“Collect some data, train a model, then deploy; a little bit of data (couple hours) is enough for many tasks”

Latent Space: The AI Engineer Podcast · Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy - Jason Ma, Dyna Robotics

Play from 30:25
30:25

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

What was described

collect a few hours of tele-operated robot data, train a model on it, then deploy the model to the robot in the real world

couple hours

Said nearby on the same subject

“Some of our people who collect data had no robotics experience they were working in retail ... they also became very well trained in robotics”
♫
Said on Latent Space: The AI Engineer Podcast
How they work Share this moment
“We have already deployed tens of robots… our goal next year is overall just deploying hundreds of robots”
♫
Said on Latent Space: The AI Engineer Podcast
Why it worked tens of robots deployed today → hundreds next year 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
“We then collected those corrections and then basically fed them back into the training data, retrained the models. They then got better.”
♫
Said on The Leadership in Insurance Podcast - Insurtech & Innovation
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.