As an end user, I want to run a single simulation as in the Krum paper Machine learning with adversaries: byzantine tolerant gradient descent (single parameter server case), for an arbitrary choice of model, aggregator and attack.
Implementation should follow the docs/adr-2026-05-22.
Requires a minimal implementation of the package primitives
As an end user, I want to run a single simulation as in the Krum paper Machine learning with adversaries: byzantine tolerant gradient descent (single parameter server case), for an arbitrary choice of model, aggregator and attack.
Implementation should follow the docs/adr-2026-05-22.
Requires a minimal implementation of the package
primitives