datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Gui-agent
Gui-Agent — GUI trajectories in LIBERO/VLA format
Human GUI demonstrations from four sources, unified into a single VLA-style
intermediate representation and written as LIBERO-layout HDF5, so LIBERO/VLA
dataloaders run against GUI data unchanged.
raw source ──[adapter]──> GuiEpisode ──[writer]──> LIBERO-style HDF5
per-source the IR format- what you train on
only specific
25,872 episodes / 453,264 steps / 235 GB… See the full description on the dataset page: https://huggingface.co/datasets/Yushi123/Gui-agent.easyr1-grounding-dataset-30k-not_grounded-SE-GUI-3B-2MPguiowl-aw-mix-full
GUI-Owl AndroidWorld SFT Mix — FULL / generalization
Purpose: AndroidWorld (116-task) SFT for a GUI-Owl-1.5-2B block-diffusion VLA.
This dataset is an action-balanced, source-mixed SFT corpus assembled from five Android
GUI-agent trajectory sources. It is built for in-domain supervised fine-tuning ahead of RL.
The mix deliberately includes AndroidWorld task-family coverage (via the openmobile anchor,
whose app field holds AW task-family names) and accepts in-domain overlap by… See the full description on the dataset page: https://huggingface.co/datasets/KMK040412/guiowl-aw-mix-full.guitarset
GuitarSet
GuitarSet v1.1.0 (Xi et al. 2018) on HuggingFace. 360 rows × 4 audio captures per row + canonical labels derived from the original JAMS. Three documented upstream errata are corrected (see below); the original file-based distribution lives on Zenodo with a permanent DOI. CC-BY 4.0.
Schema
Column
Type
Notes
track_id
string
e.g. 00_BN1-129-Eb_comp
player
int32
0–5
style
string
comp | solo
tempo_bpm
float64
key, mode, key_mode
string
e.g. Eb… See the full description on the dataset page: https://huggingface.co/datasets/jhartquist/guitarset.MolmoPoint-GUISyn
MolmoPoint-GUISyn
MolmoPoint-GUISyn is a large-scale synthetic dataset of 36K GUI screenshots with dense pointing annotations for training GUI grounding agents. Each screenshot is a realistic simulation of a digital environment (desktop apps, mobile apps, websites) generated entirely from code, with an average of 54 annotated UI elements per image.
The data is generated using the MolmoPoint-GUISyn pipeline, with Claude Sonnet 4.6 as the coding LLM.
Quick links:
Model:… See the full description on the dataset page: https://huggingface.co/datasets/allenai/MolmoPoint-GUISyn.guiowl-aw-mix-targeted
GUI-Owl AndroidWorld SFT Mix — TARGETED / in-domain
Purpose: AndroidWorld (116-task) SFT for a GUI-Owl-1.5-2B block-diffusion VLA.
This dataset is an action-balanced, source-mixed SFT corpus assembled from five Android
GUI-agent trajectory sources. It is built for in-domain supervised fine-tuning ahead of RL.
The mix deliberately includes AndroidWorld task-family coverage (via the openmobile anchor,
whose app field holds AW task-family names) and accepts in-domain overlap by… See the full description on the dataset page: https://huggingface.co/datasets/KMK040412/guiowl-aw-mix-targeted.DATA_SOURCEguiowl-aw-mix-hybrid-packed
guiowl-aw-mix-hybrid-packed
Episode-PACKED AndroidWorld-SFT mix for GUI-Owl-1.5-2B block-diffusion VLA.
600,040 steps / 57,669 episodes (whole-episode, ordered, action-trace history). 151 shards.
Source mix: openmobile 28% (AW-app anchor), gui_odyssey 26% (long/cross-app), aitw 18% (visual ballast), androidcontrol 16% (open donor), amex 12%.
Episode-presence: type 64% / swipe 57% / terminate 45% / open 13% / answer 7%. ep_len mean 10.4, p90 20, max 60.
Validated:… See the full description on the dataset page: https://huggingface.co/datasets/KMK040412/guiowl-aw-mix-hybrid-packed.GUIDEagentnet-partial-and-fail-v1
agentnet-partial-and-fail-v1
GUI state transitions (s, a, s') walked on an Ubuntu desktop by
Qwen3.8-27B, from tasks taken from AgentNet and run inside OSWorld's
Docker environment.
Walks the judge ruled partial or failed. These are the larger half and, for a world model, the more useful one: a walk that did not finish still opened dialogs, switched tabs and changed settings, and each of those is a real transition. Measured over eighty-eight walks, a failure visits 13.5 distinct… See the full description on the dataset page: https://huggingface.co/datasets/gui-wm/agentnet-partial-and-fail-v1.GUIOdyssey
cua-lite/GUIOdyssey
cua-lite preprocessed version of GUIOdyssey (hflqf88888/GUIOdyssey). A long-horizon cross-app Android mobile dataset of 8,334 task trajectories over ~128k screenshots. Produces two cohorts: use (multi-step agent episodes) and understanding (per-step screen captioning from the source description annotations).
Origin
https://huggingface.co/datasets/hflqf88888/GUIOdyssey
Load via datasets
from datasets import load_dataset
#… See the full description on the dataset page: https://huggingface.co/datasets/cua-lite/GUIOdyssey.GUIrilla-Task
GUIrilla-Task
Ground-truth Click & Type actions for macOS screenshots
Dataset Summary
GUIrilla-Task pairs real macOS screenshots with free-form natural-language instructions and precise GUI actions.
Every sample asks an agent either to:
Click a specific on-screen element, or
Type a given text into an input field.
Targets are labelled with bounding-box geometry, enabling exact evaluation of visual-language grounding models.
Data were gathered automatically by… See the full description on the dataset page: https://huggingface.co/datasets/macpaw-research/GUIrilla-Task.GUI-360
cua-lite/GUI-360
cua-lite preprocessed version of GUI-360 (vyokky/GUI-360), a large-scale dataset of computer-using-agent trajectories on Windows Microsoft-Office apps (Word / Excel / PowerPoint). Provides three desktop task types derived from the successful training trajectories: multi-step use, point grounding (intent -> element coordinate), and screen parsing (listing all interactive UI controls).
Origin
https://huggingface.co/datasets/vyokky/GUI-360… See the full description on the dataset page: https://huggingface.co/datasets/cua-lite/GUI-360.gui-odyssey-1kfineweb-atlas
FineWeb Atlas (v0.1)
FineWeb Atlas annotates 14.9 million FineWeb documents (95.5M chunks, 10.2B tokens) with 16,790 human-readable concepts spanning entities, topics, tones, and document types. Each chunk receives ~15 concept labels on average. The release includes chunk- and document-level annotations, a concept metadata table with prevalence stats, a reverse index for concept-first retrieval, and a packed cooccurrence matrix.
For background on how the atlas was built, see the… See the full description on the dataset page: https://huggingface.co/datasets/guidelabs/fineweb-atlas.guiowl-aw-mix-phoneonly-packedText_Guided_Image_Editing
Dataset Card
Dataset in ImagenHub.
Citation
Please kindly cite our paper if you use our code, data, models or results:
@article{ku2023imagenhub,
title={ImagenHub: Standardizing the evaluation of conditional image generation models},
author={Max Ku and Tianle Li and Kai Zhang and Yujie Lu and Xingyu Fu and Wenwen Zhuang and Wenhu Chen},
journal={arXiv preprint arXiv:2310.01596},
year={2023}
}
spine
GUI World Model — Spine Transitions
(s, a, s') transitions collected by walking task instructions on a live
Ubuntu desktop. Every state is captured from the running machine: a screenshot,
the accessibility tree as XML, and the rendered element table the model reads.
This set is spine only — the path an agent actually took. No branches.
Where the instructions come from
instruction_source
what it is
agentnet
Human recordings of people using their own… See the full description on the dataset page: https://huggingface.co/datasets/gui-wm/spine.c12b7guitarsetguiact-web-singleGUIEnvGUI-Perturbed
GUI-Perturbed
A step-level GUI grounding dataset built on domain-randomized web pages for diagnosing visual and spatial heuristics in VLM agents.
📄 Technical Report · 🌐 Baseline Result Viewer · 💻 Code
Overview
GUI-Perturbed is an evaluation dataset for step-level GUI element localization. It is designed to expose and precisely diagnose the failure modes of vision-language model (VLM) GUI agents to examine whether models rely on rigid visual shortcuts rather… See the full description on the dataset page: https://huggingface.co/datasets/figai/GUI-Perturbed.benchability-fig4-capability-guided
BenchAbility Figure 4 -- capability_guided
One of two training mixtures drawn from the same frozen 884,143-row candidate pool, with the
same budget (60,000 intervention + 15,000 shared replay) and the same hyperparameters. The two
differ only in how the samples are chosen, which is the whole experiment.
arm
capability_guided
selection
by capability, gap-weighted from the Figure 3 scores
intervention rows
59,999
replay rows
15,000
shards
38
pool
884,143… See the full description on the dataset page: https://huggingface.co/datasets/realzL/benchability-fig4-capability-guided.gui-libra-aw-packed
GUI-Libra AW-Packed (mobile_use, episode-packed)
Curated from GUI-Libra/GUI-Libra-81K-SFT (android subsets: aitw, amex,
android_control, gui-odyssey, coat-terminal) into the Fast-dVLM / GUI-Owl-1.5-2B
packed-parquet schema, action-balanced for AndroidWorld.
Schema matches KMK040412/guiowl-aw-mix-hybrid-packed EXACTLY:
source, episode_id, step_id, app, instruction, history, screenshot, screen_width, screen_height, target_json, action_type, coordinate, coordinate2, ep_len.… See the full description on the dataset page: https://huggingface.co/datasets/KMK040412/gui-libra-aw-packed.open-materials-guide-0210-embeddingsipfs_guinea_laws
Guinea (Guinee) Laws and Journal Officiel (sgg.gov.gn / JO / CNT)
Research snapshot of official national legislation from *Journal Officiel (journal-officiel.sgg.gov.gn) + SGG (sgg.gov.gn) + CNT (cnt.gov.gn) + ministry .gov.gn.
Not legal advice. The official gazette / authentic source prevails over this corpus.
Snapshot
Field
Value
Snapshot date
2026-09-18
Coverage
catalog-backed incomplete (JO live + SGG downloadfile + CNT archive assemblee +… See the full description on the dataset page: https://huggingface.co/datasets/endomorphosis/ipfs_guinea_laws.codocbench
Dataset Card for CoDocBench: A Dataset for Code-Documentation Alignment in Software Maintenance
Dataset Sources
Repository: https://github.com/kunpai/codocbench
Paper: https://arxiv.org/abs/2502.00519
Citation
BibTeX:
@misc{pai2025codocbenchdatasetcodedocumentationalignment,
title={CoDocBench: A Dataset for Code-Documentation Alignment in Software Maintenance},
author={Kunal Pai and Premkumar Devanbu and Toufique Ahmed}… See the full description on the dataset page: https://huggingface.co/datasets/guineapig/codocbench.gui-libra-aw-packed-phone
GUI-Libra AW-Packed PHONE-PORTRAIT (mobile_use, episode-packed)
Phone-portrait-only filter of
KMK040412/gui-libra-aw-packed,
maximizing AndroidWorld (phone-portrait benchmark) relevance.
Drop criterion: an episode is landscape (dropped) if any of its steps has
screen_height / screen_width < 1.0. Portrait episodes are kept fully and intact.
Filtering is episode-level so multi-turn episodes stay contiguous.
Schema is byte-identical to the source (same columns/dtypes, target_json… See the full description on the dataset page: https://huggingface.co/datasets/KMK040412/gui-libra-aw-packed-phone.guitar-fretboard-notes
Guitar Single-Note Recordings
A dataset of 390 single-note guitar recordings spanning 6 strings and frets 0-12, recorded by two players on acoustic and electric guitars.
Dataset Summary
This dataset contains isolated single-note recordings from a standard-tuned guitar. Each recording captures one note played on a specific string and fret combination, covering the first 12 frets across all 6 strings (78 unique notes per source). The recordings are raw, unprocessed 44100 Hz… See the full description on the dataset page: https://huggingface.co/datasets/collegefishiesd/guitar-fretboard-notes.