# Decentralized training is starting to work

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2024-05-29
- **Theme:** Decentralized AI — https://www.gregosuri.com/predictions/decentralized-ai/
- **Source:** Akash Accelerate '24: Official Livestream (Akash Network)
- **Watch at:** 01:30:48 — https://www.youtube.com/watch?v=60mmj3bNzB0&t=5448s
- **Canonical:** https://www.gregosuri.com/predictions/decentralized-ai/2024-05-29-decentralized-training-is-starting-to-work-mqe7/
- **Cite as:** Greg Osuri, "Decentralized training is starting to work," Akash Accelerate '24: Official Livestream (Akash Network), 29 May 2024, 01:30:48. https://www.youtube.com/watch?v=60mmj3bNzB0&t=5448s

## Quote — verbatim

> I think we're seeing a lot of progress in the decentralized training space... especially a lot of the research papers that are coming out, like DiLoCo and whatnot, that propose training on heterogenous, heavily distributed compute — so seems to be promising.

## Context — site annotation, not spoken

Greg responding to Eric Voorhees' warning that open-source AI's future depends on solving efficient decentralized training before Meta stops releasing frontier models. (Captions garble the paper name; DiLoCo is the likely referent.)