Tech Talk: InstaDeep

Instadeep will be discussing their joint work with NVIDIA on nucleotide transformers, a collection of new state-of-the-art large language models trained on DNA, and how it’s transforming genomics research. The four models with varying parameter sizes, from 500 million to 2.5 billion, have been trained on a diverse datasets of 3,200 human genomes and 850 genomes from various species. The talk will highlight the innovative ways in which this technology can be used to analyze and interpret complex genomic data and how JAX has been instrumental in this undertaking. 

Dr. Thomas Pierrot is a Senior Research Scientist at InstaDeep since 2018. He possesses a background in aeronautical engineering and holds a PhD in deep reinforcement learning. During his time in Paris, he built and expanded a research team from two to ten members. Recently, he relocated to the United States to establish InstaDeep Boston’s office and build a new team there. Dr. Pierrot and his team are actively engaged in conducting research on deep reinforcement learning, neuroevolution, and the application of deep learning to life sciences, with a particular focus on protein design and genomics. They aim to provide innovative solutions to complex challenges in these fields. Furthermore, the team is spearheading internal initiatives to develop, train, and deploy InstaDeep’s next generation of large language models on a large scale for life sciences applications.

Lunch will be provided, please RSVP here.

Hosted by the EECS Alliance

Details

• Date: Tuesday, April 11
• Time: 12:00 pm - 1:00 pm
• Location: Allen Room 36-462