“We went back and did trade models with these additional mutations... now Hermes will look at all of these things much more evenly”

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

Play from 1:13:40
1:13:40

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

What was described

used Hermes attention maps to spot when the model was memorizing short amino-acid nicknames, then added targeted pocket mutations to force the model to attend to the functional region

Said nearby on the same subject

“We can actually interpret where Hermes thinks the binding pocket is... it was looking at short sequences away from the functional part”
IQ
Ian Quigley Leash Bio
Latent Space: The AI Engineer Podcast ·
“We buy a known binder, run the screen again… all the chemicals that normally go in the pocket will be displaced… then we mutate the pocket and see those series drop out.”
IQ
Ian Quigley
Latent Space: The AI Engineer Podcast ·
How they work Share this moment
“We use a head coach model… regional vice presidents work with head coaches, branch coaches work with physician coaches.”
JF
Jason Fox
AI & Marketing for Home Service Pros ·
Business model Share this moment
“When we have an open model there, and people can test it out first, combine them to do application good for their targeted business, and later they can build their own model.”
ML
Ming-Yu Liu Nvidia
Practical AI ·
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.