Supporting non-IID data? #37
carsonkahn-external
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
One of the published papers cited, and HP's technical white paper, both note that a major motivation for Swarm is supporting non-IID data "unlike" Google's federated learning. How/why does Swam accomplish this better than traditional federated learning?
Beta Was this translation helpful? Give feedback.
All reactions