datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
RoboTwin2.0-Depthrobotwin2.0-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.robotwin_aloha_clean_hdf5without depth data. processed data. generated from python scripts/process_data.py $task_name $setting $expert_data_num
robotwin_3d_text_embeds_cache
RoboTwin 2.0 3D — T5 Text Embedding Cache
Precomputed UMT5-XXL text embeddings for the
1,039,891 unique task prompts of the RoboTwin 2.0 3D dataset
(flex-pi/robotwin_3d), as consumed by
Wan2.2-TI2V-5B. Precomputing these costs substantial GPU time; this cache skips it.
Cache key
Task strings from meta/tasks.jsonl are wrapped in a fixed template before encoding, and the cache
key is the sha256 of that templated prompt:
DEFAULT_PROMPT = "A video recorded from a… See the full description on the dataset page: https://huggingface.co/datasets/flex-pi/robotwin_3d_text_embeds_cache.robotwin_rolloutRoboTwin_adjust_bottle_randomizedThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "aloha",
"total_episodes": 500,
"total_frames": 68537,
"total_tasks": 424,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 15,
"splits": {
"train": "0:500"
},
"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/suz22/RoboTwin_adjust_bottle_randomized.RoboTwin_instruct-pix2pix
数据集介绍
Contributor: Shenghao Yang, Yimou Wu
数据简介
本数据集是通过https://github.com/TianxingChen/RoboTwin [1]单臂机器人模拟器在 block_hammer_beat, block_handover, blocks_stack_easy 三个任务上间隔50帧采样得到的，其数据规模如下:
block_hammer_beat block_handover blocks_stack_easy
train 100 100 100
val 10 10 10
test 10 10 10
整体数据规模如下:
train val test
num 300 30 30… See the full description on the dataset page: https://huggingface.co/datasets/YimouWu/RoboTwin_instruct-pix2pix.RoboTwin-Randomized-targzfastwam-robotwin2-tarballs
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/huiliu123/fastwam-robotwin2-tarballs.robotwin-47robotwin-stack-blocks-three-rollouts
RoboTwin stack_blocks_three — Wan2.2 TI2V Rollouts
160 text+image-to-video rollouts (10 initial conditions × 16 random seeds) for the
stack_blocks_three task from RoboTwin, generated with the Wan2.2 TI2V (5B)
diffusion model fine-tuned with a merged Vidar LoRA adapter, and scored with the
blocks_stack_v2 reward (SAM3 object tracking + IDM inverse-dynamics + FK
gripper ↔ block position matching).
Companion to the EmbodiedVideoRL / DanceGRPO
reward-model work.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/VincentNi/robotwin-stack-blocks-three-rollouts.eden_robotwin2robotwin-action-prediction-dataset
Robotic Action Prediction Dataset
Dataset Description
This dataset contains triplets of (current observation, action instruction, future observation) for training models to predict future frames of robotic actions.
Dataset Structure
Data Fields
current_frame: Input image (RGB) of the current observation
instruction: Textual description of the action to perform
future_frame: Target image (RGB) showing the expected outcome 50 frames later… See the full description on the dataset page: https://huggingface.co/datasets/bryandts/robotwin-action-prediction-dataset.robotwin-blocks-ranking-rgb-rollouts
RoboTwin blocks_ranking_rgb — Wan2.2 TI2V Rollouts
160 text+image-to-video rollouts (10 initial conditions × 16 random seeds) for the
blocks_ranking_rgb task from RoboTwin, generated with the Wan2.2 TI2V (5B)
diffusion model fine-tuned with a merged Vidar LoRA adapter, and scored with the
blocks_ranking_v2 reward (SAM3 object tracking + IDM inverse-dynamics + FK
gripper ↔ block position matching).
Companion to the EmbodiedVideoRL / DanceGRPO
reward-model work.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/VincentNi/robotwin-blocks-ranking-rgb-rollouts.RoboTwin_assetrobotwin-icil-aloha-clean
RoboTwin ICIL Aloha Clean
Training data for RoboTwin-ICIL, a
one-demonstration in-context imitation learning benchmark, in robotwin-icil-hdf5-multiview-v1 format.
Contents
50 tasks x 50 trajectories = 2,500 aloha-agilex trajectories, all of them training data. Each
episode has an HDF5 trajectory, three aligned 320x240 H.264 videos (head, left wrist, right wrist)
and an instruction JSON. The HDF5 holds joint commands, actions, end-effector poses, camera
calibration… See the full description on the dataset page: https://huggingface.co/datasets/robotensor/robotwin-icil-aloha-clean.robotwin-act
MetaAct / Robotwin ACT
This repository hosts the public release artifacts for the MetaAct Robotwin ACT project.
Files
File
Size
SHA256
Data/robotwin-act-data.tar.gz.part001 - Data/robotwin-act-data.tar.gz.part007
128,050,721,979 bytes total
6B10200A8800B70700084F9437175151886D349B1270A4A659172F0C548A6284 after merge
Docker image archives for the same release will be added to this repository later.
Download
Install the Hugging Face… See the full description on the dataset page: https://huggingface.co/datasets/MetaAct/robotwin-act.worldarena2026-robotwin-data
WorldArena-RoboTwin Dataset
Dataset: DavidxWang/worldarena2026-robotwin-dataProvider: davidwang (WorldArena Challenge 2026 Track 1 team)Last Updated: 2026-05-04Downstream Model: ABot-PhysWorld A2V VACE Training
Overview
This dataset provides action-conditioned robot manipulation videos synthesized by RoboTwin 2.0 simulation, formatted specifically for ABot-PhysWorld A2V (Action-to-Video) VACE training. It covers all 50 task categories from the CVPR 2026 WorldArena… See the full description on the dataset page: https://huggingface.co/datasets/DavidxWang/worldarena2026-robotwin-data.RoboTwin_adjust_bottle_cleanThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "aloha",
"total_episodes": 50,
"total_frames": 6902,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 15,
"splits": {
"train": "0:50"
},
"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/suz22/RoboTwin_adjust_bottle_clean.RoboTwinrobotwinrobotwin50-vae-latentsPix2Pix_RoboTwinrobotwinrobotwin_stack_blocks_two_clean50_2step_20260425
RobotWin Stack Blocks Two Clean50 2-Step Cache
This dataset repo contains a compressed latent-cache artifact for Video2VLA:
Archive: robotwin_stack_blocks_two_clean50_2step_20260425.tar.gz
Extracted directory: data/cache/robotwin_stack_blocks_two_clean50_2step_20260425
File count: 3757 .pt files
Sample latent shape: (1, 32, 4, 15, 20)
Dtype: torch.bfloat16
Representation: latent
This cache corresponds to the RobotWin stack_blocks_two clean50 setup using the 2-step denoising… See the full description on the dataset page: https://huggingface.co/datasets/RoMALab/robotwin_stack_blocks_two_clean50_2step_20260425.PIx2Pix_RoboTwin_imageRoboTwin_unified_statsRoboTwin-pi0-datarobotwin_lerobot_selectrobotwin_15_parquet