“We did clustering of building blocks on similarity and avoided entire clusters from being overlapped between test and train… performance drops when we do that”

Latent Space: The AI Engineer Podcast · Lab-in-the-Loop at Compute Speed - Ian Quigley, Leash Bio

Play from 44:23
44:23

Audio streams from the publisher’s own feed. Verbatim does not host or copy it.

What was described

clustered building blocks by similarity and ensured no whole clusters overlapped between train and test splits to avoid memorization shortcuts

Said nearby on the same subject

“You can rent a cluster and someone else can take over that cluster, that's the whole point of the fungibility and financeability of compute”
♫
Said on Monetary Matters with Jack Farley
Why it worked Share this moment
“I built, like, a home cluster, GPU cluster. And actually, I originally started kind of serving the site from my home cluster at home.”
DT
Devin Thomas
This Week in Startups ·
How they work Share this moment
“You should do something like XGBoost… we see XGBoost drop steadily as you get further away with building-block cluster approaches”
♫
Said on Latent Space: The AI Engineer Podcast
Why it worked Share this moment
“We trained the final model on the SpaceX cluster ... a few weeks of pre-training, then synthetic data NRL and then shipping it”
ML
Misha Laskin Beam
No Priors: Artificial Intelligence | Technology | Startups ·
How they work Share this moment

Moments like this, every Monday

The strongest new moments on your topics, in the speakers' own words. Free for three topics.

No card. No password. One-click unsubscribe.

Track a competitor?

Pro watches up to 25 names or phrases and emails you the morning after they are said.

See Pro, $29 a month

Verbatim quotes a short extract and links to the episode; the episode belongs to its publisher. Remove a show.