# No one has solved federated learning on heterogeneous GPUs

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2024-06-19
- **Theme:** Decentralized AI — https://www.gregosuri.com/predictions/decentralized-ai/
- **Source:** Akash: The Internet Of Compute W/ Founder Greg Osuri ($1 To $1 Million Podcast)
- **Watch at:** 00:15:57 — https://www.youtube.com/watch?v=jyGP_9A1SVs&t=957s
- **Canonical:** https://www.gregosuri.com/predictions/decentralized-ai/2024-06-19-no-one-has-solved-federated-learning-on-heterogene-mb6i/
- **Cite as:** Greg Osuri, "No one has solved federated learning on heterogeneous GPUs," Akash: The Internet Of Compute W/ Founder Greg Osuri ($1 To $1 Million Podcast), 19 Jun 2024, 00:15:57. https://www.youtube.com/watch?v=jyGP_9A1SVs&t=957s

## Quote — verbatim

> Mark my words: no one solved federated learning on heterogeneous GPUs... While it's possible to train over geographically distributed clusters, it's not practical because it takes much longer.

## Context — site annotation, not spoken

Rebutting io.net's claim of having solved it; cites Google DeepMind's DiLoCo paper and teams (Gensyn, Prime Intellect, Flock.io, Nous Research) still working on the problem.