datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
meta13sphere_IRS_DCE_Topological_Dynamics__Boundary_Dissolution_Physics
Resonance Resonance / IRS-DCE
MASTER README (FULL EXTENDED VERSION)
If you need the other data or pdf check on [https://huggingface.co/datasets/meta13sphere/phaseShift_shell_result_pdf]
[2026-09-25 Update]
The Geometry of IRS-DCE Boundary Dissolution v1.0 is now available in Korean and English.
Resolution-Dependent Component Decomposition and Dynamic Rearrangement
This release connects the existing IRS-DCE and BBRCM research to mathematical analysis and recorded… See the full description on the dataset page: https://huggingface.co/datasets/meta13sphere/meta13sphere_IRS_DCE_Topological_Dynamics__Boundary_Dissolution_Physics.hard-intersection-multimodal-sample
Hard Intersection Multimodal Samples
Release Notes
Release
Description
v1.0.0
Initial public release.
v1.1.0
Added Unreal Engine assets.Fixed issues in the OpenDRIVE map data.Updated the README to improve documentation and usability.
v1.2.0
Added OpenCRG road surface elevation data.Updated 3DGS reconstruction assets.Updated the README to document OpenCRG support.
Dataset Summary
Hard Intersection Multimodal Samples is a curated… See the full description on the dataset page: https://huggingface.co/datasets/dynamic-maps/hard-intersection-multimodal-sample.dynamic_robot_bench_dr_scripted_14k
dynamic_robot_bench_dr_scripted_14k
14,400 scripted-expert demonstrations across the 72 evaluated task
families of dynamic-robot-bench — a
conveyor-belt dynamic-manipulation benchmark (Franka Panda + wrist camera, ManiSkill 3 / SAPIEN
GPU sim). One LeRobot v2.1 dataset: 200 episodes per family,
success-filtered, every domain-randomization knob on, and belt speed uniform over
0.10–0.40 m/s.
1,118,617 frames · 209 distinct language instructions · 20 fps
The belt speed… See the full description on the dataset page: https://huggingface.co/datasets/Damin3927/dynamic_robot_bench_dr_scripted_14k.VanGogh_vs_TreeOilPainting_Torque_Brushstroke_Dynamics_EnergyField_Phase2_2026🚪 Quick Entry: Start Here
What is this dataset (in 2 sentences)
This dataset is not about what a painting looks like.
It is about what physically created it.
Instead of pattern recognition, this system forces AI to perform causal reasoning based on force, motion, and energy encoded in brushstrokes.
What you can do here
With this dataset, you can:
Reconstruct brushstroke motion from a static image
Infer pressure, torque, and stroke velocity
Test whether an AI… See the full description on the dataset page: https://huggingface.co/datasets/HaruthaiAi/VanGogh_vs_TreeOilPainting_Torque_Brushstroke_Dynamics_EnergyField_Phase2_2026.ContinuousMountainCar_Dynamic_Best_Tracking#4.5million timesteps
Included Dynamic Best Weight Tracking: Which Saves the best 15k deployment weights if performance improves to a seperate file
this allow the trained agent run infererence from the "improved performance file only" but the trained_agent viewpoint is
narrow. while the agent completes the goal, it is not super efficient due to edge usecases. #Good- #VeryGood
-Lokis Laugh is rising, like smoke above the flame, a trick without a master, a god without a name.
-Lokis Laugh is… See the full description on the dataset page: https://huggingface.co/datasets/privateboss/ContinuousMountainCar_Dynamic_Best_Tracking.DynamicVLA-Task-Examples
DynamicVLA Simulation Task Examples
打开 MP4 任务浏览页
DOM has 3 task families, not 3 task_index values: 140,932 structured instruction IDs and 207,306 episodes. This subset selects 6 demonstrations per family (18 total), preserving actual task_index/episode_index and structured instructions. All 3 published cameras are single-arm cameras: opposite, wrist, side, not top/left/right arms.
所有示例是完整 H.264 MP4。单臂相机保留官方名称：opst_cam / wrist_cam / side_cam。每条有同步合并视频、各路原视频与官方动作/状态 Parquet。… See the full description on the dataset page: https://huggingface.co/datasets/Travor278/DynamicVLA-Task-Examples.DynamicvlmDisney-Theme-Park-Queue-Dynamics
🎢 Disney World Queue Dynamics
A Comprehensive EDA & Strategic Analysis
Author: Matan Zigelman • University: Reichman University • Date: March 2026
📋 Project Introduction: Disney Theme Park Queue Dynamics
This project analyzes a numeric-heavy operational dataset from a major Disney theme park, sourced from kaggle and containing approximately 3,757,301 records. The dataset is primarily driven by time-based and operational metrics ( e.g.… See the full description on the dataset page: https://huggingface.co/datasets/matanzig/Disney-Theme-Park-Queue-Dynamics.so101_tb4_dynamichandover_scene1_receiver_blue_brush_080726Chinese_Commercial_Kitchen_Manipulation_Dataset_Preview
🍳 Chinese Commercial Kitchen Manipulation Dataset — Sample Pack v0.1
Asia's first real commercial kitchen manipulation dataset.Professional chef (20 years) · Real restaurant environment · Multi-view RGB-D · Egocentric video
📧 Request evaluation samples or full data: andy@dynamicnova.com
Overview
This sample pack contains real-world cooking demonstrations collected in an operating Chinese commercial kitchen in Zhongshan, Guangdong, China.
The data focuses on… See the full description on the dataset page: https://huggingface.co/datasets/nova-dynamics/Chinese_Commercial_Kitchen_Manipulation_Dataset_Preview.DynamicReplica_wai
DynamicReplica Dataset in WAI format
Preprocessed DynamicReplica following MapAnything.
Each scene contains the following structure when extracted:
0a5b4c-3_obj_source/
├── depth
| ├── 0a5b4c-3_obj_source_left-0000.exr
| ├── ...
├── images
| ├── 0a5b4c-3_obj_source_left-0000.png
| ├── ...
├── _process_log_backup.json
├── _process_log.json
└── scene_meta.json
Data Format Details:
depth: depth in .exr format.
images: images in .png format.
scene_meta.json: meta… See the full description on the dataset page: https://huggingface.co/datasets/ZhengGuangze/DynamicReplica_wai.ai-writing-evolutionary-dynamics
Evolutionary Dynamics of AI-Mediated Scientific Writing
Complete experimental logs and reproduction package.
Author: Arif Mohamed Khan Rabi AhamadAffiliation: School of Information Studies, Syracuse UniversityContact: arabiaha@syr.edu | ORCID: 0009-0001-0986-7570
Contents
Directory
Files
Description
logs/
0
Complete stdout from all experiments
figures/
91
All paper figures (PDF + PNG)
data/
17
Derived datasets (QTE matrices, Price components)… See the full description on the dataset page: https://huggingface.co/datasets/arifmohamedkhan/ai-writing-evolutionary-dynamics.dynamicworld-300
Dynamic World 300
A small land-cover segmentation dataset for learning and quick experiments:
300 expert-labelled Dynamic World
tiles, each paired with the Sentinel-2 scene it was labelled on. It is the
data of the GeoSave Engine
example notebooks.
classes.yaml class names and colours
train/ images/<id>.tif labels/<id>.tif 200 tiles
val/ images/<id>.tif labels/<id>.tif 50 tiles
test/ images/<id>.tif labels/<id>.tif 50 tiles
An… See the full description on the dataset page: https://huggingface.co/datasets/fatmur/dynamicworld-300.agent-failure-dynamics
AgentHazard: Process-Centric Benchmark for AI Agent Trajectory Analysis
AgentHazard is the first benchmark designed specifically for process-level analysis of AI coding agent trajectories. Unlike existing benchmarks that evaluate only final outcomes (pass/fail), AgentHazard provides standardized edit-level annotations, hazard estimation protocols, and stopping-policy evaluation tasks with unified evaluation across 85,050 trajectories from 6+ agent families.
Why… See the full description on the dataset page: https://huggingface.co/datasets/Anonymousblind/agent-failure-dynamics.DynamicObjects
Dynamic Objects Dataset
This dataset is proposed by NVFi, and used by FreeGave and TRACE.
Structure
The structure of the dataset is as:
DynObjects
| - data
| | - fallingball
| | | - train: serves as training data
| | | - val: used for evaluating novel view interpolation
| | | - test: used for evaluating future extrapolation
| | | - transforms_train.json: camera poses and other meta informations for training set
| | | - transforms_val.json: camera poses and other meta… See the full description on the dataset page: https://huggingface.co/datasets/scintigimcki/DynamicObjects.dynamic_robot_bench_dr4050_scripted_5k
dynamic_robot_bench_dr4050_scripted_5k
The fast-band (0.40–0.50 m/s) companion to
dynamic_robot_bench_dr_scripted_10k.
Scripted-expert demonstrations on a moving conveyor belt, collected with every
domain-randomization knob on, across 100 task families.
episodes
5000 (50 per family)
frames
327,466
fps
20
distinct instructions
387
belt speed
uniform over 0.40–0.50 m/s
cameras
exterior_image_1_left, exterior_image_2_left, wrist_image_left (224x224)
action… See the full description on the dataset page: https://huggingface.co/datasets/Damin3927/dynamic_robot_bench_dr4050_scripted_5k.Dynamic-RE10K
D-RE10K: Dynamic Real-Estate 10K Dataset
Overview
This dataset contains the DRE10K training split (15,467 clips, 147,422 frames) and the DRE10K mask test split (76 clips, 1,541 frames), released on Hugging Face for research on self-supervised large view synthesis in dynamic environments. The data is collected from real-estate walkthrough videos and curated specifically for training and evaluating novel view synthesis models in scenes with dynamic objects.
Our dataset… See the full description on the dataset page: https://huggingface.co/datasets/uva-cv-lab/Dynamic-RE10K.DynamicIndoorScenes
Dynamic Indoor Scenes Dataset
This dataset is proposed by NVFi, and used by FreeGave and TRACE.
Structure
The structure of the dataset is as:
DynObjects
| - data
| | - darkroom: data for Gnome House scene
| | | - train: serves as training data
| | | - val: used for evaluating novel view interpolation
| | | - test: used for evaluating future extrapolation
| | | - transforms_train.json: camera poses and other meta informations for training set
| | | - transforms_val.json:… See the full description on the dataset page: https://huggingface.co/datasets/scintigimcki/DynamicIndoorScenes.DynamicReplica
Dynamic Replica dataset for point tracking evaluation
Each scene contains the following structure when extracted:
valid/
├── 0cde48-3_obj_source_left/
| ├── images/
| | ├── 0cde48-3_obj_source_left-0000.png
| | ├── ...
| | └── done.ok
| ├── trajectories/
| | ├── 000000.pth
| | ├── ...
| | └── 000299.pth
├── ...
└── frame_annotations_valid.jgz
Please note:
20 sequences in total.
DynamicVersedynamicearthnet-taco
DynamicEarthNet
This is a repackaging, not a new dataset. It is DynamicEarthNet by Planet Labs Inc., TUM (Toker et al.), converted to TACO. Pixel values and labels are kept as released except where the description below says otherwise. All credit belongs to the original authors: if you use it, please cite them and follow their licence.
original dataset · paper · licence: CC-BY-SA-4.0
Repackaged into TACO by the Image and Signal Processing Group (ISP), Universitat de València… See the full description on the dataset page: https://huggingface.co/datasets/isp-uv-es/dynamicearthnet-taco.dynamicearthnet-planetkortowo-dynamics-image-dataset
Kortowo Dynamics Image Dataset
This is the image dataset for the Kortowo Dynamics project.
It contains frames derived from the original video dataset.
Purpose
The initial purpose of this dataset has been to train a video classification model using the strategy of analysing individual frames.
Contents
The dataset contains video frames of the Boston Dynamics' Spot robot performing different actions.
Available classes:
body_swing
crawling
jumping
looking_left… See the full description on the dataset page: https://huggingface.co/datasets/jlynxdev/kortowo-dynamics-image-dataset.gova-logit-dynamics-qwen2.5vl-7b
GOVA Logit Dynamics — Qwen2.5-VL-7B-Instruct
Per GOVA word-acquisition works item (27,155): the fill distribution text-only
(blind) vs with the image for Qwen2.5-VL-7B-Instruct. Correctness-agnostic — it captures how
vision reshapes the prediction (passive language prior + active vision).
Organized into 5 splits by the vision-role taxonomy (3 cases; Case 1 & 3 sub-split
strong/weak; threshold high>0.5, boost>0.05, re-derivable from the stored probs):
split
meaning… See the full description on the dataset page: https://huggingface.co/datasets/guangliangliu/gova-logit-dynamics-qwen2.5vl-7b.gova-logit-dynamics-octobert
GOVA Logit Dynamics — OctoBERT (World-to-Words)
Per GOVA word-acquisition works item (27,155): the fill distribution text-only
(blind) vs with the image for OctoBERT (World-to-Words). Correctness-agnostic — it captures how
vision reshapes the prediction (passive language prior + active vision).
Organized into 5 splits by the vision-role taxonomy (3 cases; Case 1 & 3 sub-split
strong/weak; threshold high>0.5, boost>0.05, re-derivable from the stored probs):
split
meaning… See the full description on the dataset page: https://huggingface.co/datasets/guangliangliu/gova-logit-dynamics-octobert.swim-360-dynamic-cv-datasetrobotwin_dynamicmsd_dsprites_dynamic
MSD dSprites-Dynamic Dataset Attribution
The Multi-factor Sequential Disentanglement benchmark includes a modified variant of the dSprites dataset, adapted to support sequential multi-factor disentanglement.
In this modified sequential version, the object’s color, shape, scale, and orientation are fixed, while its spatial position may change over time.
Original repository:
https://github.com/deepmind/dsprites-dataset
@misc{dsprites17,
author = {Loic Matthey and Irina Higgins and… See the full description on the dataset page: https://huggingface.co/datasets/TalBarami/msd_dsprites_dynamic.conveyor_apple_dynamic_scripted_posDynamicManip