The Exchanges

Every argument clarity score on this site is built from rows on this page. Each question and answer was assessed with names hidden, the host's own answers included, on four things from 1 to 5: directness (does it answer the question asked), coherence (do the ideas follow), precision (concrete details and clear references), compression (says a lot per word). The weighted mix (30/30/25/15) is the exchange score. A person's published score averages their exchange scores on raw tape only, at least 8 of them, shrunk toward the cohort mean. Full method →

Zach Nussbaum no published score: only 1 usable exchange on raw tape, and a fair score needs 8+ record → ← everyone

Every exchange below was scored with names hidden, four dimensions each from 1 to 5. An exchange's score is 0.30·directness + 0.30·coherence + 0.25·precision + 0.15·compression. The published score averages the raw tape exchange scores and shrinks small samples toward the cohort mean, so five great answers can't beat twenty good ones. Produced feed rows count only toward coarse estimates, never toward a full score.

clear all ✕
1exchanges match
1on raw tape
0redirected or not addressed
Answered raw tape D 4 · C 5 · P 4 · Cm 4 4.30

Q So maybe, uh, you mentioned, um, OpenAI Ada. Maybe, like, paint a quick, uh, picture of the space. Like, what are the other, uh, you know, embedding models that, uh, either have existed for a while or are just coming up?

A Yeah, um, so, uh, roughly in the space, uh, there, in the open source side of things, um, there's a bunch of really great models, uh, that have been released, uh, the model weights have been released, and A few of them are being hosted, uh, on, you know, embedding endpoints, um, but the problem with a lot of these models is that they have a small context length, so they're limited to 512 tokens, um, which I think, you know, nets out to maybe a few paragraphs, um, which becomes a problem when you have large, uh, large documents. Maybe you have, like, long financial documents and you want to reason over the whole document versus chunks, um, just because it gets a little confusing and it's, Uh, a little hard to, um, reason about, like, the right way to chunk up a large document if you need to, like, recall something from the first paragraph and the last paragraph. Um, so that's the, the open source side of things. Um, there's, uh, one or two long context models. Um, uh, the ones that are, are good are either too big, they're in, like, the seven billion parameter range, um, or they, uh, don't actually beat OpenAI, uh, Ada's model. Um, and then, you know, for the long contacts on the closed source, uh, uh, ADA is, you know, kind of the de facto for many people's applications today. Um, but, you know, with, with closed source, the, the big problem is, you know, you, you have no idea …

AI assessment note: “in the open source side of things, um, there's a bunch of really great models”

page 1
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.