Skip to main content

Graphcore IPUs Double Speeds On Drug Discovery

LabGenius finds drug-discovery speed doubled on Graphcore IPUs compared to GPUs for machine learning model training.
Created on April 21|Last edited on April 21
Using AI in the field of medicine is something that's seriously catching on quick recently, especially in the area of drug discovery. The race to inovate is always at it's peak in drug discovery, and Intelligence Processor Units (IPUs) are a relatively new tool that aid significantly in the goal of staying ahead of the competition.
London-based biotech firm LabGenius uses machine learning and AI to help develop new drug treatments for diseases, particularily the process of generating proteins with the potential of being used in drugs as treatment.
LabGenius had previously been using GPUs to power their machine learning models, however they chose to switch to using Graphcore IPUs through cloud services and found that the time it took to train their models was halved, going down from a month to just two weeks.
Graphcore's IPUs are just one of several ML-focused processing units on the market today, as many companies have sprung up to give their take on the machine learning computation sphere.

Find out more

Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.