| { |
| "charts": [ |
| { |
| "path": "assets/charts/data_explorer_scope_ladder.svg", |
| "question": "How do the public sample, selected 128 episodes, and Hugging Face gated full dataset version differ in scale?", |
| "title": "Scope ladder" |
| }, |
| { |
| "path": "assets/charts/data_explorer_sample_feature_modalities.svg", |
| "question": "Which modality groups dominate the one-sample task input?", |
| "title": "Public sample feature dimensions" |
| }, |
| { |
| "path": "assets/charts/data_explorer_sample_action_distribution.svg", |
| "question": "Which action labels occupy the most 20-frame windows in the sample?", |
| "title": "Public sample action distribution" |
| }, |
| { |
| "path": "assets/charts/data_explorer_selected128_split_rows.svg", |
| "question": "How many rows are available per selected-128 export and split?", |
| "title": "Selected-128 split rows" |
| }, |
| { |
| "path": "assets/charts/data_explorer_full_file_composition.svg", |
| "question": "What file types dominate the Hugging Face gated full-dataset metadata?", |
| "title": "Hugging Face full dataset file composition" |
| } |
| ], |
| "full_hf_dataset": { |
| "annotation_file_size_summary": { |
| "count": 12103, |
| "max_bytes": 2000432556, |
| "max_human": "1.86 GiB", |
| "mean_bytes": 1830083732, |
| "mean_human": "1.70 GiB", |
| "median_bytes": 1969457752, |
| "median_human": "1.83 GiB", |
| "min_bytes": 6687192, |
| "min_human": "6.38 MiB", |
| "p25_bytes": 1869556164, |
| "p75_bytes": 1989975784 |
| }, |
| "basename_counts": { |
| "README.md": 1, |
| "annotation.hdf5": 12103, |
| "fisheye_cam0.mp4": 12102, |
| "fisheye_cam1.mp4": 12102, |
| "fisheye_cam2.mp4": 12102, |
| "fisheye_cam3.mp4": 12102, |
| "stereo_left.mp4": 12102, |
| "stereo_right.mp4": 12102, |
| "visualization.rrd": 541 |
| }, |
| "card_data": { |
| "extra_gated_button_content": "I have signed the access control agreement", |
| "extra_gated_description": "Access is reviewed manually and is limited to approved non-commercial use. Completion of an external agreement-signing step may be required before approval. Please sign the agreement via [DocuSign](https://ropedia.docsend.com/view/ra7ej7gs6s98sw87)", |
| "extra_gated_heading": "Request controlled access to Xperience-10M", |
| "language": [ |
| "en" |
| ], |
| "license": "other", |
| "pretty_name": "Xperience-10M", |
| "size_categories": [ |
| "1M<n<10M" |
| ], |
| "tags": [ |
| "egocentric", |
| "first-person", |
| "multimodal", |
| "3d", |
| "4d", |
| "embodied-ai", |
| "robotics", |
| "human-motion", |
| "mocap", |
| "imu", |
| "audio", |
| "depth", |
| "captions", |
| "video" |
| ], |
| "task_categories": [ |
| "video-classification", |
| "image-to-text", |
| "depth-estimation", |
| "robotics" |
| ] |
| }, |
| "complete_episode_training_size_summary": { |
| "count": 12102, |
| "max_bytes": 2721499520, |
| "max_human": "2.53 GiB", |
| "mean_bytes": 2237801505, |
| "mean_human": "2.08 GiB", |
| "median_bytes": 2365869284, |
| "median_human": "2.20 GiB", |
| "min_bytes": 8158968, |
| "min_human": "7.78 MiB", |
| "p25_bytes": 2286228606, |
| "p75_bytes": 2412829381 |
| }, |
| "episode_count_per_session_summary": { |
| "count": 803, |
| "max": 51, |
| "mean": 15.07, |
| "median": 11, |
| "min": 1, |
| "p25": 8, |
| "p75": 13 |
| }, |
| "episode_size_summary": { |
| "count": 12103, |
| "max_bytes": 5108045286, |
| "max_human": "4.76 GiB", |
| "mean_bytes": 2318233840, |
| "mean_human": "2.16 GiB", |
| "median_bytes": 2369826102, |
| "median_human": "2.21 GiB", |
| "min_bytes": 8158968, |
| "min_human": "7.78 MiB", |
| "p25_bytes": 2292986495, |
| "p75_bytes": 2420801956 |
| }, |
| "file_type_counts": { |
| ".hdf5": 12103, |
| ".md": 1, |
| ".mp4": 72612, |
| ".rrd": 541 |
| }, |
| "gated": "manual", |
| "incomplete_episode_records": [ |
| { |
| "episode_id": "ep1", |
| "episode_path": "dc3f4139-f499-4de7-b057-e25b7dfb2d83/ep1", |
| "file_count": 1, |
| "has_all_six_videos": false, |
| "has_annotation": true, |
| "has_fisheye_cam0": false, |
| "has_visualization_rrd": false, |
| "is_complete": false, |
| "is_degraded_valid": false, |
| "missing_required_files": [ |
| "fisheye_cam0.mp4", |
| "fisheye_cam1.mp4", |
| "fisheye_cam2.mp4", |
| "fisheye_cam3.mp4", |
| "stereo_left.mp4", |
| "stereo_right.mp4" |
| ], |
| "top_level_session": "dc3f4139-f499-4de7-b057-e25b7dfb2d83", |
| "total_bytes": 1418232696, |
| "training_bytes_excluding_visualization_rrd": 1418232696, |
| "video_count": 0 |
| } |
| ], |
| "last_modified": "2026-04-21T05:03:45+00:00", |
| "metadata_note": "The official dataset is gated on Hugging Face. These full-corpus figures use authenticated Hugging Face Hub file metadata for the HF-hosted dataset version only; they do not inspect private row content or redistribute raw MP4/HDF5/RRD files.", |
| "pilot_scale_estimates": { |
| "all_complete_episodes_windows_at_256_each": 3098112, |
| "episode_100_windows_at_256_each": 25600, |
| "episode_32_windows_at_256_each": 8192, |
| "episode_500_windows_at_256_each": 128000, |
| "mean_based_32_episode_training_bytes": 71609648160, |
| "mean_based_32_episode_training_human": "66.69 GiB", |
| "median_based_32_episode_training_bytes": 75707817088, |
| "median_based_32_episode_training_human": "70.51 GiB", |
| "windows_per_episode": 256 |
| }, |
| "repo_id": "ropedia-ai/xperience-10m", |
| "repo_sha": "ce943cf271a758b60240084892d05cf6dc12dd90", |
| "summary": { |
| "all_complete_episode_training_bytes_excluding_visualization_rrd": 27081873816281, |
| "all_complete_episode_training_human_excluding_visualization_rrd": "24.63 TiB", |
| "annotation_hdf5_count": 12103, |
| "complete_episode_count": 12102, |
| "complete_episode_pct": 99.9917, |
| "complete_sessions": 802, |
| "degraded_valid_episode_count": 12102, |
| "degraded_valid_episode_pct": 99.9917, |
| "degraded_valid_sessions": 802, |
| "episode_like_folder_count": 12103, |
| "file_count_excluding_gitattributes": 85257, |
| "mp4_count": 72612, |
| "sibling_count": 85258, |
| "top_level_session_count": 804, |
| "total_bytes_from_file_metadata": 28057584187079, |
| "total_human_from_file_metadata": "25.52 TiB", |
| "training_bytes_excluding_visualization_rrd": 27083292060675, |
| "training_human_excluding_visualization_rrd": "24.63 TiB", |
| "visualization_rrd_bytes": 974292126404, |
| "visualization_rrd_count": 541, |
| "visualization_rrd_human": "907.38 GiB" |
| }, |
| "video_count_histogram": { |
| "0": 1, |
| "6": 12102 |
| } |
| }, |
| "generated_at_utc": "2026-06-23T11:37:29Z", |
| "public_sample": { |
| "dataset": { |
| "episode_id": "xperience-10m-sample", |
| "hosting_policy": "Raw files are streamed or downloaded from the official public sample dataset. This task-suite repo mirrors derived artifacts and metadata only.", |
| "license": "cc-by-nc-4.0", |
| "repo_commit": "2d80b5af8085a18ffe702e1420968308007f6204", |
| "repo_id": "ropedia-ai/xperience-10m-sample", |
| "repo_url": "https://huggingface.co/datasets/ropedia-ai/xperience-10m-sample" |
| }, |
| "feature_blocks": [ |
| { |
| "dim": 1092, |
| "display": "Body Joints", |
| "modality": "motion_capture", |
| "name": "body_joints" |
| }, |
| { |
| "dim": 980, |
| "display": "Depth + Confidence", |
| "modality": "depth", |
| "name": "depth_confidence" |
| }, |
| { |
| "dim": 896, |
| "display": "Language Text", |
| "modality": "language", |
| "name": "caption_objects_interaction_text" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Fisheye Cam0", |
| "modality": "video", |
| "name": "video_fisheye_cam0" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Fisheye Cam1", |
| "modality": "video", |
| "name": "video_fisheye_cam1" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Fisheye Cam2", |
| "modality": "video", |
| "name": "video_fisheye_cam2" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Fisheye Cam3", |
| "modality": "video", |
| "name": "video_fisheye_cam3" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Stereo Left", |
| "modality": "video", |
| "name": "video_stereo_left" |
| }, |
| { |
| "dim": 686, |
| "display": "Video Stereo Right", |
| "modality": "video", |
| "name": "video_stereo_right" |
| }, |
| { |
| "dim": 441, |
| "display": "Left Hand", |
| "modality": "motion_capture", |
| "name": "hand_left_joints" |
| }, |
| { |
| "dim": 441, |
| "display": "Right Hand", |
| "modality": "motion_capture", |
| "name": "hand_right_joints" |
| }, |
| { |
| "dim": 168, |
| "display": "Audio", |
| "modality": "audio", |
| "name": "audio_fisheye_cam0_aac" |
| }, |
| { |
| "dim": 147, |
| "display": "Body Contacts", |
| "modality": "motion_capture", |
| "name": "body_contacts" |
| }, |
| { |
| "dim": 117, |
| "display": "Calibration", |
| "modality": "pose_slam", |
| "name": "calibration" |
| }, |
| { |
| "dim": 63, |
| "display": "Camera Rotation", |
| "modality": "pose_slam", |
| "name": "camera_rotation_matrix" |
| }, |
| { |
| "dim": 42, |
| "display": "IMU Accel/Gyro", |
| "modality": "inertial", |
| "name": "imu_accel_gyro" |
| }, |
| { |
| "dim": 22, |
| "display": "SLAM Point Cloud", |
| "modality": "pose_slam", |
| "name": "slam_point_cloud" |
| }, |
| { |
| "dim": 21, |
| "display": "Camera Translation", |
| "modality": "pose_slam", |
| "name": "camera_translation" |
| } |
| ], |
| "feature_dim_by_modality": { |
| "audio": 168, |
| "depth": 980, |
| "inertial": 42, |
| "language": 896, |
| "motion_capture": 2121, |
| "pose_slam": 223, |
| "video": 4116 |
| }, |
| "file_bytes_by_kind": { |
| "hdf5_annotation_container": { |
| "bytes": 1931496028, |
| "count": 1, |
| "human": "1.80 GiB" |
| }, |
| "rerun_viewer_recording": { |
| "bytes": 2702924036, |
| "count": 1, |
| "human": "2.52 GiB" |
| }, |
| "video": { |
| "bytes": 383606257, |
| "count": 5, |
| "human": "365.84 MiB" |
| }, |
| "video_with_audio": { |
| "bytes": 89842251, |
| "count": 1, |
| "human": "85.68 MiB" |
| } |
| }, |
| "file_count": 8, |
| "hdf5_groups": [ |
| { |
| "description": "Camera intrinsics/extrinsics and static calibration values used to align sensor streams.", |
| "group": "calibration" |
| }, |
| { |
| "description": "JSON text with task config, action segments, object mentions, interaction descriptions, and global summary.", |
| "group": "caption" |
| }, |
| { |
| "description": "Depth maps and confidence channels aligned to the episode timeline.", |
| "group": "depth" |
| }, |
| { |
| "description": "Full-body joint and contact signals for human motion modeling.", |
| "group": "full_body_mocap" |
| }, |
| { |
| "description": "Left and right hand joint trajectories used by hand-motion and forecast tasks.", |
| "group": "hand_mocap" |
| }, |
| { |
| "description": "Accelerometer and gyroscope streams sampled at higher frequency than the video frames.", |
| "group": "imu" |
| }, |
| { |
| "description": "Episode metadata, frame indexing, and source bookkeeping.", |
| "group": "metadata" |
| }, |
| { |
| "description": "Camera trajectory, pose, and sparse SLAM point-cloud information.", |
| "group": "slam" |
| }, |
| { |
| "description": "Video stream metadata and per-frame alignment information.", |
| "group": "video" |
| } |
| ], |
| "object_vocab_count": 34, |
| "segment_count": 35, |
| "source_policy": "Window-level labels, features, predictions, and diagnostics are embedded here. Official raw MP4/HDF5/RRD files are linked from the Raw Sample Browser, with compact browser-preview clips for immediate MP4/audio playback.", |
| "top_actions": [ |
| { |
| "count": 159, |
| "name": "Hold gooseneck kettle" |
| }, |
| { |
| "count": 159, |
| "name": "Pour coffee" |
| }, |
| { |
| "count": 127, |
| "name": "Position kettle to pour" |
| }, |
| { |
| "count": 95, |
| "name": "Lift gooseneck kettle" |
| }, |
| { |
| "count": 95, |
| "name": "Close bottle cap" |
| }, |
| { |
| "count": 95, |
| "name": "Wait/Prepare for pouring" |
| }, |
| { |
| "count": 87, |
| "name": "Transfer coffee to dripper" |
| }, |
| { |
| "count": 63, |
| "name": "Grasp coffee scoop" |
| }, |
| { |
| "count": 47, |
| "name": "Hold coffee carafe" |
| }, |
| { |
| "count": 39, |
| "name": "Move kettle" |
| }, |
| { |
| "count": 31, |
| "name": "Move kettle away" |
| }, |
| { |
| "count": 30, |
| "name": "Pick up kettle" |
| } |
| ], |
| "top_objects": [ |
| { |
| "count": 636, |
| "name": "coffee dripper" |
| }, |
| { |
| "count": 517, |
| "name": "scale" |
| }, |
| { |
| "count": 446, |
| "name": "gooseneck kettle" |
| }, |
| { |
| "count": 333, |
| "name": "digital scale" |
| }, |
| { |
| "count": 238, |
| "name": "water bottle" |
| }, |
| { |
| "count": 217, |
| "name": "white mug" |
| }, |
| { |
| "count": 203, |
| "name": "white bottle" |
| }, |
| { |
| "count": 170, |
| "name": "coffee mug" |
| }, |
| { |
| "count": 158, |
| "name": "bottle" |
| }, |
| { |
| "count": 145, |
| "name": "dripper" |
| }, |
| { |
| "count": 126, |
| "name": "glass carafe" |
| }, |
| { |
| "count": 123, |
| "name": "kettle" |
| } |
| ], |
| "top_subtasks": [ |
| { |
| "count": 166, |
| "name": "Handle gooseneck kettle" |
| }, |
| { |
| "count": 151, |
| "name": "Pour coffee" |
| }, |
| { |
| "count": 150, |
| "name": "Pick up and position kettle" |
| }, |
| { |
| "count": 119, |
| "name": "Prepare for pouring" |
| }, |
| { |
| "count": 111, |
| "name": "Prepare coffee equipment and scoop grounds" |
| }, |
| { |
| "count": 111, |
| "name": "Lift gooseneck kettle" |
| }, |
| { |
| "count": 111, |
| "name": "Pour and close white bottle" |
| }, |
| { |
| "count": 79, |
| "name": "Transfer coffee grounds to dripper" |
| }, |
| { |
| "count": 79, |
| "name": "Set down kettle and retrieve white bottle" |
| }, |
| { |
| "count": 39, |
| "name": "Move kettle" |
| }, |
| { |
| "count": 14, |
| "name": "" |
| }, |
| { |
| "count": 10, |
| "name": "Pour milk into coffee" |
| } |
| ], |
| "total_bytes": 5107868572, |
| "total_human": "4.76 GiB", |
| "windowization": { |
| "audio_sample_rate": 16000, |
| "duration_human": "4.85 min", |
| "duration_sec": 291.03000343524565, |
| "feature_dim": 8546, |
| "fps_observed": 20.00137419266181, |
| "note": "The public task suite converts the raw episode into overlapping 20-frame windows with a 5-frame stride.", |
| "num_frames": 5821, |
| "num_windows": 1161, |
| "stride_frames": 5, |
| "window_frames": 20 |
| } |
| }, |
| "selected_128": { |
| "bytes_by_size_band": { |
| "long": { |
| "bytes": 75928334176, |
| "human": "70.71 GiB" |
| }, |
| "lower_mid": { |
| "bytes": 75730496234, |
| "human": "70.53 GiB" |
| }, |
| "short": { |
| "bytes": 70693191518, |
| "human": "65.84 GiB" |
| }, |
| "upper_mid": { |
| "bytes": 75837737335, |
| "human": "70.63 GiB" |
| } |
| }, |
| "bytes_by_split": { |
| "test": { |
| "bytes": 37243648205, |
| "human": "34.69 GiB" |
| }, |
| "train": { |
| "bytes": 223733297061, |
| "human": "208.37 GiB" |
| }, |
| "val": { |
| "bytes": 37212813997, |
| "human": "34.66 GiB" |
| } |
| }, |
| "exports": [ |
| { |
| "episodes": 119, |
| "key": "sparse_export", |
| "label": "Sparse selected-128 export", |
| "samples": 3808, |
| "scale_counts": {}, |
| "split_counts": { |
| "test": 448, |
| "train": 2848, |
| "val": 512 |
| } |
| }, |
| { |
| "episodes": 119, |
| "key": "qwen_v6_multiscale_export", |
| "label": "Qwen3-Omni v6 multiscale JSONL", |
| "samples": 34269, |
| "scale_counts": {}, |
| "split_counts": { |
| "test": 4032, |
| "train": 25629, |
| "val": 4608 |
| } |
| }, |
| { |
| "episodes": 119, |
| "key": "dense_multiscale_compact_export", |
| "label": "Dense multiscale compact export", |
| "samples": 106095, |
| "scale_counts": { |
| "dense_20f_stride10": 60725, |
| "long_80f_stride40": 15067, |
| "medium_40f_stride20": 30303 |
| }, |
| "split_counts": { |
| "test": 11795, |
| "train": 79593, |
| "val": 14707 |
| } |
| } |
| ], |
| "metadata_matrix_sparse": { |
| "bytes": 217477, |
| "feature_dim": 394, |
| "human": "212.38 KiB", |
| "row_count": 3808, |
| "sha256": "cdaa55ea3bb229d70bee845f05628ed7a984dd46f95080358d51943b4b8975df", |
| "split_counts": { |
| "test": 448, |
| "train": 2848, |
| "val": 512 |
| } |
| }, |
| "metadata_matrix_v2": { |
| "bytes": 3375193, |
| "feature_dim": 394, |
| "human": "3.22 MiB", |
| "row_count": 34269, |
| "sha256": "3c2a23b681c9f257a4c04710c72441650844ede6279d3e67be57ccc3abd4fa90", |
| "split_counts": { |
| "test": 4032, |
| "train": 25629, |
| "val": 4608 |
| } |
| }, |
| "official_dataset": { |
| "access_note": "Episode directory pages are browsable on Hugging Face. Raw annotation/video file downloads require access to the gated official dataset and are not redistributed here.", |
| "gated_resolve_base_url": "https://huggingface.co/datasets/ropedia-ai/xperience-10m/resolve/main", |
| "repo_id": "ropedia-ai/xperience-10m", |
| "repo_sha": "ce943cf271a758b60240084892d05cf6dc12dd90", |
| "tree_base_url": "https://huggingface.co/datasets/ropedia-ai/xperience-10m/tree/main" |
| }, |
| "raw20_proxy_tasks": [ |
| "camera_view_sync_retrieval", |
| "interaction_text_prediction" |
| ], |
| "raw20_result_records": 40, |
| "selection_summary": { |
| "one_episode_per_top_level_session": true, |
| "selected_download_size_excluding_visualization_rrd_bytes": 298189759263, |
| "selected_episode_count": 128, |
| "selected_split_counts": { |
| "test": 16, |
| "train": 96, |
| "val": 16 |
| }, |
| "size_band_counts": { |
| "long": 32, |
| "lower_mid": 32, |
| "short": 32, |
| "upper_mid": 32 |
| } |
| }, |
| "size_band_counts": { |
| "long": 32, |
| "lower_mid": 32, |
| "short": 32, |
| "upper_mid": 32 |
| }, |
| "sparse_main_task_counts": [ |
| { |
| "count": 96, |
| "name": "Stocking grocery shelves" |
| }, |
| { |
| "count": 96, |
| "name": "Stocking grocery store shelves" |
| }, |
| { |
| "count": 96, |
| "name": "Measuring and marking cardboard" |
| }, |
| { |
| "count": 64, |
| "name": "Stocking supermarket shelves" |
| }, |
| { |
| "count": 64, |
| "name": "Stocking retail store shelves" |
| }, |
| { |
| "count": 64, |
| "name": "Assembling a jigsaw puzzle" |
| }, |
| { |
| "count": 64, |
| "name": "Crafting cardboard shapes" |
| }, |
| { |
| "count": 64, |
| "name": "Stocking shelves in a retail store" |
| }, |
| { |
| "count": 64, |
| "name": "Playing Mahjong" |
| }, |
| { |
| "count": 64, |
| "name": "Stocking canned goods on a supermarket shelf" |
| }, |
| { |
| "count": 32, |
| "name": "Sorting and organizing colorful buttons on a table" |
| }, |
| { |
| "count": 32, |
| "name": "Crafting cardboard shapes and preparing workspace" |
| } |
| ], |
| "sparse_windows_by_split": { |
| "test": 448, |
| "train": 2848, |
| "val": 512 |
| }, |
| "split_episode_counts": { |
| "test": 16, |
| "train": 96, |
| "val": 16 |
| } |
| }, |
| "sources": [ |
| { |
| "path": "docs/data/raw_sample_files.json", |
| "scope": "public_sample" |
| }, |
| { |
| "path": "docs/data/single_episode_explorer.json", |
| "scope": "public_sample" |
| }, |
| { |
| "path": "results/episode_task_suite/windows.csv", |
| "scope": "public_sample" |
| }, |
| { |
| "path": "docs/data/xperience10m_128_episode_feature_index.json", |
| "scope": "selected_128" |
| }, |
| { |
| "path": "results/omni_finetune/xperience10m_128_episode_selection.csv", |
| "scope": "selected_128" |
| }, |
| { |
| "path": "results/omni_finetune/multi_episode_128_task_baselines/windows.csv", |
| "scope": "selected_128" |
| }, |
| { |
| "path": "results/omni_finetune/full_dataset_metadata_audit.json", |
| "scope": "full_hf_dataset" |
| } |
| ], |
| "status": "pass" |
| } |
|
|
