datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ScaleEdit-12M
ScaleEdit-12M: Scaling Open-Source Image Editing Data Generation via Multi-Agent Framework
📌 Overview
The largest open-source instruction-based image editing dataset to date.
ScaleEdit-12M contains 11.8 million rigorously verified instruction–image pairs spanning 23 task families across diverse real and synthetic visual domains. It was constructed using ScaleEditor, a fully open-source hierarchical multi-agent framework that eliminates… See the full description on the dataset page: https://huggingface.co/datasets/InternVL-U/ScaleEdit-12M.tiny-scalesThis repo contains the tinyHLE dataset, a list of items to use as a subset of the Humanity's Last Exam benchmark in order to make evaluation more efficient.
The repo contains two files:
tiny_hle.json: a file containing a list of question IDs and weights for three different sample sizes (0.5%, 1.0%, 2.0%)
clean_scales_embedding_hle.parquet: a file containing embeddings representing each item of the HLE benchmark along 16 cognitive scales dimensions, used to create the subsets
Since these are… See the full description on the dataset page: https://huggingface.co/datasets/ambean-tr/tiny-scales.paired-llama-3.2-1b-embeddings-lmsys-chat-1m
Paired Llama 3.2 1B Token Embeddings (LMSYS-Chat-1M)
This dataset contains paired activations corresponding to single token locations extracted from Meta's Llama 3.2 1B Instruct on conversations from LMSYS-Chat-1M.
Embeddings are provided for layers 5 through 14, which capture the most interesting intermediate representations.
This dataset was built to study things like:
Learning different basis for activations at a given layer
Studying if there are cases where position encodes… See the full description on the dataset page: https://huggingface.co/datasets/scaleinvariant/paired-llama-3.2-1b-embeddings-lmsys-chat-1m.colinear_scaling_models
Collinear/Non-Collinear Scaling Models
Checkpoint repository for scaling law experiments comparing collinear (CO) and non-collinear (NC) experimental designs for the paper Tokens-per-Parameter Coverage Is Critical for Robust LLM Scaling Law Extrapolation under review for NeurIPS 2026.
Code
Anonymized code repository (reproduces all tables): anonymous.4open.science
Directory Structure
{dataset}/{design}/N_{param_count}/
Dataset: wikipedia, pes2o, cosmopedia… See the full description on the dataset page: https://huggingface.co/datasets/TPPIsCriticalFor/colinear_scaling_models.scannet_mcmc_3dgs
Data Statistics
Scenes
Mean PSNR ↑
Mean SSIM ↑
Mean LPIPS ↓
Mean Depth L1 ↓
Mean #3DGS
Total #3DGS
1,613
30.17 dB
0.875
0.221
0.0151 m
1.000M
1.613B
License Notice:This dataset is derived from ScanNet and follows the ScanNet Terms of Use.See ScanNet Terms for details.
Scandium-Dataset
Dataset Card — Scandium-Dataset v1.0.0
Summary
Scandium-Dataset provides a harmonized, quality-scored foundation of DFT-computed structural and thermodynamic properties across 267,230 materials from Materials Project, OQMD, and JARVIS-DFT. It supports the early screening stage of battery materials discovery — filtering by phase stability, electronic structure, and structural family — before downstream property prediction (ionic conductivity, mechanical stability… See the full description on the dataset page: https://huggingface.co/datasets/Scandium-Labs/Scandium-Dataset.scannetpp_v2_mcmc_3dgs
Data Statistics
Scenes
Mean PSNR ↑
Mean SSIM ↑
Mean LPIPS ↓
Mean Depth L1 ↓
Mean #3DGS
Total #3DGS
956
35.08 dB
0.959
0.062
0.0031 m
1.500M
1.434B
License Notice:This dataset is derived from ScanNet++ and follows the ScanNet++ Terms of Use.See ScanNet++ Terms for details.
scaling-laws-cacheThis dataset is my cache for the scaling-laws related to the gemstone models.
In data_cache is the approach 3 data cache with the mins for delta=1e-4, the mins for delta=1e-3 are in mins_1e-3.
This is the code I used to upload it:
import pandas as pd
from datasets import Dataset
import os
import gc
def get_data_dict(path):
contents = os.listdir(path)
ds_store = {}
for i, file in enumerate(contents):
gc.collect()
df = pd.read_parquet(f"{path}{file}")
for… See the full description on the dataset page: https://huggingface.co/datasets/smcleish/scaling-laws-cache.KernelBench
KernelBench
A benchmark designed to evaluate the ability of LLMs to generate efficient GPU kernels for optimizing neural network performance
Version
[07-21-2025] This HF dataset version has been updated to v0.1
Citation
@misc{ouyang2024kernelbench,
title={KernelBench: Can LLMs Write GPU Kernels?},
author={Anne Ouyang and Simon Guo and Azalia Mirhoseini},
year={2024},
url={https://scalingintelligence.stanford.edu/blogs/kernelbench/},
}
sae-activations-llama-3.1-8b-layer19-lmsys-chat-1m
SAE Feature Activations — Llama 3.1 8B Instruct, Layer 19 (LMSYS-Chat-1M)
This dataset contains Sparse Autoencoder (SAE) feature activations extracted from layer 19 of Meta's Llama 3.1 8B Instruct on conversations from LMSYS-Chat-1M.
It also has natural language explainations of features generated by GPT OSS 120B. See subset 4 for details.
The SAE used is Goodfire/Llama-3.1-8B-Instruct-SAE-l19, which decomposes layer-19 residual stream activations into interpretable sparse features.… See the full description on the dataset page: https://huggingface.co/datasets/scaleinvariant/sae-activations-llama-3.1-8b-layer19-lmsys-chat-1m.scannetpp_v1_mcmc_3dgs
Data Statistics
Scenes
Mean PSNR ↑
Mean SSIM ↑
Mean LPIPS ↓
Mean Depth L1 ↓
Mean #3DGS
Total #3DGS
330
28.89 dB
0.917
0.144
0.0201 m
1.552M
512.282M
License Notice:This dataset is derived from ScanNet++ and follows the ScanNet++ Terms of Use.See ScanNet++ Terms for details.
ps4mas-ps-scale
PS4MAS PS scale
Current PS / scale experiment catalog
This generated section is authoritative. Older tables above are historical.
Each experiments/<name>/ contains unified episodes.parquet, meta.json, and an unmodified summary.json when available. Raw full-hop traces.jsonl and run_config.json are separate files; scores are never injected into raw traces.
Partial snapshots have immutable content-derived names. Existing experiments are skipped unless… See the full description on the dataset page: https://huggingface.co/datasets/yinita/ps4mas-ps-scale.LDS-retrain-bank-adamw-N16k-bs256-scale0.25
Retrain bank: plan_adam_eps1e17_16k_scale0.25
This repository contains 100 fully retrained language models, not just scores.
Each model is GPT-2 (gpt2) fine-tuned on the same 16,000-document corpus with a different random 1% (160 documents) held out, from the same seed and the same data order as the base model in retrained/base. Retraining is deterministic within one environment, so the models differ only by the documents removed.
That is the expensive part of any leave-k-out… See the full description on the dataset page: https://huggingface.co/datasets/EleutherAI/LDS-retrain-bank-adamw-N16k-bs256-scale0.25.tpch_tables_scale_1
polars-tpch
This repo contains the code used for performance evaluation of polars. The benchmarks are TPC-standardised queries and data designed to test the performance of "real" workflows.
From the TPC website:
TPC-H is a decision support benchmark. It consists of a suite of business-oriented ad hoc queries and concurrent data modifications. The queries and the data populating the database have been chosen to have broad industry-wide relevance. This benchmark illustrates… See the full description on the dataset page: https://huggingface.co/datasets/kunishou/tpch_tables_scale_1.VietPET-RoI
VietPET-RoI
VietPET-RoI is a Vietnamese whole-body PET/CT dataset containing paired
cropped 3D volumes, regional reports, and modality-specific 3D ROI bounding
boxes. It is intended for medical multimodal research, report generation,
visual question answering, and ROI grounding.
Research use only. This dataset is not intended for diagnosis, treatment
decisions, or direct patient care.
Summary
Split
Patients
CT/PET region pairs
ROIs
Train
160
480
1,544… See the full description on the dataset page: https://huggingface.co/datasets/scarlettlin/VietPET-RoI.AIRBOT_MMK2_chop_the_scallions
AIRBOT_MMK2_chop_the_scallions
📋 Overview
This dataset uses an extended format based on LeRobot and is fully compatible with LeRobot.
Robot Type: discover_robotics_aitbot_mmk2
| 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
cut
📊 Dataset Statistics… See the full description on the dataset page: https://huggingface.co/datasets/RoboCOIN/AIRBOT_MMK2_chop_the_scallions.strawberry_picking_dataset_scara_extra
Strawberry Picking Dataset — SCARA Extra
Relationship to the paper: These are extra SCARA demonstrations collected
for other tasks and experimental setups. None of the data in this repository
were used in the training, evaluation, or other experiments reported in
Learning to Pick: A Visuomotor Policy for Clustered Strawberry Picking.
The paper is cited as related context for the robot platform and research area.
LeRobot v3.0 dataset converted from ACT/ALOHA HDF5 demos on… See the full description on the dataset page: https://huggingface.co/datasets/zfff/strawberry_picking_dataset_scara_extra.ps4mas-castle-scale
PS4MAS CASTLE scale
Current CASTLE / scale experiment catalog
This generated section is authoritative. Older tables above are historical.
Each experiments/<name>/ contains unified episodes.parquet, meta.json, and an unmodified summary.json when available. Raw full-hop traces.jsonl and run_config.json are separate files; scores are never injected into raw traces.
Partial snapshots have immutable content-derived names. Existing experiments are skipped unless… See the full description on the dataset page: https://huggingface.co/datasets/yinita/ps4mas-castle-scale.ctr-scan-object-uniform50-20260917
IdleMask review — passed (2026-09-19)
Reviewed by the dataset owner: observation.arm_active_mask is correct and this revision is a formally usable CTR dataset. This section supersedes previous active/idle-mask descriptions below.
For each of the 50 episodes and each physical arm, only the initial contiguous scheduling delay may have mask 0. From first duty through the final frame the mask is always 1. Scan synchronization waits, cooperative holds, the shared scan tail and… See the full description on the dataset page: https://huggingface.co/datasets/Shiki42/ctr-scan-object-uniform50-20260917.field-scale-dataset
Field-Scale Dataset
A large-scale benchmark dataset of field-scale 3D subsurface velocity
volumes (SOS-smoothed, depth-truncated to 619 samples) paired with 2D
velocity slices, their corresponding acoustic wavefields, and multi-source
shot-gather cubes. The dataset spans multiple geological settings and covers
five frequency bands (3-6, 3-8.5, 3-12, 3-17.5, 3-25 Hz), supporting wavefield
prediction, seismic inversion, source-aware training from the 64-source
shot-gather cubes, and… See the full description on the dataset page: https://huggingface.co/datasets/subsurfacegen/field-scale-dataset.BioRiskEvalscale_up_swegymskill-scarcity-index
Datamata Skill Scarcity Index
Which tech skills are genuinely hard to hire for: a daily composite scarcity score per skill built from how long roles stay open (time-to-fill), the salary premium employers pay over the category median and how often the same role is re-posted after failing to fill. Computed from active job listings across public company career pages and job boards.
Latest snapshot: 2026-09-23
Rows in this release: 16902
Updated: daily
Licence: CC BY 4.0 — free to… See the full description on the dataset page: https://huggingface.co/datasets/datamatastudios/skill-scarcity-index.strawberry_picking_dataset_scara_clustered
Strawberry Picking Dataset — SCARA Clustered
This LeRobot v3.0 dataset contains human-teleoperated demonstrations of a
4-DoF SCARA robot performing clustered strawberry picking in a controlled
physical mock-up. It accompanies the work
Learning to Pick: A Visuomotor Policy for Clustered Strawberry Picking,
which studies imitation-learned visuomotor control for reaching and picking a
target strawberry amid leaves, stems, and neighboring fruit.
The scene uses artificial strawberry… See the full description on the dataset page: https://huggingface.co/datasets/zfff/strawberry_picking_dataset_scara_clustered.tts-scaling-ladder-de-en
Balanced Expressive Speech: German–English Scaling Views
A nested set of ten cumulative training views for reproducible expressive TTS and audio-model studies. Eight base tiers were designed to be approximately 50/50 German/English and, within each language, 50/50 real/synthetic; each cell reserves half its volume for coverage of 40 EmoNet emotion categories and half for 57 VoiceNet dimensions × 10 value bins. Two later, non-overlapping extensions yield cumulative views of 49… See the full description on the dataset page: https://huggingface.co/datasets/laion/tts-scaling-ladder-de-en.ur5e_scanner_fp_deltaTCPThis dataset was created using LeRobot format.
Dataset Structure
{
"codebase_version": "v2.1",
"robot_type": "ur5e",
"total_episodes": 60,
"total_frames": 31577,
"total_tasks": 1,
"total_videos": 120,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:59"
},
"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/ns69956/ur5e_scanner_fp_deltaTCP.europeana-it-scans_beirThis is a copy of https://huggingface.co/datasets/jinaai/europeana-it-scans reformatted into the BEIR format. For any further information like license, please refer to the original dataset.
Disclaimer
This dataset may contain publicly available images or text data. All data is provided for research and educational purposes only. If you are the rights holder of any content and have concerns regarding intellectual property or copyright, please contact us at "support-data (at) jina.ai"… See the full description on the dataset page: https://huggingface.co/datasets/jinaai/europeana-it-scans_beir.math_stratos_scale_judged_and_annotated_with_difficultyllm-xray-lesion-scans
VG1 status · 22 September 2026 (Europe/Istanbul)
The 7B class opened on 21 September 2026. Any registered account may scan the repositories on the eligible list of the scope page — exact Apache-2.0 revisions, listed there with their status — within the free allowance of 20 browser scans and five distinct source models per calendar month. 7B-class repositories run as single-model quantization simulations. A comparison of two 7B-class checkpoints is currently accepted by the… See the full description on the dataset page: https://huggingface.co/datasets/tetracta/llm-xray-lesion-scans.ur5e_bt_rtde_10MA_scale_18.7_14.5This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "ur5e",
"total_episodes": 200,
"total_frames": 49277,
"total_tasks": 1,
"total_videos": 200,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:200"
},
"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/gribes02/ur5e_bt_rtde_10MA_scale_18.7_14.5.