“We can actually interpret where Hermes thinks the binding pocket is... it was looking at short sequences away from the functional part”

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

Play from 1:12:09
1:12:09

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

What was described

applied Hermes (a protein-language model) to generate cross-attention maps that reveal which amino-acid positions the model focuses on

Said nearby on the same subject

“The model can also generate videos… the model can also generate its own training data so that becomes a self-fulfilling flywheel”
♫
Said on Latent Space: The AI Engineer Podcast
Automation Share this moment
“We went back and did trade models with these additional mutations... now Hermes will look at all of these things much more evenly”
IQ
Ian Quigley Leash Bio
Latent Space: The AI Engineer Podcast ·
How they work Share this moment
“Epic ... banned handwritten code ... switching to an AI-first production model even down to the art which will be entirely generated with full AI workflows.”
♫
Said on The AI Daily Brief: Artificial Intelligence News and Analysis
How they work Share this moment
“There are… neurons that seem to track model confidence where you can even like steer and make the model more or less confident generically about whatever concept it's learning.”
EB
Eric Bigelow research team
"The Cognitive Revolution" ·
Why it worked 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.