10122025 Box 01

A subset of the MolmoAct2-BimanualYAM dataset, a large-scale collection for robot learning. It contains more than 720 hours of training demonstrations for diverse tabletop manipulation tasks. The dataset was created by AllenAI using LeRobot and includes annotated language instructions.

Use Cases

• Training robot learning models based on bimanual manipulation demonstrations.
• Developing instruction-following robotic agents based on annotated language commands.
• Benchmarking multi-arm robot policies on tabletop tasks.
• Studying human-like dexterous manipulation strategies from recorded demonstrations.

Strengths

• Contains more than 720 hours of training demonstrations across the full collection.
• Focuses on bimanual manipulation, a complex and relevant robotics domain.
• Includes annotated language instructions, enabling language-conditioned learning.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.

Provenance

Source

AllenAI

Collection Method

Created using LeRobot; consists of robot manipulation demonstrations.

Freshness

Last updated 2026-05-05 14:43:51; freshness should be verified.