“We're going to build an open source stack for SFT for mid-training ... we can store the trajectories ... and you just do this for a year of a lab and then check in every six months”

The a16z Show · Can Open Models Keep Up With the AI Frontier? | Nathan Lambert

Play from 40:12
40:12

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

What was described

Built an open-source stack for SFT mid-training, using SkyRL and other tools, storing model trajectories, task configs and evaluation results, and publishing a snapshot every six months

Said nearby on the same subject

“Claude downloaded every single model, every single tool we use, FFmpeg, all the Python scripts that it's using to do all the editing”
♫
Said on YouTube Creators Hub
Automation Share this moment
“We built the team, imagined stores… 18 months later we opened our first store.”
♫
Said on B2B Go-To-Market Leaders
How they work Share this moment
“A really good gut check is every two years to re-evaluate the systems and tools. I don't think you should do it every year.”
MS
Mandy Savoca
In-House Outliers ·
Why it worked Share this moment
“We asked inkling to fine tune itself using tinker. The model wrote its own fine tuning job, ran it and evaluated the result.”
♫
Said on Everyday AI Podcast - An AI and ChatGPT Podcast
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.