datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
WebLINX-full
WebLINX: Real-World Website Navigation with Multi-Turn Dialogue
WARNING: This is not the main WebLINX data card! You might want to use the main WebLINX data card instead:
WebLINX: Real-World Website Navigation with Multi-Turn Dialogue
WebLINX: Real-World Website Navigation with Multi-Turn Dialogue
Xing Han Lù*, Zdeněk Kasner*, Siva Reddy
💾Code
📄Paper
🌐Website
📓Colab
🤖Models
💻Explorer
🐦Tweets
🏆Leaderboard
Your browser does not support the… See the full description on the dataset page: https://huggingface.co/datasets/McGill-NLP/WebLINX-full.WebSight
Dataset Card for WebSight
Dataset Description
WebSight is a large synthetic dataset containing HTML/CSS codes representing synthetically generated English websites, each accompanied by a corresponding screenshot.
This dataset serves as a valuable resource for tasks such as generating UI codes from a screenshot.
It comes in two versions:
v0.1: Websites are coded with HTML + CSS. They do not include real images.
v0.2: Websites are coded with HTML + Tailwind CSS. They do… See the full description on the dataset page: https://huggingface.co/datasets/HuggingFaceM4/WebSight.conceptual-captions-12m-webdatasetEdge-Agent-Reasoning-WebSearch-260K
Edge Agent Reasoning WebSearch 260K
Abstract
The Edge-Agent-Reasoning-WebSearch-260K dataset is a massive, synthetically expert-engineered corpus of over 700 Million tokens, designed to train small, local models (SLMs) and edge-deployed agents in advanced problem deconstruction and self-aware reasoning.
Rather than training a model to execute instructions directly—which often leads to hallucinations when context is missing—this dataset trains a model to act as a… See the full description on the dataset page: https://huggingface.co/datasets/yatin-superintelligence/Edge-Agent-Reasoning-WebSearch-260K.webcode2m_purifiedWebCode2M: A Real-World Dataset for Code Generation from Webpage Designs
Features:
image: the screenshot of the webpage.
bbox: the layout information, i.e., the bounding boxes (Bbox) of all the elements in the webpage, which contains the size, position, and hierarchy information.
text: the webpage code text including HTML/CSS code.
scale: the scale of the screenshot, in the format [width, height].
lang: the main language of the text content displayed on the rendered page (excluding HTML/CSS… See the full description on the dataset page: https://huggingface.co/datasets/xcodemind/webcode2m_purified.food-dataset
Food Dataset
An image classification dataset of food photos organized into 201 categories (folders), with 35,046 images total (~924 MB).
Each top-level folder is a category (e.g. adana kebab, sushi, waffles, tiramisu, ...) containing JPEG images of that food/dish. This follows the standard Hugging Face imagefolder layout, so it loads directly with:
from datasets import load_dataset
ds = load_dataset("webbrain-one/food-dataset")
Structure
<category… See the full description on the dataset page: https://huggingface.co/datasets/webbrain-one/food-dataset.pickapic_v2_webdatasetwebdataset archive of yuvalkirstain/pickapic_v2.
Dataloading code can be found here.
webcode2mWebCode2M: A Real-World Dataset for Code Generation from Webpage Designs with Layouts
(This dataset is also called Vision2UI.)
Automatically generating webpage code from webpage designscan significantly reduce the workload of front-end developers, andrecent Multimodal Large Language Models (MLLMs) have shownpromising potential in this area. However, our investigation revealsthat most existing MLLMs are constrained by the absence of highquality, large-scale, real-world datasets, resulting in… See the full description on the dataset page: https://huggingface.co/datasets/xcodemind/webcode2m.ShowUI_Web
Dataset Card for ShowUI_Web
This is a FiftyOne dataset with 21988 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/ShowUI_Web")
# Launch the App
session = fo.launch_app(dataset)
Web Dataset from ShowUI… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/ShowUI_Web.danbooru2023-webp-4Mpixel-224The data set is just resized to 224*224
https://huggingface.co/datasets/KBlueLeaf/danbooru2023-webp-4Mpixel
Pseudo code for processing
def resize_image(file_path):
with Image.open(file_path) as img:
resized_img = img.resize((224, 224))
resized_img.save(file_path)
WebCompass
WebCompass
A unified multimodal benchmark for evaluating LLMs' ability to generate, edit, and repair functional web pages. WebCompass spans three input modalities — text design documents, reference screenshots, and video demonstrations — and three task families — generation, editing, and repair.
GitHub: NJU-LINK/WebCompass
Project Page: nju-link.github.io/WebCompass
Quick Start
from datasets import load_dataset
# Generation tasks (existing)
ds_text =… See the full description on the dataset page: https://huggingface.co/datasets/NJU-LINK/WebCompass.cc12m-webdataset
CC12M WebDataset
这是CC12M数据集的WebDataset格式版本。
数据集信息
文件数量: 1098
总大小: 888796.33 MB
上传时间: 2025-03-18 14:45:49
使用方法
import webdataset as wds
dataset = wds.WebDataset("https://huggingface.co/yangyang857658468/cc12m-webdataset/resolve/main/cc12m_*.tar")
stable-diffusion-webui
Stable Diffusion web UI
A browser interface based on Gradio library for Stable Diffusion.
Features
Detailed feature showcase with images:
Original txt2img and img2img modes
One click install and run script (but you still must install python and git)
Outpainting
Inpainting
Color Sketch
Prompt Matrix
Stable Diffusion Upscale
Attention, specify parts of text that the model should pay more attention to
a man in a ((tuxedo)) - will pay more attention to tuxedo
a man in a… See the full description on the dataset page: https://huggingface.co/datasets/PennyJX/stable-diffusion-webui.webarena-infinity-trajectories
WebArena-Infinity: Browser Agent Trajectories
Successful browser-agent trajectories collected from the WebArena-Infinity benchmark. WebArena-Infinity is a scalable approach for automatically generating realistic web environments paired with verifiable tasks, enabling robust training and evaluation of general-purpose browser agents. Each trajectory records a browser agent completing a real web-application task — including step-by-step screenshots, agent reasoning, and actions.… See the full description on the dataset page: https://huggingface.co/datasets/webarena-x/webarena-infinity-trajectories.dan-webp-newguiact_websingle_test
Dataset Card for GUIAct Web-Single Dataset - Test Set
This is a FiftyOne dataset with 1410 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/guiact_websingle_test")
# Launch the App
session = fo.launch_app(dataset)… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/guiact_websingle_test.Web2Code
Dataset Details
Our Web2Code instruction tuning dataset construction and instruction generation process involves four key components: (1) Creation of new webpage image-code pair data: We generated high-quality HTML webpage-code pairs following the CodeAlpaca prompt using GPT-3.5 and convert them into instruction-following data. (2) Refinement of existing webpage code generation data: We transform existing datasets including into an instruction-following data format similar to LLaVA… See the full description on the dataset page: https://huggingface.co/datasets/MBZUAI/Web2Code.1k_Website_Screenshots_and_Metadata
Dataset Card for 1000 Website Screenshots with Metadata
Dataset Summary
Silatus is sharing, for free, a segment of a dataset that we are using to train a generative AI model for text-to-mockup conversions. This dataset was collected in December 2022 and early January 2023, so it contains very recent data from 1,000 of the world's most popular websites. You can get our larger 10,000 website dataset for free at: https://silatus.com/datasets
This dataset includes:
High-res… See the full description on the dataset page: https://huggingface.co/datasets/silatus/1k_Website_Screenshots_and_Metadata.webui
WebUI
A large-scale dataset pairing real-world UI screenshots with their original HTML, CSS, and JavaScript source code, per-viewport bounding boxes for every visible DOM element, and GPT-4.1 vision descriptions. Every sample is rendered at three responsive breakpoints. Built from public design systems, component libraries, open-source projects, and community code — not synthetically generated.
Overview
Stat
Value
Total rows
36,807
Unique UI samples
12… See the full description on the dataset page: https://huggingface.co/datasets/ronantakizawa/webui.InternVL-SA1B-Caption-WebDatasetThis repo contains the recaptioned SA1B images in webdataset format. The recaptioned prompts are from https://huggingface.co/datasets/OpenGVLab/InternVL-SA-1B-Caption
distilled-web
Dataset Card for lenamerkli/distilled-web
This dataset consists of web-scraped data using a custom crawler purpose-built for each website.
Dataset Details
Dataset Sources
Repository: https://github.com/lenamerkli/distilled-web
Uses
This dataset is useful for training large language models.
The train split provides instruction-following and chat data for supervised fine-tuning (SFT) and instruction tuning.
The pretrain split… See the full description on the dataset page: https://huggingface.co/datasets/lenamerkli/distilled-web.MegaSynth-webdatasetM3-webWebSTAR
WebSTAR: WebVoyager Step-Level Trajectories with Augmented Reasoning
Dataset Description
WebSTAR (WebVoyager Step-Level Trajectories with Augmented Reasoning) is a large-scale dataset for training and evaluating computer use agents with step-level quality scores. This dataset is part of the research presented in "Scalable Data Synthesis for Computer Use Agents with Step-Level Filtering" (He et al., 2025).
Unlike traditional trajectory-level filtering approaches… See the full description on the dataset page: https://huggingface.co/datasets/microsoft/WebSTAR.typeface_dataset
Typeface Dataset
Designed by COLIGNUM/webXOS (x.com/colignum)
Under Development.
github.com/webxos for more info.
Dataset Information
Name: colignum_typeface_dataset_512px_2026-01-22T01-21-36-081Z
Total Characters: 155
Resolution: 512×512 pixels
Format: PNG + CSV/Parquet
Generated: 2026-01-22
Character Sets Included
A-Z Uppercase (26 characters)
a-z Lowercase (26 characters)
0-9 Numbers (10 characters)
Symbols (32 characters):… See the full description on the dataset page: https://huggingface.co/datasets/webxos/typeface_dataset.WebGym
cua-lite/WebGym
WebGym gpt-5.5 trajectories collected on WebGym OmniBoxes tasks and judge via the cua-lite webgym runtime — 25/60/15 site-start difficulty tiers + curated popular-2102 pool, balanced prompt, serp-only+loop cleaned
Origin
https://github.com/microsoft/webgym
https://huggingface.co/datasets/microsoft/webgym_tasks
https://arxiv.org/abs/2601.02439
Load via datasets
from datasets import load_dataset
# entire dataset
ds =… See the full description on the dataset page: https://huggingface.co/datasets/cua-lite/WebGym.mind2web_multimodal_test_website
Dataset Card for Multimodal Mind2Web "Cross-Website" Test Split
Note: This dataset is the test split of the Cross-Website dataset introduced in the paper.
This is a FiftyOne dataset with 1019 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 =… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/mind2web_multimodal_test_website.WebSightDescribed
Dataset Card for WebSightDescribed
Dataset Description
WebSightDescribed is a subset of WebSight v0.1, augmenting the dataset with
synthetically generated natural language descriptions of the websites.
This dataset serves as a valuable resource for the task of generating html code from a natural language description.
Details for WebSightDescribed
Data Fields
An example of a sample appears as follows:
{
'image': PIL.Image,
'id': int,
'html':… See the full description on the dataset page: https://huggingface.co/datasets/haidark1/WebSightDescribed.website-media
OpenRAL — website media
Video clips shown in the "See it run" section of openral.com
(benchmarks, simulation and on-hardware deployment runs).
Each clip lives under <category>/<benchmark>_<rskill>_<success|fail>/ with three
web-optimised assets:
poster.jpg — first-frame thumbnail
preview.mp4 — square 640px, muted (the autoscroll strip)
full.mp4 — native aspect, ≤1080p, with audio (the expand modal)
Generated and published by scripts/build-media.mjs in the
website repo.… See the full description on the dataset page: https://huggingface.co/datasets/OpenRAL/website-media.phishing-website-screenshots
Phishing Website Screenshots
A dataset of 8,370 full-page website screenshots labelled as legitimate or phishing, intended for training and evaluating visual phishing-detection models.
Contents
Label
label
Images
legitimate
0
7,924
phishing
1
446
Total
8,370
Screenshots were captured at a desktop viewport (1920×1080) as PNG images.
Structure
legitimate/<brand>/<page>.png
phishing/<source>/<page>.png
metadata.csv
metadata.csv… See the full description on the dataset page: https://huggingface.co/datasets/shresthsamyak/phishing-website-screenshots.