Photonic computing for extreme-scale AI
Quand ? |
Le 13/12/2021, de 11:00 à 12:00 |
---|---|
Où ? | Salle Condorcet |
Participants |
Laurent Daudet |
Ajouter un événement au calendrier |
vCal iCal |
Recent « foundation » Machine Learning models, in the wake of OpenAi’s GPT-3 model for NLP, offer tremendous potential for research and applications. However, training such extreme-scale models requires massive amounts of computing resources, already challenging the capacity of some of the largest supercomputing architectures. In this talk I will present LightOn’s view on how future AI hardware and software should be co-designed, to address some of the hardest computing challenges, such for scaling up language models, recommender systems, or graph neural networks. In particular, I will discuss how Optical Processing Units (OPUs) can already be seamlessly integrated into a variety of hybrid photonics / silicon pipelines, leveraging randomized algorithms for Machine Learning and Scientific Computing.