datasets for complex software engineering tasks. Our mission is to enable autonomous agents that solve real-world coding challenges and advance toward Artificial General Intelligence (AGI). • 🔮 Core Belief: Scaling RL in real-world environments is the path to human-like intelligence• 🛠️ Purpose: Create RL data infrastructure for autonomous software engineering agents Join us in advancing the next generation of… See the full description on the dataset page: https://huggingface.co/datasets/ByteDance-Seed/Multi-SWE-RL.

_gr1_unified

PhysicalAI-Robotics-GR00T-X-Embodiment-Sim Github Repo: Isaac GR00T N1 We provide a set of datasets used for post-training of GR00T N1. Each dataset is a collection of trajectories from different robot embodiments and tasks. Cross-embodied bimanual manipulation: 9k trajectories Dataset Name #trajectories bimanual_panda_gripper.Threading 1000 bimanual_panda_hand.LiftTray 1000 bimanual_panda_gripper.ThreePieceAssembly 1000… See the full description on the dataset page: https://huggingface.co/datasets/nhatchung/_gr1_unified.

ZwZ-RL-VQA

ZwZ-RL-VQA: Region-to-Image Distilled Training Data for Fine-Grained Perception This dataset contains 74K high-quality VQA pairs generated via Region-to-Image Distillation (R2I) for training multimodal large language models (MLLMs) on fine-grained perception tasks without test-time tool use. 📖 Overview The Zooming without Zooming (ZwZ) method transforms "zooming" from an inference-time tool into a training-time primitive: Zoom-in Synthesis: Strong teacher models… See the full description on the dataset page: https://huggingface.co/datasets/inclusionAI/ZwZ-RL-VQA.

G1_Dex3_ToastedBread_Dataset

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex3_ToastedBread_Dataset.

control-arena-32217692-archive-20260312

control-arena 32217692 archive Source host: 32217692 Created: 2026-03-12T00:47:36.659953+00:00 Contents: archived failed-run directories from the instance, uploaded as separate folders under failed_runs/. Manifest: 32217692_hf_manifest.csv.

g1-rl-locomotion

G1 RL Locomotion Dataset Unitree G1 humanoid locomotion RL training data collected on NVIDIA Jetson AGX Thor. Robot: Unitree G1 (29-DOF actuated, 71-dim state) Backends: MuJoCo (primary), Newton GPU (experimental) Total transitions: ~270,000 (combined master) State dim: 71 (qpos:36 + qvel:35) Action dim: 29 (all actuated joints) Format: NPZ (SB3 replay buffer compatible) Collection period: March 2026

DR_dataset

DR_dataset Simulated robot manipulation dataset for imitation learning. Dataset Description This dataset contains 200 episodes of a Trossen WXAI robotic arm performing food transfer tasks in MuJoCo simulation. Task Description The robot performs a scooping motion to transfer food from a source container to target bowls: HOME - Start at home position APPROACH_CONTAINER - Move above the source container REACH_CONTAINER - Lower into container SCOOP - Perform… See the full description on the dataset page: https://huggingface.co/datasets/ganatrask/DR_dataset.

drone-audio-detection-samples

Dataset Description Drone Audio Detection Samples (DADS) is currently the largest publicly available drone audio database, specifically designed for developing drone detection systems using deep learning techniques. All audio files are standardized to a sample rate of 16,000 Hz, 16-bit depth, mono-channel, and vary in length from 500 milliseconds to several minutes. Most drone audio files were manually trimmed to ensure that a drone was always present in the recording. However, some… See the full description on the dataset page: https://huggingface.co/datasets/geronimobasso/drone-audio-detection-samples.

pick-up-cube-robotai-team2-level-1

Dataset from darikod. Tags: modality:image, region:us.

leju_robot_moving_parts_n

leju_robot_moving_parts_n 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 492… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_n.

leju_robot_moving_parts_s

leju_robot_moving_parts_s 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 495… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_s.

G1_Dex1_Stack_Block

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Stack_Block.

EmbodiedGenData

https://huggingface.co/spaces/HorizonRobotics/EmbodiedGen-Gallery-Explorer

ManiSkill_Demonstrations

ManiSkill Demonstrations This dataset repo contains all of the latest ManiSkill demonstration datasets as well as some pretained model weights used to generate some demonstrations. To download by environment ID run python -m mani_skill.utils.download_demo env_id You can also replace env_id with "all" to download all demos. For the previous ManiSkill2 paper's demonstration data (v0.5.3 of the mani_skill package) see https://huggingface.co/datasets/haosulab/ManiSkill2 The… See the full description on the dataset page: https://huggingface.co/datasets/haosulab/ManiSkill_Demonstrations.

rl-python

Dataset from satpalsr. Tags: size_categories:10K<n<100K, format:json, modality:text, library:datasets, library:pandas.

G1_Dex1_Pour_Medicine

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Pour_Medicine.

libero-rlds

Dataset Structure These datasets are used for MemoryVLA training. This is the standard LIBERO setting and can be directly used for other models as well.All data follow the RLDS format from the LIBERO benchmark, where each task initially contains 50 trajectories and failed rollouts are filtered out.NOTE: LIBERO-90 is also included. libero_spatial_no_noops — 10 tasks libero_object_no_noops — 10 tasks libero_goal_no_noops — 10 tasks libero_10_no_noops — 10 tasks… See the full description on the dataset page: https://huggingface.co/datasets/shihao1895/libero-rlds.

robotwin2.0-fastwam

RobotWin 2.0 (Preprocessed LeRobot v2.1 Release) This repository releases our preprocessed RoboTwin / RobotWin 2.0 dataset in LeRobot v2.1 format for the open-source release of Fast-WAM: Do World Action Models Need Test-time Future Imagination? This is not the official upstream RoboTwin release. It is our paper-specific processed version prepared to support training, evaluation, and reproducibility for our project. This Hugging Face repository distributes the dataset as split… See the full description on the dataset page: https://huggingface.co/datasets/yuanty/robotwin2.0-fastwam.

habitat_web_pose_train

Dataset from Aasdfip. Tags: size_categories:10K<n<100K, format:parquet, format:optimized-parquet, modality:image, modality:text.

Paired_Compressible_Boussinesq_Flow_Simulation_with_Random_Temperature_BCs

Dataset from NurshatMenglik. Tags: license:cc-by-4.0, size_categories:1K<n<10K, format:imagefolder, modality:image, library:datasets.

G1_Dex1_Erase_Board

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Erase_Board.

carla-dataset-ped

CARLA Stage 2 Pedestrian Dataset A large-scale driving dataset (Stage 2) focused on pedestrians, captured from CARLA simulator. This dataset contains RGB images and depth maps with camera parameters, specifically tracking pedestrian movements. Tar File Structure This dataset is stored in WebDataset format for efficient streaming and loading. Sharding Strategy 2 scenes per shard: Each .tar file contains exactly 2 complete scenes 200 frames per scene: Each scene… See the full description on the dataset page: https://huggingface.co/datasets/mkxdxd/carla-dataset-ped.

hidden_phys_vqa_datasets-v1

Dataset from mlcf-robot. Tags: region:us.

G1_Dex1_Organize_Tools

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Organize_Tools.

DroneAudioSet

Dataset from ahlab-drone-project. Tags: license:mit, size_categories:1K<n<10K, format:parquet, modality:text, library:datasets.

drifting-vla-rlbench

Dataset from christian0420. Tags: region:us.

G1_WBT_Inspire_Put_Clothes_into_Washing_Machine

Real Unitree G1 whole-body teleoperated dataset: the humanoid loads clothes into a washing machine using Inspire hands. Laundry-pipeline task, upstream of folding.

RLAIF-V-Dataset

Dataset Card for RLAIF-V-Dataset This dataset was introduced in RLAIF-V: Open-Source AI Feedback Leads to Super GPT-4V Trustworthiness. GitHub This dataset was also used in MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe News: [2025.09.18] 🎉 Our data is used in the powerful MiniCPM-V 4.5 model, which represents a state-of-the-art end-side MLLM achieving GPT-4o level performance! [2025.03.01] 🎉 RLAIF-V is accepted by CVPR 2025!… See the full description on the dataset page: https://huggingface.co/datasets/openbmb/RLAIF-V-Dataset.

G1_Dex1_Pack_PencilBox

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Pack_PencilBox.

ManiSkill_bridge_v2_real2sim

Assets for Real2Sim evaluation of the Bridge v2 dataset from https://github.com/simpler-env/SimplerEnv/

ball_to_cup

ball_to_cup This dataset was created using the teleop-i2rt-robot framework and follows the LeRobot v2.1 format. Dataset Information Robot Type: i2rt_yam_pro Total Episodes: 456 Total Frames: 87,711 FPS: 30 Codebase Version: v2.1 Cameras Features This dataset contains 11 features: observation.images.wrist: shape=[480, 640, 3], dtype=video observation.images.base: shape=[480, 640, 3], dtype=video observation.images.env: shape=[480, 640, 3]… See the full description on the dataset page: https://huggingface.co/datasets/Neotix-Robotics/ball_to_cup.

data_embody

Dataset from Atarilab. Tags: size_categories:n<1K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

Canny-Control-Dataset

Canny Control Dataset A few example clips for ControlNet-style training using canny-edge-guided prompts. Videos taken from the DAVIS dataset. This dataset was used to train the Canny IC-Lora model for LTXV. https://huggingface.co/Lightricks/LTX-Video-ICLoRA-canny-13b-0.9.7

leju_robot_moving_parts_o

leju_robot_moving_parts_o 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 490… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_o.

rl_training

Dataset from tnnanh1005. Tags: region:us.

G1_Dex1_Pack_PingPong

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Pack_PingPong.

robot_control

Robot Control Dataset Overview The TD-NIRS and EEG data for this dataset was collected using a Kernel Flow headset. Participants were asked to: Clench left fist Clench right fist Clench both fists Tap tongue Relax Data The data folder contains numpy files. Each numpy file represents a 15s chunk of data. Timing t=0 t=3 t=15 rest period stimulus presented end of data Format You can use the… See the full description on the dataset page: https://huggingface.co/datasets/KernelCo/robot_control.

results_v2

Dataset Card for "results_v2" Leaderboard More Information needed

omnigibson-robot-assets

omnigibson-robot-assets This repo contains the robot assets and the state mesh assets for OmniGibson. Pushing updates To generate an updated zipped version for the zipped versions repo, you can first build the asset bundle: git archive --format=zip --prefix=assets/ --output omnigibson-robot-assets.zip main Then you can upload the file using the HuggingFace website.

leju_robot_moving_parts_t

leju_robot_moving_parts_t 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 493… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_t.

G1_Dex1_Wipe_Table

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Wipe_Table.

open_pose_controlnet

Dataset for training controlnet models with conditioning images as Human Pose the entries have been taken from this dataset ptx0/photo-concept-bucket the open pose images have been generated with controlnet_aux for the scripts to download the files, generate the open pose and the dataset please refer to: raulc0399/dataset_scripts

Motion-Robot-Dataset

Dataset from xiaoheng22. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

controlnet-testing

Dataset from YiYiXu. Tags: license:openrail, size_categories:n<1K, format:imagefolder, modality:image, library:datasets.

leju_robot_moving_parts_f

leju_robot_moving_parts_f 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 489… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_f.

leju_robot_moving_parts_b

leju_robot_moving_parts_b 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 490… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_b.

leju_robot_part_placement

leju_robot_part_placement 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 538… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_part_placement.

dr-tulu-rl-data

[!NOTE] For full information, go check out the Dr Tulu paper here. DR Tulu RL Data This dataset contains the RL training data for DR Tulu, containing prompts and search-based rubrics generated from OpenScholar and SearchArena prompts, with rubrics generated using GPT-4.1. Important: This does not contain the RaR datasets we use in final RL training, but only the OpenScholar and SearchArena subsets. For the RaR data, we use data from: anisha2102/RaR-Science-20k-o3-mini… See the full description on the dataset page: https://huggingface.co/datasets/rl-research/dr-tulu-rl-data.

ZwZ-RL-VQA-mini

ZwZ-RL-VQA-mini Default split points to train_crop.parquet.

leju_robot_moving_parts_c

leju_robot_moving_parts_c 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 490… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_c.

PDM_Lite_Carla_LB2

PDM-Lite Dataset for CARLA Leaderboard 2.0 Description PDM-Lite is a state-of-the-art rule-based expert system for autonomous urban driving in CARLA Leaderboard 2.0, and the first to successfully navigate all scenarios. This dataset was used to create the QA dataset for DriveLM-Carla, a benchmark for evaluating end-to-end autonomous driving algorithms with Graph Visual Question Answering (GVQA). DriveLM introduces GVQA as a novel approach, modeling perception, prediction… See the full description on the dataset page: https://huggingface.co/datasets/autonomousvision/PDM_Lite_Carla_LB2.

G1_Dex1_Bag_Insert

This dataset was created using LeRobot. Due to the inability to precisely describe spatial positions, adjust the scene to closely match the first frame of the dataset after installing the hardware as specified in Part 5 of AVP Teleoperation Documentation. Data collection is not completed in a single session, and variations between data entries exist. Ensure these variations are accounted for during model training. Dataset Structure meta/info.json: { "codebase_version":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_Bag_Insert.

Thyme-RL

[📖 Home Page] [📖 Github Repo] [📖 Technique Report] [📊 Thyme SFT Model] [📊 Thyme RL Model] [📝 SFT Data] [📝 RL Data] 🔥 News 2025.08.15 🌟 We are excited to introduce Thyme: Think Beyond Images. Thyme transcends traditional ``thinking with images'' paradigms by autonomously generating and executing diverse image processing and computational operations through executable code, significantly enhancing performance on high-resolution perception and complex… See the full description on the dataset page: https://huggingface.co/datasets/Kwai-Keye/Thyme-RL.

ai2thor-sideview-only

Dataset from linjieli222. Tags: size_categories:10K<n<100K, format:parquet, modality:image, modality:text, library:datasets.

DrivingDojo

Dataset from Yuqi1997. Tags: task_categories:image-to-video, language:en, license:apache-2.0, size_categories:n>1T, region:us.

taste-rob-bathroom

TASTE-Rob Bathroom Video Dataset This dataset contains 103 videos from the Bathroom scene of TASTE-Rob dataset. Usage from huggingface_hub import hf_hub_download # Download a specific video video_path = hf_hub_download( repo_id="charlychan123/taste-rob-bathroom", filename="video_name.mp4", repo_type="dataset" ) Citation If you use this dataset, please cite the original TASTE-Rob paper.

sd

Capstone VLA Datasets LeRobot v2.1 datasets for a liquid pouring task on a KuavoV4Pro humanoid robot (34-DOF, bimanual). Used to benchmark GR00T N1.6, pi0.5, and Diffusion Policy under varying real-to-synthetic data ratios. Mixture Configurations Dataset Type Episodes human unknown 200 Dataset Structure (LeRobot v2.1) <dataset>/ ├── meta/ │ ├── info.json │ ├── tasks.jsonl │ ├── episodes.jsonl │ └── stats.safetensors ├── data/ │ └──… See the full description on the dataset page: https://huggingface.co/datasets/Lusmse/sd.

PAPO_ViRL39K_train

This is the official release of the training data for paper PAPO: Perception-Aware Policy Optimization for Multimodal Reasoning. Hugging Face Paper: https://huggingface.co/papers/2507.06448 Project page: https://mikewangwzhl.github.io/PAPO/ This dataset is the train split of the training dataset for PAPO. (Optional) To include validate set, you may use our adapted val split PAPOGalaxy/PAPO_MMK12_test. Data Source Training We adapt the multimodal benchmark… See the full description on the dataset page: https://huggingface.co/datasets/PAPOGalaxy/PAPO_ViRL39K_train.

PhysicalAI-Robotics-Manipulation-Objects

PhysicalAI-Robotics-Manipulation-Objects is a dataset of automatic generated motions of robots performing operations such as picking and placing objects in a kitchen environment. The dataset was generated in IsaacSim leveraging reasoning algorithms and optimization-based motion planning to find solutions to the tasks automatically [1, 3]. The dataset includes a bimanual manipulator built with Kinova Gen3 arms. The environments are kitchen scenes where the furniture and appliances were… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-Manipulation-Objects.

mikasa-robo

This repository contains ready-made datasets and pre-trained oracle agent checkpoints for the MIKASA-Robo benchmark suite, a comprehensive collection of memory-intensive robotic manipulation tasks (see our Website, Paper, and Github for details). Instructions for collecting datasets using oracle agent checkpoints and a description of the datasets are available at https://github.com/CognitiveAISystems/MIKASA-Robo. If you find our datasets useful, please cite our paper :)… See the full description on the dataset page: https://huggingface.co/datasets/avanturist/mikasa-robo.

PhysicalAI-Robotics-Manipulation-Kitchen

PhysicalAI Robotics Manipulation in the Kitchen Dataset Description: PhysicalAI-Robotics-Manipulation-Kitchen is a dataset of automatic generated motions of robots performing operations such as opening and closing cabinets, drawers, dishwashers and fridges. The dataset was generated in IsaacSim leveraging reasoning algorithms and optimization-based motion planning to find solutions to the tasks automatically [1, 3]. The dataset includes a bimanual manipulator built with… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-Manipulation-Kitchen.

GR1-Manipulation-IsaacSim-Rendered

GR1 Manipulation — Isaac Sim Rendered Dataset Dataset Description Multi-modal rendered dataset of NVIDIA GR1 humanoid robot performing tabletop manipulation tasks in Isaac Sim. Each episode is rendered as 3 synchronized MP4 videos: RGB — ray-traced color video Depth — normalized depth map (grayscale, 0–10m range) Segmentation — semantic segmentation with per-object class coloring This dataset is designed as input for NVIDIA Cosmos Transfer 2.5 to generate photorealistic… See the full description on the dataset page: https://huggingface.co/datasets/khang123452/GR1-Manipulation-IsaacSim-Rendered.

ControlNets

Dataset from tyDiffusion. Tags: region:us.

RLHF-V

Borrowed from: https://huggingface.co/datasets/openbmb/RLHF-V-Dataset You can use it in LLaMA Factory by specifying dataset: rlhf_v.

taste-rob-bedroom

TASTE-Rob Bedroom Video Dataset This dataset contains 157 videos from the Bedroom scene of TASTE-Rob dataset. Usage from huggingface_hub import hf_hub_download # Download a specific video video_path = hf_hub_download( repo_id="charlychan123/taste-rob-bedroom", filename="video_name.mp4", repo_type="dataset" ) Citation If you use this dataset, please cite the original TASTE-Rob paper.

G1_WBT_Inspire_Collect_Clothes_MainCamOnly

Real Unitree G1 whole-body teleoperated (WBT) dataset: the humanoid picks up and collects clothes using Inspire hands (multi-finger, dexterous). Main-camera-only variant — lower storage footprint. Part of Unitree's laundry-pipeline data collection.

leju_robot_box_storage_parcel_c

leju_robot_box_storage_parcel_c 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 493… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_c.

hh-rlhf-helpful-base

HH-RLHF-Helpful-Base Dataset Summary The HH-RLHF-Helpful-Base dataset is a processed version of Anthropic's HH-RLHF dataset, specifically curated to train models using the TRL library for preference learning and alignment tasks. It contains pairs of text samples, each labeled as either "chosen" or "rejected," based on human preferences regarding the helpfulness of the responses. This dataset enables models to learn human preferences in generating helpful responses… See the full description on the dataset page: https://huggingface.co/datasets/trl-lib/hh-rlhf-helpful-base.

robotwin

RoboTwin Bimanual Robotic Manipulation Platform Lastest Version: RoboTwin 2.0🤲 Webpage | Document | Paper | Community | Leaderboard https://private-user-images.githubusercontent.com/88101805/463126988-e3ba1575-4411-4a36-ad65-f0b2f49890c3.mp4 [2.0 Version (lastest)] RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic ManipulationUnder Review 2025: Webpage | Document | PDF | arXiv | Talk (in Chinese) | 机器之心 | Leaderboard… See the full description on the dataset page: https://huggingface.co/datasets/sdvkasc/robotwin.

drone_detections

Video links for the tracked drones: Video 1: https://youtu.be/4oP8efiyYFo Video 2: https://youtu.be/yX8kFZaAmwQ Dataset Name: drone_detections Content: Frames from two drone videos containing one detection or more. Source videos: Two test input videos split into frames using ffmpeg. Detector used for scanning: YOLOv8 trained on a custom drone dataset (https://universe.roboflow.com/iium-xzks3/drone-detection-uyucu/dataset/6) Format: Parquet file and a detections/ folder with .jpg images.… See the full description on the dataset page: https://huggingface.co/datasets/AhmadAlBajes/drone_detections.

lehome0322

EXP02a_collect This dataset repo stores the original folder contents of EXP02a_collect. Upload marker: lehome0322 Upload date: 2026-03-22 Source path: /root/data/lehome-challenge/Datasets/dataflywheel/EXP02a_collect Size at upload: about 5.1G Structure: 12 sub-datasets (001 to 012) Each sub-dataset keeps the original LeRobot-style layout, including: data/ images/ videos/ meta/ This repo preserves the raw collected structure and does not merge the 001 to 012 sub-datasets into a… See the full description on the dataset page: https://huggingface.co/datasets/20060618yf/lehome0322.

leju_robot_box_storage_parcel_g

leju_robot_box_storage_parcel_g 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 492… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_g.

ReconDreamer-RL

Dataset from Ni1111. Tags: region:us.

seraphim-drone-detection-dataset

Seraphim Drone Detection Dataset Dataset Overview This is a comprehensive drone image dataset curated from 23 open-source datasets and processed through a custom cleaning pipeline. The dataset is designed for training object detection models to identify drones in various environments and conditions. The majority of images feature rotary-wing (multi-rotor) unmanned aerial vehicles (UAVs), with a smaller portion representing fixed-wing and hybrid.… See the full description on the dataset page: https://huggingface.co/datasets/lgrzybowski/seraphim-drone-detection-dataset.

Xlang-Embodied-dataset

Dataset from EricsXt. Tags: license:apache-2.0, region:us, EmbodiedAI, Robot, RealworldData.

Assignment3_Attallah_Drone_Detections

UAV Drone Detection and Tracking This project detects UAV drones in video using a fine-tuned YOLO detector and tracks their trajectories using a Kalman filter. Dataset Training Dataset: https://universe.roboflow.com/dronedetectionpitt-nwyps/drone-detection-yhkcr YouTube videos used as evaluation inputs: https://www.youtube.com/watch?v=DhmZ6W1UAv4 https://www.youtube.com/watch?v=YrydHPwRelI These two videos are unlabeled out-of-distribution (OOD) inputs. Because the… See the full description on the dataset page: https://huggingface.co/datasets/grugg/Assignment3_Attallah_Drone_Detections.

leju_robot_box_storage_parcel_d

leju_robot_box_storage_parcel_d 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 212… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_d.

maniskill_pusht_sim_train_rfm

Dataset from ACriv. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

nyu_franka_play_dataset_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "franka", "total_episodes": 365, "total_frames": 34448, "total_tasks": 1, "total_videos": 730, "total_chunks": 1, "chunks_size": 1000, "fps": 3, "splits": { "train": "0:365" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/ygtxr1997/nyu_franka_play_dataset_lerobot.

robotwin-pi0-eval-result-supp

Dataset from RogersPyke. Tags: size_categories:1K<n<10K, modality:video, library:datasets, library:mlcroissant, region:us.

Motion-Robot-Video-Dataset

Dataset from xiaoheng22. Tags: region:us.

rlbench-18-tasks

RLBench 18 Tasks Dataset Overview This repository provides the RLBench dataset for 18 tasks, originally hosted by PerAct in Google Drive. Since downloading large files from Google Drive via terminal can be problematic due to various limits, we have mirrored the dataset on Hugging Face for easier access. To know more about the details of this dataset, please refer to PerAct. Dataset Structure The dataset is organized into three splits: data/ ├── train/ #… See the full description on the dataset page: https://huggingface.co/datasets/hqfang/rlbench-18-tasks.

Nemotron-RL-Super-Training-Blends

Dataset Description: Nemotron-3-Super-RL-Training-Blends contains the dataset blends used to train the Nemotron-3-Super-120B-A12B model. RL training for the Nemotron-3-Super-120B-A12B model is done in 6 stages: RLVR 1, RLVR 2, RLVR 3, SWE 1, SWE 2, and RLHF. The blends for each stage consist of data from various datasets, which we detail below. The percentages in parentheses indicate the mixing ratios of the dataset components. Note that the model was also trained on additional data… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Super-Training-Blends.

ReplicaCAD_dataset

What is ReplicaCAD? Vist the ReplicaCAD Homepage The ReplicaCAD dataset is an artist recreation of the scanned “FRL apartment” variations from the Replica dataset. This dataset is intended for use in the Habitat simulator for embodied in-home interaction tasks such as object re-arrangement. All materials are licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) Public License. Dataset Contents: We provide two dataset downloads with different… See the full description on the dataset page: https://huggingface.co/datasets/ai-habitat/ReplicaCAD_dataset.

touch_in_the_wild-dataset

📦 Touch in the Wild Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper 🔍 Overview This dataset supports the paper Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper. It contains a wide range of visuo-tactile demonstrations collected using a portable visuo-tactile gripper, spanning both structured lab settings and unstructured real-world environments. The dataset is divided into three major categories: Four… See the full description on the dataset page: https://huggingface.co/datasets/binghaohuang-robot/touch_in_the_wild-dataset.

leju_robot_box_storage_parcel_b

leju_robot_box_storage_parcel_b 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 497… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_b.

OpenHumnoidActuatedFaceData

🤖 Open Humanoid Actuated Face Dataset Dataset Summary The Open Humanoid Actuated Face Dataset is designed for researchers working onfacial‑actuation control, robotics, reinforcement learning, and human–computer interaction. Origin – collected during a reinforcement‑learning (RL) training loop whose objective was to reproduce human facial expressions. Platform – a modified i2Head InMoov humanoid head with a silicone skin. Control – 16 actuators driving facial… See the full description on the dataset page: https://huggingface.co/datasets/infosysLimited/OpenHumnoidActuatedFaceData.

opcdimage-qwen3vl4b-zwz-rl-vqa-mini

Dataset from muyuho. Tags: size_categories:1K<n<10K, format:parquet, modality:image, modality:text, library:datasets.

nyu_franka_play_dataset

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 456, "total_frames": 44875, "total_tasks": 1, "total_videos": 912, "total_chunks": 1, "chunks_size": 1000, "fps": 5, "splits": { "train": "0:456" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/nyu_franka_play_dataset.

lerobot-simulation-koch-arm-over-the-barrier-01

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "so_arm_101", "total_episodes": 1, "total_frames": 424, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:1" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/OpenGraphLabs-Research/lerobot-simulation-koch-arm-over-the-barrier-01.

G1_WBT_Brainco_Pickup_Pillow

Data Structure Observations observation.state.ee_state (12) End-effector states of the robot. Computed via forward kinematics (FK) from the root link to the left and right end-effectors. Includes the contribution of the waist. Represented as concatenated poses of both end-effectors. observation.state.hand_state (12 or 2) Finger states for both hands. The dimensionality depends on the hand type. Inspire Hand (range: 0.0 – 1.0, open → close)… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_WBT_Brainco_Pickup_Pillow.

leju_robot_pass_the_cleaner_e

leju_robot_pass_the_cleaner_e 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home industry 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_pass_the_cleaner_e.

berkeley_fanuc_manipulation

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.0", "robot_type": "unknown", "total_episodes": 415, "total_frames": 62613, "total_tasks": 32, "total_videos": 830, "total_chunks": 1, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:415" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/berkeley_fanuc_manipulation.

leju_robot_box_storage_parcel_i

leju_robot_box_storage_parcel_i 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 222… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_i.

robotwin2-agilex

Dataset from Oooer8. Tags: size_categories:1K<n<10K, format:imagefolder, modality:image, modality:video, library:datasets.

lerobot_rlbench_rvt_push_buttons_variation0

Dataset from tt1225. Tags: size_categories:n<1K, format:parquet, modality:tabular, modality:timeseries, modality:video.

Nemotron-RL-Agentic-Conversational-Tool-Use-Pivot-v1

Dataset Description: We created an RL dataset for conversational tool-use by utilizing existing expert tool-use trajectories. We pose each assistant step of the trajectory as a separate behavior cloning problem where the policy model is incentivized to match the tool call choices of the expert model. Each trajectory includes the use of tools for authentication, data lookup, servicing (i.e. booking reservations, changing them, getting discounts, etc), and more across 838 different… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Conversational-Tool-Use-Pivot-v1.

build_block_tower-advantages

Advantage Values for villekuosmanen/build_block_tower Pre-computed advantage values for offline RL training. Source Dataset: villekuosmanen/build_block_tower Value Model: villekuosmanen/rewact_build_block_tower_all_3 N-step lookahead: 50 Files This dataset contains per-episode parquet files with advantage values for each frame. Usage from pathlib import Path import pandas as pd # Load advantages for a specific episode advantage_df =… See the full description on the dataset page: https://huggingface.co/datasets/villekuosmanen/build_block_tower-advantages.

LeWAM_community_dataset

LeWAM Community Dataset A large-scale community-contributed robotics dataset for robot learning research featuring 321 datasets from the LeRobot community. Derived from Community Dataset v2, and converted to LeRobotDataset v3.0. The dataset is an artifact from the LeWAM project. Dataset Statistics Metric Value Total Datasets 321 Weighted Average FPS 30.0 Primary Tasks Manipulation, Pick & Place, Sorting Robot Types SO-100 and 101 (various colors)… See the full description on the dataset page: https://huggingface.co/datasets/ehalicki/LeWAM_community_dataset.

robotwin2-clean-lerobot

Dataset from Hoshipu. Tags: size_categories:100K<n<1M, format:parquet, modality:image, modality:timeseries, library:datasets.

push_block_isaaclab

Dataset from theconstruct-ai. Tags: size_categories:100K<n<1M, format:parquet, modality:tabular, modality:video, library:datasets.

hh-rlhf-49k-ja-single-turn

This dataset was created by automatically translating part of "Anthropic/hh-rlhf" into Japanese, and selected for single turn conversations.You can use this dataset for RLHF and DPO. hh-rlhf repository https://github.com/anthropics/hh-rlhf Anthropic/hh-rlhf https://huggingface.co/datasets/Anthropic/hh-rlhf

Nemotron-Cascade-2-RL-data

Dataset Description: The Nemotron-Cascade-2-RL dataset is a curated reinforcement learning (RL) dataset blend used to train Nemotron-Cascade-2-30B-A3B model. It includes instruction-following RL, multi-domain RL, on-policy distillation, and software engineering RL (SWE-RL) data. This dataset is ready for commercial use. The dataset contains the following subset: IF-RL Contains 45,879 training samples for instruction-following RL. Our curation process mainly resolves… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Cascade-2-RL-data.

LR_Robotics

Dataset from Latoff. Tags: size_categories:10K<n<100K, format:parquet, modality:tabular, modality:text, modality:video.

leju_robot_pass_the_cleaner_b

leju_robot_pass_the_cleaner_b 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home industry 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_pass_the_cleaner_b.

lerobot_rlbench_rvt_open_drawer_variation0

Dataset from tt1225. Tags: size_categories:n<1K, format:parquet, modality:tabular, modality:timeseries, modality:video.

leju_robot_box_storage_parcel_a

leju_robot_box_storage_parcel_a 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 493… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_a.

humanoid-everyday

Humanoid Everyday A Comprehensive Robotic Dataset for Open-World Humanoid Manipulation Overview Humanoid Everyday is a large-scale, diverse humanoid manipulation dataset designed for open-world robotic learning and embodied intelligence. It contains over 260 tasks across 7 major categories, covering dexterous manipulation, human–humanoid interaction, and locomotion-integrated activities.All data were collected through a human-supervised teleoperation pipeline, recording… See the full description on the dataset page: https://huggingface.co/datasets/Fizzi789/humanoid-everyday.

rlbench

Dataset from haibozzz. Tags: region:us.

PhysicalAI-Robotics-GR00T-GR1

GR1-100 92 Videos of a Fourier GR1-T2 robot performing various tasks in the lab from the third person perspective. This dataset is ready for commercial/non-commercial use. Dataset Owner(s): NVIDIA Corporation (GEAR Lab) Dataset Creation Date: May 1, 2025 License/Terms of Use: This dataset is governed by the Creative Commons Attribution 4.0 International License (CC-BY-4.0). This dataset was created using a Fourier GR1-T2 robot. Intended… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-GR00T-GR1.

openpi_datasets

openpi_datasets version folder data v0.1.0 put_the_box image_higt,image_left,state,action v0.2.0 pi0_put_the_yellow_mango_on_the_blue_plate image_higt,image_left,state,action

G1_WBT_Inspire_Put_Clothes_Into_Basket

Real Unitree G1 whole-body teleoperated dataset: the humanoid picks clothes and places them into a basket, using Inspire hands. Part of Unitree's laundry-pipeline data collection.

reachy-mini-dances-library

Dataset from pollen-robotics. Tags: size_categories:n<1K, format:json, modality:text, library:datasets, library:dask.

RoboTwin-Clean

Dataset from StarVLA. Tags: size_categories:n<1K, modality:video, library:datasets, library:mlcroissant, region:us.

robotwin_aloha_clean_hdf5

without depth data. processed data. generated from python scripts/process_data.py $task_name $setting $expert_data_num

terminology-sentences-sft-rlhf-test

Dataset from ChengSong. Tags: size_categories:100K<n<1M, format:parquet, modality:text, library:datasets, library:dask.

frankapickplace_200episodes

Dataset from Pidiya. Tags: size_categories:10K<n<100K, format:parquet, modality:tabular, modality:timeseries, modality:video.

MOSAIC_Dataset

MOSAIC Dataset Project Page | Paper | Code | Dataset | Model This repository releases the built-in MOSAIC multi-source motion dataset in the following paper: MOSAIC: Bridging the Sim-to-Real Gap in Generalist Humanoid Motion Tracking and Teleoperation with Rapid Residual Adaptation The dataset is organized into: Human motions stored in an AMASS-style format Unitree G1 motions retargeted from human motions and converted to NPZ for training/visualization It includes motions from:… See the full description on the dataset page: https://huggingface.co/datasets/BAAI-Humanoid/MOSAIC_Dataset.

kunkado

Kunnafonidilaw ka cadeau 🇲🇱 A messy‑real Bambara ASR corpus for developing modern speech models & code‑switch studies Quick Facts value Total duration 161.15 h Reviewed subset 39.3 h (≈ 25 %) Total segments 118 925 Languages Bambara (majority) • French (code‑switch) • misc. Arabic (translit) LICENSE CC‑BY‑SA 4.0 kunkado aims to mirror how Malians speak bambara today: fast, informal, and full of French code‑switching. We hope it fuels robust… See the full description on the dataset page: https://huggingface.co/datasets/RobotsMali/kunkado.

rlds_full

Dataset from czxlovesu03. Tags: region:us.

leju_robot_box_storage_parcel_j

leju_robot_box_storage_parcel_j 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 226… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_j.

VLA-OS-Dataset

Dataset Card This is the training dataset used in the paper VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models. Source Project Page: https://nus-lins-lab.github.io/vlaos/ Paper: https://arxiv.org/abs/2506.17561 Code: https://github.com/HeegerGao/VLA-OS Model: https://huggingface.co/Linslab/VLA-OS Usage Ensure you have installed git lfs: curl -s… See the full description on the dataset page: https://huggingface.co/datasets/Linslab/VLA-OS-Dataset.

ManiSkill2

ManiSkill2 Data Update: ManiSkill 3 has been released https://github.com/haosulab/ManiSkill/. It uses different datasets than ManiSkill2 so the data here is not expected to transfer over ManiSkill2 is a unified benchmark for learning generalizable robotic manipulation skills powered by SAPIEN. It features 20 out-of-box task families with 2000+ diverse object models and 4M+ demonstration frames. Moreover, it empowers fast visual input learning algorithms so that a CNN-based… See the full description on the dataset page: https://huggingface.co/datasets/haosulab/ManiSkill2.

computer-use

Computer Use Trajectories Successful computer-use agent trajectories collected on OSWorld tasks. Dataset Details Rows: 160 (one per task trajectory) Steps: 1,378 total across all trajectories (avg ~8.6 steps/task) Agent: Gemini 3 Flash Preview with linearized accessibility-tree grounding Score filter: Only trajectories with score = 1.0 (fully successful) Domains Domain Tasks Description chrome 21 Web browsing tasks in Google Chrome gimp 15 Image… See the full description on the dataset page: https://huggingface.co/datasets/markov-ai/computer-use.

G1_WBT_Inspire_Pickup_Pillow_MainCamOnly

Data Structure Observations observation.state.ee_state (12) End-effector states of the robot. Computed via forward kinematics (FK) from the root link to the left and right end-effectors. Includes the contribution of the waist. Represented as concatenated poses of both end-effectors. observation.state.hand_state (12 or 2) Finger states for both hands. The dimensionality depends on the hand type. Inspire Hand (range: 0.0 – 1.0, open → close)… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_WBT_Inspire_Pickup_Pillow_MainCamOnly.

RoboTwin_clean

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "aloha", "total_episodes": 2500, "total_frames": 549787, "total_tasks": 2409, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 15, "splits": { "train": "0:2500"}, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/SUZ-tsinghua/RoboTwin_clean.

robocasa-x-xp3k-pnp

RoboCasa-X XP-3K PnP Counter to Sink / PnP Sink to Counter Task: PnP Counter to Sink / PnP Sink to Counter TFDS builder id after download: robocasa_x_xp3k_pnp Robots in dataset: Panda, Panda-OG, Jaco Number of demos: 6000

guru-RL-92k-extra-info-compressed

Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective Note for this extra-info-compressed data version! The dataset provided in this repository is specifically intended for use with the latest release of VeRL (v0.4.0). Since VeRL rl_dataset.py processes datasets as datasets.Dataset, it is essential that the structure of all Parquet files remains fully consistent. This repository is designed to meet that requirement. In this repo, the… See the full description on the dataset page: https://huggingface.co/datasets/LLM360/guru-RL-92k-extra-info-compressed.

AgiBot-g1_robotic_arm_picks_up_battery

AgiBot-g1_robotic_arm_picks_up_battery 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: ruantong_a2d | Codebase Version: v2.1 End-Effector Type: two_finger_gripper 🏠 Scene Types This dataset covers the following scene types: factory 🤖 Atomic Actions This dataset includes the following atomic actions: place pick grasp 📊 Dataset Statistics Metric Value… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/AgiBot-g1_robotic_arm_picks_up_battery.

DeepEyesV2_RL

Dataset from honglyhly. Tags: size_categories:n<1K, format:parquet, modality:text, library:datasets, library:pandas.

data

Dataset Card for BEHAVIOR Robot Suite (BRS) Data This dataset provides robotic trajectories for five real-world household tasks. These tasks are: Clean house after a wild party; Clean the toilet; Take trash outside; Put items onto shelves; Lay clothes out. These data are first collected and used in the paper BEHAVIOR Robot Suite: Streamlining Real-World Whole-Body Manipulation for Everyday Household Activities . Dataset Details Dataset Sources… See the full description on the dataset page: https://huggingface.co/datasets/behavior-robot-suite/data.

lerobot-simulation-koch-arm-to-the-shelf-01

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "so_arm_101", "total_episodes": 1, "total_frames": 478, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "fps": 30, "splits": { "train": "0:1" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/OpenGraphLabs-Research/lerobot-simulation-koch-arm-to-the-shelf-01.

VPPO_ViRL39K_train

Dataset Card for VPPO_ViRL39K_train Dataset Details Dataset Description This dataset is the official training split used to fine-tune the VPPO-7B and VPPO-32B models presented in our paper, "Spotlight on Token Perception for Multimodal Reinforcement Learning". This is a direct copy of the TIGER-Lab/ViRL39K dataset. We have isolated it here to ensure the exact version used in our experiments is publicly available, guaranteeing reproducibility for our research.… See the full description on the dataset page: https://huggingface.co/datasets/chamber111/VPPO_ViRL39K_train.

trace_robotwin_rand_val

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "aloha", "total_episodes": 2400, "total_frames": 436031, "total_tasks": 2359, "chunks_size": 1000, "fps": 50, "splits": { "train": "0:2400" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/quincyu/trace_robotwin_rand_val.

PWBC

PWBC Motion Dataset (TextOpTracker-Ready) Overview This dataset provides processed motion data for the PWBC project. Directory structure The dataset is organized as: converted_from_holosoma/ sub*_largebox_*_original_mj_w_obj_omnirt/ motion.npz Each motion is stored in its own folder and contains one motion.npz file. Intended usage Use this dataset as artifacts/converted_from_holosoma in TextOpTracker. Example (train/play side):… See the full description on the dataset page: https://huggingface.co/datasets/Haongchen/PWBC.

rlvr_gsm8k_zs

Dataset from ai2-adapt-dev. Tags: size_categories:1K<n<10K, format:parquet, modality:text, library:datasets, library:pandas.

ScanBot

ScanBot: Towards Intelligent Surface Scanning in Embodied Robotic Systems This dataset is presented in the paper ScanBot: Towards Intelligent Surface Scanning in Embodied Robotic Systems. 🧠 Dataset Summary ScanBot is a dataset for instruction-conditioned, high-precision surface scanning with robots. Unlike existing datasets that focus on coarse tasks like grasping or navigation, ScanBot targets industrial laser scanning, where sub-millimeter accuracy and parameter… See the full description on the dataset page: https://huggingface.co/datasets/ed1son/ScanBot.

arm-robot

Dataset from lashik. Tags: size_categories:1K<n<10K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

trajectory-prediction-waymo

Waymo Trajectory Prediction Dataset Description This dataset contains preprocessed trajectory prediction samples for autonomous driving research, formatted for use with DiscoBench's TrajectoryPrediction task. Original Dataset: Waymo Open Motion Dataset Number of Samples: 850 Format: Pickle files with numpy arrays Task: Multi-modal trajectory prediction Dataset Structure Each sample is a pickle file containing: obj_trajs (32, 21, 2): Past trajectories of… See the full description on the dataset page: https://huggingface.co/datasets/saeedrmd/trajectory-prediction-waymo.

pitch-control-tracking

Pitch Control Tracking Data Per-player per-frame pitch control values from ~38 million rows of professional soccer tracking data across 20 matches from three providers. Computed using the Spearman (2017) physics-based model — each row contains one player's position, velocity, and the home-team control probability at that location. Part of the (Right! Luxury!) Lakehouse soccer analytics platform. Quick Start from datasets import load_dataset ds =… See the full description on the dataset page: https://huggingface.co/datasets/luxury-lakehouse/pitch-control-tracking.

EA-Temporal

Dataset from EmbodiedArena. Tags: size_categories:1K<n<10K, format:imagefolder, modality:image, library:datasets, library:mlcroissant.

leju_robot_hotel_services_d

leju_robot_hotel_services_d 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 324… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_hotel_services_d.

visual_distracting_control_suite

Visual Distracting Control Suite Benchmark This dataset contains expert trajectories generated by a Proximal Policy Optimization (PPO) reinforcement learning agent trained on 4 environments of the Distracting Control Suite. For each environment we collect data with different levels of distraction, which we define below, and masks for the agent. Levels of distraction: None: Vanilla DeepMind Control Suite without visual distractions. The environment uses the default static background… See the full description on the dataset page: https://huggingface.co/datasets/EpicPinkPenguin/visual_distracting_control_suite.

legal-rag-bench

Legal RAG Bench ‍⚖️ Legal RAG Bench by Isaacus is a reasoning-intensive benchmark for assessing the end-to-end, real-world performance of production-grade legal RAG systems. Legal RAG Bench is composed of 4,876 passages sampled from the Judicial College of Victoria’s Criminal Charge Book alongside 100 complex, handwritten questions demanding expert-level knowledge of Victorian criminal law and procedure to be answered correctly. Legal RAG Bench is the first open dataset for the… See the full description on the dataset page: https://huggingface.co/datasets/isaacus/legal-rag-bench.

OmniRetarget_Dataset

OmniRetarget Dataset: Humanoid Loco-Manipulation & Scene Interaction Paper | Project Page This dataset contains motion trajectories of a G1 humanoid robot interacting with objects and complex terrains. It was generated by OMNIRETARGET, an interaction-preserving data generation engine that produces high-quality, kinematically feasible trajectories free of common artifacts like foot-skating and penetration. Dataset Structure Due to licensing restrictions, we cannot… See the full description on the dataset page: https://huggingface.co/datasets/omniretarget/OmniRetarget_Dataset.

conq_hose_manipulation

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "unknown", "total_episodes": 139, "total_frames": 8277, "total_tasks": 3, "chunks_size": 1000, "fps": 30, "splits": { "train": "0:139" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/conq_hose_manipulation.

RL_ManiObj_Aug

Dataset from xinjjj. Tags: size_categories:n<1K, format:imagefolder, modality:3d, modality:image, modality:video.

leju_robot_box_storage_parcel_f

leju_robot_box_storage_parcel_f 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 497… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_box_storage_parcel_f.

roboturk

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "robot_type": "unknown", "total_episodes": 1995, "total_frames": 187507, "total_tasks": 3, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:1995" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4"… See the full description on the dataset page: https://huggingface.co/datasets/lerobot/roboturk.

leju_robot_hotel_services_c

leju_robot_hotel_services_c 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp pick place 📊 Dataset Statistics Metric Value Total Episodes 334… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_hotel_services_c.

VideoRFT-Data

🎥 VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning Paper | Code | CoT Dataset (on Hugging Face) | RL Dataset (on Hugging Face) | Models (on Hugging Face) Abstract Reinforcement fine-tuning (RFT) has shown great promise in achieving humanlevel reasoning capabilities of Large Language Models (LLMs), and has recently been extended to MLLMs. Nevertheless, reasoning about videos, which is a fundamental aspect of human intelligence… See the full description on the dataset page: https://huggingface.co/datasets/QiWang98/VideoRFT-Data.

mcp-agent-trajectory-benchmark

MCP Agent Trajectory Benchmark A benchmark dataset of 49 MCP (Model Context Protocol) agent trajectories (38 single-pass + 11 multi-conv) with complete tool-use traces in the ATIF v1.2 (Agent Trajectory Interchange Format) format. Each agent operates in a distinct business domain with custom tools, realistic user conversations, and full execution traces. Designed for training and evaluating tool-use / function-calling capabilities of LLMs. Overview Item Details… See the full description on the dataset page: https://huggingface.co/datasets/obaydata/mcp-agent-trajectory-benchmark.

leju_robot_moving_parts_j

leju_robot_moving_parts_j 📋 Overview This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot. Robot Type: leju_robot | Codebase Version: v2.1 End-Effector Type: five_finger_hand 🏠 Scene Types This dataset covers the following scene types: home 🤖 Atomic Actions This dataset includes the following atomic actions: grasp place pick 📊 Dataset Statistics Metric Value Total Episodes 153… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/leju_robot_moving_parts_j.

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.

robotwin-pi0-supp-data

Dataset from RogersPyke. Tags: size_categories:10K<n<100K, format:parquet, modality:image, modality:timeseries, modality:video.

atari-chopper_command-dataset

Atari-Chopper Command Dataset This is a large dataset of 10M video frames and actions collected from the Chopper Command atari environment (Bellemare et al., 2012) in order to train world models.The dataset enables reproducible, large-scale experiments in action-conditioned video prediction. It is meant to be used with Jasmine, our JAX-based world modeling codebase. Dataset Summary Environment: Atari Learning Environment Frames: 10 million Resolution: 84 × 84… See the full description on the dataset page: https://huggingface.co/datasets/p-doom/atari-chopper_command-dataset.

merged_strawberries_tictactoe

SO101 Merged Strawberries & Tic-Tac-Toe Dataset Dataset Description This dataset contains robot manipulation episodes combining two distinct tasks using SO101 robot. The dataset includes 27 episodes of robot manipulation with multi-view camera observations. Tasks Task 0: Pick up strawberries and put them in a tray Task 1: Pick up an X piece and place it on the tic-tac-toe board to complete the winning line Robot: SO101 Follower robot with 6 degrees of freedom… See the full description on the dataset page: https://huggingface.co/datasets/kinam0252/merged_strawberries_tictactoe.

OceanGym

🌊 OceanGym 🦾 A Benchmark Environment for Underwater Embodied Agents 🌐 Home Page 📄 ArXiv Paper 🤗 Hugging Face ☁️ Google Drive ☁️ Baidu Drive OceanGym is a high-fidelity embodied underwater environment that simulates a realistic ocean setting with diverse scenes. As illustrated in figure, OceanGym establishes a robust benchmark for evaluating autonomous agents through a series of challenging tasks, encompassing various perception analyses and decision-making… See the full description on the dataset page: https://huggingface.co/datasets/zjunlp/OceanGym.

Med-eval-logs

MED Evaluation Logs Evaluation logs for the MED (Measure-Explain-Diagnose) framework analyzing vision tool-use reinforcement learning. Paper What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom 📄 Paper: arXiv:2602.01334 💻 Code: github.com/GAIR-NLP/Med Dataset Description This dataset contains evaluation results from vision tool-use RL experiments, tracking model performance across… See the full description on the dataset page: https://huggingface.co/datasets/Med2026/Med-eval-logs.

teleoperation-pick-and-place

Dataset from axel-kaliff. Tags: region:us.

dw_pose_controlnet

Important Notice This is a copy of raulc0399/open_pose_controlnet replacing openpose conditioning images with DW pose. Dataset for training controlnet models with conditioning images as Human Pose the entries have been taken from this dataset ptx0/photo-concept-bucket the open pose images have been generated with controlnet_aux for the scripts to download the files, generate the openpose and the dataset please refer to: raulc0399/dataset_scripts Openpose to… See the full description on the dataset page: https://huggingface.co/datasets/dimitribarbot/dw_pose_controlnet.

OmniAction-LIBERO

RoboOmni: Proactive Robot Manipulation in Omni-modal Context 📖 arXiv Paper (Accepted to ICLR 2026 🎉) | 🌐 Website | 🤗 Model | 🤗 Dataset | 🛠️ Github | Recent advances in Multimodal Large Language Models (MLLMs) have driven rapid progress in Vision–Language–Action (VLA) models for robotic manipulation. Although effective in many scenarios, current approaches largely rely on explicit instructions, whereas in real-world interactions, humans rarely issue… See the full description on the dataset page: https://huggingface.co/datasets/OpenMOSS-Team/OmniAction-LIBERO.

gen2_profile

Dataset from ultra-ai-robotics. Tags: size_categories:10M<n<100M, format:parquet, modality:tabular, modality:text, modality:timeseries.

VPPO_MMK12_validation

Dataset Card for VPPO_MMK12_validation Dataset Details Dataset Description This dataset is the official validation split used to fine-tune the VPPO-7B and VPPO-32B models presented in our paper, "Spotlight on Token Perception for Multimodal Reinforcement Learning". This is a direct copy of the test split of FanqingM/MMK12 dataset. We have isolated it here to ensure the exact version used in our experiments is publicly available, guaranteeing reproducibility for… See the full description on the dataset page: https://huggingface.co/datasets/chamber111/VPPO_MMK12_validation.

isaaclab-arena-franka-dataset

Dataset from umegan. Tags: size_categories:n<1K, modality:video, library:datasets, library:mlcroissant, region:us.

VisionReasoner_multi_object_7k_840

Seg-Zero: Reasoning-Chain Guided Segmentation Dataset This repository contains a training dataset for the models presented in "Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement" and "VisionReasoner: Unified Visual Perception and Reasoning via Reinforcement Learning". Seg-Zero is a novel framework that demonstrates remarkable generalizability and derives explicit chain-of-thought reasoning through cognitive reinforcement. It introduces a decoupled architecture… See the full description on the dataset page: https://huggingface.co/datasets/Ricky06662/VisionReasoner_multi_object_7k_840.

G1_Dex1_MountCameraRedGripper_Dataset

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "Unitree_G1_Gripper", "total_episodes": 201, "total_frames": 172649, "total_tasks": 1, "total_videos": 804, "total_chunks": 1, "chunks_size": 1000, "fps": 30, "splits": { "train": "0:201"}, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/unitreerobotics/G1_Dex1_MountCameraRedGripper_Dataset.

trajectory-prediction-argoverse2

Dataset from saeedrmd. Tags: region:us.

CausalVerse_Image

CausalVerse Image Dataset This dataset contains two families of splits: Physics splits: Fall, Refraction, Slope, Spring Static image generation: scene1, scene2, scene3, scene4 All splits share the same columns: image (binary image; datasets.Image) render_path (string; original image filename/path) metavalue (string; per-sample metadata; schema varies by split) Paper: CausalVerse: Benchmarking Causal Representation Learning with Configurable High-Fidelity Simulations Project… See the full description on the dataset page: https://huggingface.co/datasets/CausalVerse/CausalVerse_Image.

dronedata_new_lerobot

This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v2.1", "robot_type": "uav", "total_episodes": 799, "total_frames": 79900, "total_tasks": 2, "total_videos": 0, "total_chunks": 1, "chunks_size": 1000, "fps": 10, "splits": { "train": "0:799" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path":… See the full description on the dataset page: https://huggingface.co/datasets/Celina717/dronedata_new_lerobot.

fractal-rlds

Dataset Structure These datasets are used for MemoryVLA training. This is the standard setting and can be directly used for other models as well.All data follow the RLDS format from the Fractal dataset. fractal20220817_data — 130k+ episodes, google robot NOTE: The upload has not been fully completed yet.

PhysicalAI-Robotics-NuRec

Dataset Description The Physical AI NuRec dataset seeks to empower robotic researchers to build the next generation of physical AI based end-to-end robotic models. This dataset includes various 3DGUT in USD files that can be loaded in Isaac Sim. Some datasets also include a mesh and occupancy map. The Mesh components are used for collision detection while the 3DGUT components provide realistic rendering. The asset can also be used with Isaac Sim Extensions like MobilityGen for AMR… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-NuRec.

RLVR-GSM-MATH-IF-Mixed-Constraints

GSM/MATH/IF Data - RLVR Formatted Note that this collection is licensed under ODC-BY-1.0 license; different licenses apply to subsets of the data. This dataset contains data formatted for use with open-instruct - specifically reinforcement learning with verifiable rewards. It was used to train the final Tulu 3 models with RL, and contains the following subsets: GSM8k (7,473 samples): The GSM8k train set formatted for use with RLVR and open-instruct. MIT License. MATH (7,500… See the full description on the dataset page: https://huggingface.co/datasets/allenai/RLVR-GSM-MATH-IF-Mixed-Constraints.

Skywork-OR1-RL-Data

🤔 Skywork-OR1-RL-Data 🔥 News April 15, 2025: We are excited to release our RL training dataset Skywork-OR1-RL-Data For our final training phase, we filtered problems based on their difficulty levels (0-16, higher values indicate harder problems) relative to specific model variants (DeepSeek-R1-Distill-Qwen-{1.5,7,32}B. For each model variant, we excluded problems with difficulty values of 0 and 16 specific to that model from its training data.You can… See the full description on the dataset page: https://huggingface.co/datasets/Skywork/Skywork-OR1-RL-Data.

Eurus-2-RL-Data

Eurus-2-RL-Data Links 📜 Paper 📜 Blog 🤗 PRIME Collection Introduction Eurus-2-RL-Data is a high-quality RL training dataset of mathematics and coding problems with outcome verifiers (LaTeX answers for math and test cases for coding). For math, we source from NuminaMath-CoT. The problems span from Chinese high school mathematics to International Mathematical Olympiad competition questions. For coding, we source from APPS, CodeContests, TACO, and Codeforces.… See the full description on the dataset page: https://huggingface.co/datasets/PRIME-RL/Eurus-2-RL-Data.

qwen3-32b-rl-step200-r2e-gym-trajectories

Dataset from AxT-dev. Tags: region:us.

rlbench

Dataset from simpack. Tags: region:us.

Dolci-Think-RL-7B

Dolci-Think-RL-7B Dataset Summary Dolci-Think-RL-7B is the reinforcement learning dataset used to train the Olmo-3-7B-Think model.It contains 102,014 prompts designed to elicit deep reasoning across: Math Coding Precise Instruction Following General Chat It blends high-quality curated sources with filtering designed for deliberate reasoning. Dataset Composition Total Samples: 102,014 Original Dataset Contribution… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-Think-RL-7B.

carla-autopilot-multimodal-dataset

CARLA Autopilot Multimodal Dataset This dataset contains synchronized multimodal driving data collected in the CARLA simulator using the autopilot feature. It provides RGB images from multiple cameras, semantic segmentation, LiDAR point clouds, 2D bounding boxes, and ego-vehicle state/control signals across varied weather, maps, and traffic densities. The dataset is designed for research in autonomous driving, sensor fusion, imitation learning, and self-driving evaluation.… See the full description on the dataset page: https://huggingface.co/datasets/immanuelpeter/carla-autopilot-multimodal-dataset.

lanternfly-data

CMU Campus Spotted Lanternfly Geolocated Dataset Dataset Description This dataset contains geolocated photographs of spotted lanternflies (Lycorma delicatula) collected around Carnegie Mellon University campus in Pittsburgh, Pennsylvania. The dataset was created as part of a class project (Project 1) focused on invasive species monitoring and geospatial data collection. Each entry includes high-resolution photographs with precise GPS coordinates, timestamps, and metadata… See the full description on the dataset page: https://huggingface.co/datasets/rlogh/lanternfly-data.

PhysicalAI-Robotics-GraspGen

GraspGen: Scaling Sim2Real Grasping GraspGen is a large-scale simulated grasp dataset for multiple robot embodiments and grippers. We release over 57 million grasps, computed for a subset of 8515 objects from the Objaverse XL (LVIS) dataset. These grasps are specific to three grippers: Franka Panda, the Robotiq-2f-140 industrial gripper, and a single-contact suction gripper (30mm radius). Dataset Format The dataset is released in the WebDataset format. The… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicalAI-Robotics-GraspGen.

IsaacLab-Arena_RoboTwin_Data

Dataset from littleTang. Tags: size_categories:n<1K, modality:video, library:datasets, library:mlcroissant, region:us.

embodied_reasoner

Embodied-Reasoner Dataset Dataset Overview Embodied-Reasoner is a multimodal reasoning dataset designed for embodied interactive tasks. It contains 9,390 Observation-Thought-Action trajectories for training and evaluating multimodal models capable of performing complex embodied tasks in indoor environments. Key Features 📸 Rich Visual Data: Contains 64,000 first-person perspective interaction images🤔 Deep Reasoning Capabilities: 8 million thought… See the full description on the dataset page: https://huggingface.co/datasets/zwq2018/embodied_reasoner.

0318_franka_teleop

Dataset from periphanes. Tags: size_categories:n<1K, modality:video, library:datasets, library:mlcroissant, region:us.

Nemotron-RL-Agentic-Function-Calling-Pivot-v1

Dataset Description: This is a RL dataset for general function-calling by utilizing existing expert tool-use trajectories. We pose each assistant step of the trajectory as a separate behavior cloning problem where the policy model is incentivized to match the tool call choices of the expert model. This dataset is released as part of NVIDIA NeMo Gym, a framework for building reinforcement learning environments to train large language models. NeMo Gym contains a growing collection of… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Function-Calling-Pivot-v1.

RoboMIND-lerobot

RoboMIND-lerobot (Unofficial Partial Conversion) This is an unofficial conversion of a subset of the RoboMIND dataset into LeRobot v3.0 format. This repository does NOT contain the full RoboMIND dataset. Only a selection of embodiments has been converted. For the complete official dataset, please visit: x-humanoid-robomind/RoboMIND. Included Subsets Config Name Robot Episodes Frames Tasks Cameras Arm Config agilex_3rgb AgileX 8,638 5,431,759 53 front… See the full description on the dataset page: https://huggingface.co/datasets/Traly/RoboMIND-lerobot.

metaworld-expert

Dataset from CooperL. Tags: region:us.