LIBERO Fastwam: Preprocessed Robot Learning Demonstrations for World Action Models

LIBERO MuJoCo 3.3.2 preprocessed robot demonstrations released by author yuanty in March 2026. This processed dataset supports training and evaluation for the Fast-WAM research project on world action models. It is distributed as four .tar.gz archives on Hugging Face.

Use Cases

• Training world action models based on preprocessed robot demonstration trajectories.
• Evaluating model performance on the LIBERO benchmark tasks based on the provided processed data.
• Reproducing results from the Fast-WAM research paper based on the paper-specific processed release.

Strengths

• Data is preprocessed and formatted for the LeRobot v2.1 framework, which may reduce setup time.
• Dataset is specifically released to support reproducibility for the associated research paper.
• Last updated on 2026-03-23, suggesting recent maintenance.

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, file size, and license information are unknown, which may limit suitability assessment.

Provenance

Source

Processed from the upstream LIBERO dataset, released by author yuanty on Hugging Face.

Collection Method

Preprocessed for the Fast-WAM research project; specific processing steps are not detailed in the provided input.

Freshness

2026-03-23 12:40:54