GUIMid training data

A released GUI-agent mid-training mixture and supporting domain datasets, together with a post-training trajectory file and image archives for web and mobile interaction.

AI-assisted editorial review against a primary source. This is a catalog annotation, not an independent reproduction or endorsement.

What is this resource?

Resource type

Datasets

Environment

web, android

Topics

imitation-learning, planning, screen-understanding

Record publication date

Not recorded

Evidence retrieved

2026-10-17T11:23:07Z

Resource ID

url:422010da7a5c845a18f6

What licenses are recorded?

Code, training data, weights, and documents can have different terms. A blank declaration means unknown, not unrestricted use.

Code

Not recorded

Data

mit (Publisher dataset-card declaration for the released packaging; original source datasets retain their own terms.)

Document

Not recorded

Weights

Not recorded

License review notes

• Data: Publisher dataset-card declaration for the released packaging; original source datasets retain their own terms. Source 1

Dataset profile

Supervision

mid-training instruction supervision, GUI post-training trajectories

Modalities

text, images

Formats

JSON, tar.gz image archives

Splits

train configurations for named mid-training domains

Scale and counting units

Do not treat Hub size bands or the inconsistent introductory task count as a sample total. Some constituent-domain image packages remain forthcoming.

Compare source-reviewed profiles

Recorded specifications

availability

Ungated mixture/post-training JSON and image-archive manifests; some per-domain images and training code remain pending.

gated

false

hub id

hkust-nlp/GUIMid

release revision

b6a360815c843b5733674191b6e6d7fa68b2ea8c

Explore related topics

Scope and limitations

The public manifest includes GUIMid.json, mavis.tar.gz, GUI_trajectory.json, traj.tar.gz and domain-specific JSON files. The card says image data for some individual domains is still forthcoming and its overview inconsistently says nine tasks while later enumerating seven vision-language plus four language-only domains; no aggregate task/sample count is inferred. Training code remains marked coming soon in the repository. MIT is the publisher dataset declaration, not a relicensing of all upstream content.

Related resources

• documented by: Breaking the Data Barrier -- Building GUI Agents Through Task Generalization

Primary evidence and verification

Source retrieval is separate from experiment reproduction. Use the original publication or publisher page when citing claims about the resource.

1. editorial primary review

   Retrieved 2026-10-17T11:23:07Z. Supports: name, kind, summary, environments, topics, scope, notes, relations, details.dataset_profile, licenses.data.

   Content fingerprint

   1cdf702c819ecf370f4191c6c71fae7b579cfbaf766908b42156964a69eddacc
2. editorial primary review

   Retrieved 2026-10-17T11:23:07Z. Supports: identity, release_status, details.hub_id, details.gated, details.release_revision.

   Content fingerprint

   ededa2b3c8f21af8188152b028758c800db76cdbb8461ba82b8574cab375914e
3. editorial primary review

   Retrieved 2026-10-17T11:18:19Z. Supports: identity.

   Content fingerprint

   31b836da66c6907ff9a75ebc7f06da0db67e7d9fe9ca92d0c770f5f15f90a39e
4. editorial primary review

   Retrieved 2026-10-17T11:18:19Z. Supports: notes, relations.

   Content fingerprint

   d9f63a03088c37f7145725c7a182100d1593fceadd26a9bf817d434ca8d6297e

Suggest a correction

Open a source-backed correction. Include this record ID and the original source. Read our editorial policy and coverage boundaries.