The a16z Show · Can Open Models Keep Up With the AI Frontier? | Nathan Lambert
Audio streams from the publisher’s own feed. Verbatim does not host or copy it.
treat prompts as the key unit of post-training, using them to generate completions for further fine-tuning
“…prompt the unit tests to see and push the limits… then I went to Claude Code… built Claude Code skills that for me are rinse and repeat off of the unit test prompts…”
“An agent identity needs to be treated as a first-class identity… it needs to have least privilege… just-in-time access… a complete record of everything privileged.”
“As you get further… smaller, as you get closer… larger… when you find your person, you collect them into your crystal”
“We asked inkling to fine tune itself using tinker. The model wrote its own fine tuning job, ran it and evaluated the result.”
The strongest new moments on your topics, in the speakers' own words. Free for three topics.
No card. No password. One-click unsubscribe.
Pro watches up to 25 names or phrases and emails you the morning after they are said.
See Pro, $29 a monthVerbatim quotes a short extract and links to the episode; the episode belongs to its publisher. Remove a show.