Shawn Wang: uncertainty

Transcript context

…Yeah. Okay. Yeah. I mean, it's a small ops team. We employ a lot of fractional ops people for various reasons, but a lot of it is you can pay someone three to seven dollars an hour to look at generations and understand what went wrong. Yeah. Got it. And then at a high level for eval, I assume you build most of this yourself. Did you look at what's out there? I don't know what is in the comparison set for you, like human, you know, like, or whatever scale has skill spellbook. Yeah. Or did you just like, you just not bother evaluating things from other companies or other vendors? Yeah, I think we definitely, I don't know, necessarily want to call out the specific vendors. But yeah, we, we have used for different things. We use different products and then some of this has to be run on like Google Sheets. Yeah. We do a lot of our model graded evaluation in the workflow itself, so we don't necessarily need something like, you know, open layer. We have worked with some of the platforms where you can, gives you a nice interface for evals as well.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.