We are introducing Podcast 14M, an open-source, restricted-access dataset released under the Fathom1 licence, through which we retain exclusive rights to determine how and with whom the dataset may be shared. Podcast 14M was inspired by existing closed-source datasets used to develop open-source models, despite the underlying training data remaining inaccessible to the research community. Our goal is to bridge this gap by providing access to a portion of such data, enabling researchers to train and develop high-quality, smaller-scale models.
Podcast 14M is released under Project Imagination, an initiative dedicated to creating and sharing innovative datasets that contribute meaningfully to the model training and research community.
Developed by Sleeping AI.
- Downloads last month
- 35