LR COASTER | Loss Landscape morphology & dynamics visualization | Deep Learning

LR Coaster visualizes a learning rate stress test during the training of a convnet. I use extreme changes in the learning rate to illustrate how the morphology and dynamics of the loss landscape change in response to the changes in the learning rate. The resolution used (300K loss values calculated per frame) allow us to understand the change in morphology. The net is using batchnorm, which during the training phase produces perturbations in the landscape. More details and related analysis about this and other visualizations will be published in the future.
Loss Landscape generated with real data: real data, convnet, imagenette dataset, sgd-adam, bs=16, bn, lr sched, train mod, 300k pts, 0.5 w range, log scaled (orig loss nums) & vis-adapted, net trained with fast.ai
In the intersection between research and art, the A.I LL project explores the morphology and dynamics of the fingerprints left by deep learning optimization training processes.
The project goes deep into the training phase of these processes and generates high quality visualizations, using some of the latest deep learning and machine learning research and producing inspiring animations that can both inform and inspire the community.
As the weight space changes through the optimization process, loss landscapes become alive, organic entities that challenge us to unlock the mysteries of learning.
How do these multidimensional entities behave and change as we modify hyperparameters and other elements of our networks?
How can we best tame these wild beasts as we cross their edge horizon on our way to the deepest convexity they hold?
losslandscape.com