How Meta's Muse works, revealed by the 6.8 GB filesystem it sent me

Meta's Muse is a research project focused on developing a general-purpose AI model that can be used for a variety of tasks. A 6.8 GB filesystem was sent to a researcher, containing the Muse runtime, which revealed details about the project. The filesystem includes a number of directories, including 'docs', 'experiments', and 'data', which suggest that the project is still in its early stages. The 'docs' directory contains a README file that describes the project's goals and objectives, while the 'experiments' directory contains a number of test runs and results. The 'data' directory contains a large amount of raw data that the project is likely using to train its model. The filesystem also includes a number of Python scripts and configuration files that suggest the project is built on top of the Hugging Face Transformers library.