datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
gazeta
Dataset Card for Gazeta
Dataset Summary
Dataset for automatic summarization of Russian news. News and their summaries are from the Gazeta website. Summaries were parsed as the content of an HTML tag with “description” property. Additional selection of good summaries was performed. There are two versions of this dataset.
Supported Tasks and Leaderboards
Leaderboard on Papers With Code: text-summarization-on-gazeta.
Please use the original evaluation script with… See the full description on the dataset page: https://huggingface.co/datasets/IlyaGusev/gazeta.gazzetta-ufficiale
Gazzetta Ufficiale 👩🏻⚖️⚖️🏛️📜🇮🇹
La Gazzetta Ufficiale della Repubblica Italiana, quale fonte ufficiale di conoscenza delle norme in vigore in Italia e strumento di diffusione, informazione e ufficializzazione di testi legislativi, atti pubblici e privati, è edita dall’Istituto Poligrafico e Zecca dello Stato e pubblicata in collaborazione con il Ministero della Giustizia, il quale provvede alla direzione e redazione della stessa. L'Istituto Poligrafico e Zecca dello Stato… See the full description on the dataset page: https://huggingface.co/datasets/mii-llm/gazzetta-ufficiale.retail_gaze
Dataset Card for retail_gaze
This is a FiftyOne dataset with 42 samples.
Installation
If you haven't already, install FiftyOne:
pip install -U fiftyone
Usage
import fiftyone as fo
from fiftyone.utils.huggingface import load_from_hub
# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = load_from_hub("Voxel51/retail_gaze")
# Launch the App
session = fo.launch_app(dataset)
Dataset Details
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/retail_gaze.trudovaya-gazeta
Newspaper «Trudovaya Gazeta»
Description
Newspaper of the Alatyr uezd committee of the RCP(b) and the Alatyr uezd executive committee, 1922–1929. From 17 Oct 1929 renamed «Red Chuvashia». Source: digitized 1923–1929.
Source
National Library of the Chuvash Republic: http://nbchr.ru/
Mirror: https://huggingface.co/chuvash-data
Газета «Трудовая газета»
Описание
Газета Алатырского уездного комитета РКП(б) и уездного исполкома, 1922–1929 гг. С… See the full description on the dataset page: https://huggingface.co/datasets/chuvash-data/trudovaya-gazeta.GAZE_soundspace
Unity + SoundSpaces 2.0 Spatial Audio Dataset (Replica)
Audio-visual spatial-audio dataset built by placing sound-emitting objects in
17 Replica indoor scenes with a Unity/PhysX physical placement stage, then
rendering binaural room impulse responses (RIR) with
SoundSpaces 2.0 (habitat-sim
AudioSensor, per-class acoustic materials). Each case ships the listener's
first-person render, per-source binaural RIRs, dry source audio, the
RIR-convolved per-source audio, a mixed binaural… See the full description on the dataset page: https://huggingface.co/datasets/mumbumble/GAZE_soundspace.GazeVaLMFull paper can be found here.
GazeVaLM: A Multi-Observer Eye-Tracking Benchmark for Evaluating Clinical Realism in AI-Generated X-Rays
This is the official dataset repository for the paper GazeVaLM: A Multi-Observer Eye-Tracking Benchmark for Evaluating Clinical Realism in AI-Generated X-Rays.
GazeVaLM is a public eye-tracking dataset for studying clinical perception during chest radiograph authenticity assessment. The dataset comprises 960 gaze recordings from 16 expert… See the full description on the dataset page: https://huggingface.co/datasets/davidcwong/GazeVaLM.Landscape-DataThis repository contains the dataset used in the paper Landscape of Thoughts: Visualizing the Reasoning Process of Large Language Models.
Code: https://github.com/tmlr-group/landscape-of-thoughts
This project is licensed under the MIT License. See the LICENSE.md file for details.
fold_towelbulgarian-state-gazette-ocreye-gaze-data
Eye Gaze Data
Eye gaze dataset collected as part of the Nymeria project. Contains per-frame gaze measurements from egocentric recordings across multiple participants and activities.
Dataset Structure
default (1099 files)
The main configuration. Contains binocular gaze data with separate left/right eye measurements.
Columns:
tracking_timestamp_us — timestamp in microseconds
sequence_uid, session_uid — recording identifiers
gaze_type — gaze classification… See the full description on the dataset page: https://huggingface.co/datasets/nymeriagazedata/eye-gaze-data.gaze_dataset_full
gaze_dataset_full — StreamGaze_v2 + EgoGazeVQA + HD-EPIC
A single repository containing three complementary benchmarks for
evaluating multimodal LLMs on gaze-grounded egocentric video question
answering:
Subfolder
Source
Held-out (val/test)
Questions
StreamGaze_v2/
egoexolearn, holoassist, egtea
egtea
8 MCQ tasks (4-opt) — gaze-conditioned past/present/future
EgoGazeVQA/
ego4d, egoexo, egtea
egtea
causal / spatial / temporal (5-opt)
HD-EPIC/
P01–P09
P09… See the full description on the dataset page: https://huggingface.co/datasets/Peanuttoad/gaze_dataset_full.GEM_gaze-assisted-ego-motion-in-drivingVRGaze
Dataset sample
VRGaze dataset sample:
Gaze360-split50_gazeuse_10h30_23-09_20260923_103845This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"action": {
"dtype": "float32",
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow_flex.pos",
"wrist_flex.pos",
"wrist_roll.pos",
"gripper.pos"
],
"shape": [
6… See the full description on the dataset page: https://huggingface.co/datasets/Tridex/50_gazeuse_10h30_23-09_20260923_103845.I2I-Paint
I2I-Paint
概要
AI画像生成をしていて「I2Iの為に画像を加工したいけど
フォトショやクリスタを起動するほどではない」 「液タブの電源を入れるほどではない」というような
ちょっとした加筆修正の時に使えるペイントソフトのpythonコードです。
出来た画像はクリップボードにコピーしてA1111のI2Iウインドウや
ComfyUIのload Imageノードにペーストできます。
実行にはpythonのインストールが必要です。
Win11で動作確認しています。私の5700Xの環境では高速でカーソルを動かすと線が途切れるので同等以上のCPU推奨です。
また、ペンタブ操作はウインドウが動いてしまうのでマウスモードへの切り替えが必要になりました。
（追記：モニタの設定を見直したら改善しました。https://www.xp-pen.jp/faq-0-222.html）
※2025/03/11 描画がもっさりしすぎだったので画面書き換え回数と補間の点の数を増やしました。ご自分のCPUに合わせてコード冒頭の以下の数値を書き換えてください。… See the full description on the dataset page: https://huggingface.co/datasets/Gazou-Seiri-Bu/I2I-Paint.gazette-hu
Magyar Közlöny — Hungarian official gazette (public-domain legal text)
Legislation, decrees and official notices from Magyar Közlöny,
Hungary's official gazette, cleaned from the source PDFs to raw prose. One row per paragraph,
with issue metadata (number, date, year, era) and a best-effort section (the current act
heading). Text only.
License
CC0-1.0. Under the Hungarian Copyright Act (Szjt., Act LXXVI/1999) §1(4), legislation
and other official documents… See the full description on the dataset page: https://huggingface.co/datasets/lazos/gazette-hu.turkish-resmi-gazete
Turkish Resmi Gazete (Official Gazette) — Full Corpus
Türkiye Resmî Gazete'sinin 2019-01-02'den 2026-07-17'ye kadar yayımlanan tüm sayılarının tam metnini içeren bir derlemdir. Kanunlar, yönetmelikler, Cumhurbaşkanı kararları, tebliğler, Anayasa Mahkemesi kararları, kurul kararları ve atama kararnameleri dahil olmak üzere Resmî Gazete'de o tarih aralığında yayımlanmış her türden resmî belge yer almaktadır.
İçerik ve Kaynak
Resmî Gazete, günlük yayınlarını… See the full description on the dataset page: https://huggingface.co/datasets/moganai/turkish-resmi-gazete.gazebo_push_box_3dof_v1Gaze-Co-Benchmark
Gaze-Co Benchmark
Paper: Gaze Target Estimation Anywhere with Concepts
Evaluation set for Promptable Gaze Target Estimation (PGE), the task introduced in
Gaze Target Estimation Anywhere with Concepts (CVPR 2026).
Each instance pairs an image with a natural-language description of one person and the
gaze target that person is looking at. A model receives the image and the text prompt — no
head bounding box, no detector, no pose — and must localize the subject and predict where… See the full description on the dataset page: https://huggingface.co/datasets/IrohXu/Gaze-Co-Benchmark.blueberry-cubes-gaze-1camThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "blueberry_ros",
"total_episodes": 222,
"total_frames": 121873,
"total_tasks": 3,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 15,
"splits": {
"train": "0:222"
},
"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/fecasado/blueberry-cubes-gaze-1cam.gaze-as-grounding-evidence
Gaze as Evidence for Common Grounding
Processed, window-level gaze features for Gaze as Evidence for Common
Grounding: A Cross-Corpus Analysis of MapTask and MUNDEX by Nan Li,
Albert Gatt and Massimo Poesio (MINT 2026).
Paper on arXiv ·
Hugging Face paper page ·
GitHub: data and analysis code
The dataset connects gaze measurements with reference-alignment annotations in
MapTask and retrospective understanding judgments in MUNDEX. Both corpora use
discrete behavioral gaze… See the full description on the dataset page: https://huggingface.co/datasets/chnln/gaze-as-grounding-evidence.my-gazebo-nav-dataset-v2GazeCtrl_dataset
GazeCtrl — human-curated gaze labels for body reorientation
Gaze labels for turnaround frames generated from single photos, used to train a
gaze-conditioned ControlNet that reorients a person's whole body toward a target gaze.
This repository contains labels only — no images. That is deliberate; see
Why there are no images here. The frames are regenerated
from the pipeline in the code repository, which makes every row here reproducible without
redistributing anyone's… See the full description on the dataset page: https://huggingface.co/datasets/baki0115/GazeCtrl_dataset.cm-gaza
cm-gaza
Verified in-the-wild phone footage with capture metadata, from a worldwide network of mappers · first regional
release: Gaza · v0.1 preview · Collective Memory
The global corpus
Collective Memory runs an active, worldwide network of mappers: individuals who document places and events where
they are, through an app that records the account, device time and GPS at the moment of capture and checks the media
and its metadata on submission. As of 2026-09… See the full description on the dataset page: https://huggingface.co/datasets/collectivememory/cm-gaza.NMT_Rwandan-Gazette_parallel_data_en_kin
Dataset Details
Dataset Description
This is a curated parallel dataset from the Official Gazette of the Republic of Rwanda. It has been curated to extract corresponding English and Kinyarwanda text and in the future we shall add French to the mix
Curated by: Digital Umuganda
Language(s) (NLP): Kinyarwanda and English
License: cc-by-4.0
Dataset Sources [optional]
The dataset original content was retrieved from the Rwandan ministry of Justice website… See the full description on the dataset page: https://huggingface.co/datasets/DigitalUmuganda/NMT_Rwandan-Gazette_parallel_data_en_kin.hd_epic_interaction_anticipation_bench_som_gaze
Dataset Card for HD-EPIC Gaze Interaction Anticipation: SoM & Gaze-Augmented Clips
Dataset Summary
This dataset provides visually-augmented versions of the Gaze Interaction Anticipation subset of the HD-EPIC VQA benchmark (Perrett et al., CVPR 2025), built for the paper:
Leveraging Gaze and Set-of-Mark in VLLMs for Human-Object Interaction Anticipation from Egocentric Videos
Daniele Materia, Francesco Ragusa, Giovanni Maria Farinella — ICPR 2026
The original… See the full description on the dataset page: https://huggingface.co/datasets/danielemateria/hd_epic_interaction_anticipation_bench_som_gaze.gazecaptureblueberry-cubes-gazeThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "blueberry_ros",
"total_episodes": 222,
"total_frames": 121873,
"total_tasks": 3,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 15,
"splits": {
"train": "0:222"
},
"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/fecasado/blueberry-cubes-gaze.gazebo_push_box_3dof_v2