Google launched the Model Card Toolkit designed to facilitate processes for AI-based model
transparency reporting used by downstream users, regulators, and developers. The toolkit is built on the Model Cards framework that is used for reporting on ethics-informed evaluation, usage, and provenance. It aims to provide a detailed overview of the model’s potential limitations and uses.
The latest Toolkit aims to make it easier for third-party vendors to easily develop Model cards by combining necessary data and helping in the development of easy interfaces for all users. The developers can specify the graphs and metrics needed to be displayed in the finished Model Card. This will include data on the model’s productivity which could divert from the final performance.