datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
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.MolmoPoint-TrackSyn
MolmoPoint-TrackSyn Dataset
Synthetic point tracking annotations for procedurally generated videos generated with Blender.
Each example contains an expression describing an object, per-frame point trajectories, and video metadata. All videos are encoded as 6 FPS and points are sampled at 2 FPS.
Dataset Statistics
Video Source
Unique Annotations
Unique Videos
static-camera
34,324
11,629
dyna-camera
41,841
14,158
Total
76,165
25,787
Schema… See the full description on the dataset page: https://huggingface.co/datasets/allenai/MolmoPoint-TrackSyn.MolmoPoint-TrackAny
MolmoPoint-TrackAny Dataset
Point tracking annotations for natural videos from YouTube and MammalNet sources.
Each example contains an expression describing an object, per-frame point trajectories, and metadata linking back to the source video.
Two sampling-FPS variants (1 and 2) are concatenated into a single train split.
Dataset Statistics
The dataset contains 17,437 unique annotations across 13,050 videos. Each annotation appears twice — once sampled at 1 FPS and once… See the full description on the dataset page: https://huggingface.co/datasets/allenai/MolmoPoint-TrackAny.