World Wires · story 109848 · corroborated · 1 source(s)

FedReason: LLMs as weight-space meta-aggregators for federated learning

Researchers propose FedReason, a method where large language models select aggregation rules in heterogeneous federated learning environments.

Open in the desk

Coverage

• FedReason: LLMs as weight-space meta-aggregators for federated learning arXiv: new preprints (security, AI, physics) · tier C · clear net · 2026-11-05 18:00 UTC

What this site indexes