Tools

Search

Explore

Videos Channels Figures

Atmrix

About

Tools

Search

Explore

Videos Channels Figures

Atmrix

About

JD

Jeff Dean

08/08/17

@ Y Combinator

We aim to reduce the time it takes to train models, allowing engineers to conduct multiple experiments simultaneously and improve workflow efficiency.

Video

YC

Jeff Dean’s Lecture for YC AI

@ Y Combinator

08/08/17

Related Takeaways

JD

Jeff Dean

08/08/17

@ Y Combinator

The goal is to make the time to result as short as possible, transforming how engineers approach machine learning experiments.

MK

Mike Krieger

05/09/25

@ Sequoia Capital

Our goal is to enable models to work autonomously for longer periods, which will require advancements in memory, tool use, and organizational onboarding.

LF

Lex Fridman Cast

02/03/25

@ Lex Fridman

If you reduce the time from when your model is done training, you want to get it out as soon as possible to maximize the perceived quality of your outputs. DMC does this so well.

DP

Dwarkesh Patel Cast

05/23/25

@ Dwarkesh Patel

We often underutilize models, expecting them to solve tasks quickly without giving them enough time to iterate and improve.

LQ

Lin Qiao

05/19/25

@ Sequoia Capital

We have developed a self-serve platform that allows developers to easily tune AI applications for speed and quality, facilitating rapid scaling and experimentation.

VM

Varun Mohan

05/02/25

@ Y Combinator

We quickly improved our training infrastructure and trained our own models, which allowed us to pull ahead in terms of quality and capabilities compared to GitHub Copilot.

O

OpenAI Cast

04/11/25

@ OpenAI

The effort to parallelize work aggressively during the run was crucial to avoid performance cliffs that could make the model untrainable.

JD

Jeff Dean

08/08/17

@ Y Combinator

Currently, we train bespoke models for each problem, but the goal is to develop models that can generate solutions for multiple problems, enhancing architecture search.

DR

Dan Roberts

05/09/25

@ Sequoia Capital

OpenAI's model improved not only with training time compute but also with test time compute, indicating that the model benefits from additional reasoning time during testing.