Evidence receipt / belief
Published · transcript-backedWilliam Beauchamp: belief
26 Jan 2025 Latent Space Outlasting Noam Shazeer, crowdsourcing Chai AI with >1.4m DAU, and becoming the "Western DeepSeek" — with William Beauchamp, Chai Research
“Lots of cool stuff. So really quickly, I think inference is very, very important.”
Source trail
Everything needed to verify it.
- Speaker
- William Beauchamp
- Attribution
- Verified speaker
- Claim type
- belief
- Recorded
- 26 Jan 2025
- Publisher
- Latent Space
Transcript context
…Yeah, it's exactly, exactly. Yeah. I just realized we skipped one important part. So you spent $10 million on compute last year. You say you're going to probably triple that. Yeah. I'm sure you're doing a lot of work on custom kernels, kind of like inference optimization, any cool stuff. Yeah. That you want to share there. Yeah. Lots of cool stuff. So really quickly, I think inference is very, very important. It's super important. It's massively underlooked and we can look at all the different foundation models and the techniques, the differences in the foundation models on how well they perform from a cost perspective with inference. Mixture of experts, for example, tend to do really, really good from like a cost perspective. We've worked with a very talented team called. MK1 and we, so I saw, I saw them in the Chaiverse logs. What are they?…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.