datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
lab-bench
LAB-Bench
The Language Agent Biology Benchmark, or LAB-Bench, is an evaluation dataset for AI systems intended to benchmark capabilities foundational to scientific research in biology. The dataset currently consists of 8 broad categories, comprising 30 narrower subtasks, including extracting information from the scientific literature (LitQA2), retrieving information from databases (DbQA) and supplementary information (SuppQA), reasoning about scientific figures (FigQA) and tables… See the full description on the dataset page: https://huggingface.co/datasets/futurehouse/lab-bench.3D-FUTURE-FCOhle_futurehouse_bronzeQ-Bench-HFQ-Bench2-HFhle_futurehouse_bronzehle-gold-bio-chem
Humanity's Last Exam (HLE) Bio/Chem Gold
Humanity’s Last Exam (HLE) is a challenging question-answering AI benchmark covering advanced academic fields including Math, Physics, Chemistry, Biology, Engineering, and Computer Science.
At FutureHouse, we audited the biology and chemistry subsets of HLE using a combination of expert human evaluators and our in-house research agent, and found that around 30% of the questions contain answers directly contradicted by peer-reviewed… See the full description on the dataset page: https://huggingface.co/datasets/futurehouse/hle-gold-bio-chem.hle_futurehouse_goldpusht-privilegedThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "2d pointer",
"total_episodes": 1,
"total_frames": 91,
"total_tasks": 1,
"total_videos": 0,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 10,
"splits": {
"train": "0:1"
},
"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/the-future-dev/pusht-privileged.pusht-image-v1
PushT Dataset
This dataset contains demonstrations for the PushT environment, a robotic manipulation task where an agent needs to push a T-shaped object onto a matching target surface.
Each episode is initialized with a randomized T-block and target position and orientation to ensure a more realistic scenario.
Environment Details
The dataset was collected using the gym-pusht environment, which provides a simple 2D robotic pushing task.
Task: Push the T-shaped… See the full description on the dataset page: https://huggingface.co/datasets/the-future-dev/pusht-image-v1.q-align-datasetsq-bench2
This is the benchmark dataset for "A Benchmark for Multi-modal Foundation Models on Low-level Vision: from Single Images to Pairs"
The structure of the jsonl files is as follows:
q-bench2-a1-dev.jsonl (with img_path, question, answer_candidates, correct_answer)
q-bench2-a1-test.jsonl (with img_path, question, answer_candidates, without correct_answer)
q-bench2-a2.jsonl (with img_path, empty response)
The img_path is organized as prefix + img1 + _cat_ + img2… See the full description on the dataset page: https://huggingface.co/datasets/q-future/q-bench2.omy_insert_manual_qactual_state7_future1
OMY Insert Manual Dual View q_actual State7 Future1
This is the converted dual-view LeRobot dataset for the MuJoCo plug insertion task.
Semantics
fps: 20 Hz
observation.image: agent-view RGB image, 256 x 256
observation.wrist_image: wrist/egocentric RGB image, 256 x 256
observation.state[t]: measured joint state q_actual[t], 7D [joint1..joint6, gripper_actual]
action[t]: measured next-frame joint state q_actual[t+1], 7D [joint1..joint6, gripper_actual]
With… See the full description on the dataset page: https://huggingface.co/datasets/wkal/omy_insert_manual_qactual_state7_future1.behavioral_stability
Behavioral Stability (Unsteered Generations)
Unsteered baseline generations and their measured behavior, used as the reference point for steering.
This dataset is part of the data release for the paper Predicting Future Behaviors in Reasoning Models Enables Better Steering.
The data is organized as <model>/<dataset>/.... Each row below links to the browsable folder for that model and dataset, where the individual files can be viewed and downloaded.
Data… See the full description on the dataset page: https://huggingface.co/datasets/future-probes/behavioral_stability.UI-Redline-bench
UI-Redline-bench
This dataset contains the benchmark data for the paper "UI-Redline-bench: 赤入れ指示によるWebUIコード修正ベンチマーク".
The benchmark evaluates the capability of Vision-Language Models (VLMs) to modify Web UI code (HTML/CSS) based on visual "redline" instructions (handwritten or digital) drawn on screenshots.
📄 Paper | 💻 GitHub Repository (Evaluation Code & Runnable Environment)
Usage
This Hugging Face dataset contains only the images and the corresponding metadata.… See the full description on the dataset page: https://huggingface.co/datasets/future-architect/UI-Redline-bench.winml-test-set
WinML Test Set
Dataset Summary
WinML Test Set is an evaluation‑only collection for validating model accuracy and stability on Windows ML / DirectML / ONNX Runtime pipelines. It aggregates several permissively‑licensed sources and harmonizes schema for reproducible, regression‑grade testing across backends and versions. Not intended for training.
Intended Use
Accuracy and regression benchmarking of Windows ML / DirectML / ONNX Runtime pipelines.… See the full description on the dataset page: https://huggingface.co/datasets/Futuremark/winml-test-set.creategenClothFold206This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 206,
"total_frames": 20600,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:206"
},
"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/the-future-dev/ClothFold206.A-BenchProject Page
Glance at A-Bench Performance
For open-source models, LLaVA-NeXT (Qwen-110B) takes the first place. For closed-source models, GEMINI 1.5 PRO takes the first place.
Evaluate your model on A-Bench
First download the dataset and meta information from Huggingface.
The imgs.zip contains all the AI-generated images and Abench.json contains all the meta information including the img_path, questions, answers, and categories. The item of Abench.json is… See the full description on the dataset page: https://huggingface.co/datasets/q-future/A-Bench.hle_futurehouse_goldClothFold75This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 75,
"total_frames": 7500,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:75"
},
"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/the-future-dev/ClothFold75.realhouseClothFold150This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 150,
"total_frames": 15000,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:150"
},
"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/the-future-dev/ClothFold150.DryClothThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 8000,
"total_frames": 48000,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:8000"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path": null… See the full description on the dataset page: https://huggingface.co/datasets/the-future-dev/DryCloth.ClothFold125This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 125,
"total_frames": 12500,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:125"
},
"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/the-future-dev/ClothFold125.pusht_image_keypointsThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "unknown",
"total_episodes": 206,
"total_frames": 25650,
"total_tasks": 1,
"chunks_size": 1000,
"fps": 10,
"splits": {
"train": "0:206"
},
"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/the-future-dev/pusht_image_keypoints.RopeFlatten150This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 150,
"total_frames": 11250,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:150"
},
"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/the-future-dev/RopeFlatten150.hle_futurehouse_bronzeClothFold200epThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 200,
"total_frames": 20000,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:200"
},
"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/the-future-dev/ClothFold200ep.ClothFold100This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": null,
"total_episodes": 100,
"total_frames": 10000,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 10,
"splits": {
"train": "0:100"
},
"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/the-future-dev/ClothFold100.