Open science and model cards in machine learning, Nov 3, 2026

Description  

Impactful and reproducible research requires that researchers can transfer their knowledge to others in an effective manner. When it comes to published machine learning models, they are nowadays accompanied by model cards that describe the model for other researchers and possible users in a clear and organized form. Model cards contain information such as benchmarks, limitations and intended use cases of your model, and they are used all throughout the ML and AI fields. 

This lecture will teach you how to read and write model cards and how you can use them to plan your model creation process. 

What is reproducibility and why do we need to care about it in machine learning research? What are the best practices for sharing machine learning models in your research? This lecture seeks to educate both machine learning scientists and non-scientists alike about transparency and reproducibility in machine learning.  

The lecture builds around the concept of a machine-learning model card. The idea of a machine learning model card was introduced in 2018 by Mitchell et al.. The basic idea of a model card is that it should contain information on the intended use cases of the model and possible limitations of the model. Since then, the concept has been used by all of the major players in the field: Hugging Face, OpenAI, Google, Meta to name a few. 

This lecture is a shortened, more generalized version of the Reproducibility and Open Science practices in machine learning event arranged by Aalto Scientific Computing and Lumi AI Factory: https://www.aalto.fi/en/events/reproducibility-and-open-science-practices-in-machine-learning-20may2026  

 Who can participate? 

 The webinar is aimed at staff and students. No need for prior expertise. Some presented solutions may be specific to Aalto University.  

Format  

A 1.5-hour webinar including a live Q&A discussion. The webinar is free of charge.  

Schedule and location  

The training will be held at the LUMI AI Factory hub TUAS Maarintie 8, and online via Zoom on November 3, at 1–2.30 PM Eastern European Time (EET)