Eduard (Edward) Dulharu

speaker

420 appearances 1 recordings 1 series first heard Aug 2026 last heard 7 Aug

Eduard (Edward) Dulharu’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Aug 2026 with 1.

Appearances

newest first · ▶ plays the moment

Heavy Networking · HN837: Agentic AI to Reduce MTTR · 7 Aug 2026 podcast

So it's a cosine function between your question represented in vectors and with the vectors which are stored within this database.

And the idea is to catch up the um to retrieve from this database.

Only those vectors which ha which are uh which have basically cosine of zero is one, closer to one, which means that they are semantically similar.

And I didn't know trigonometry was coming into this conversation.

It is, it is.

Okay.

And um one limitation of Rog, uh just for the sake of uh well.

Is that you can afford to extract a certain number of chunks from the database.

Uh-huh.

So you have a uh parameter called k and um which has a finite, it's a finite number, and you can extract the first k matches with your uh which have uh semantic similarity with your question, which means that if your uh

real answer or the best answer is beyond with K, you are missing the answer.

So this is one limitation of rug.

There are various uh rug flavors which somehow mitigate uh this cake problem.

Um but Rug, as I mentioned, uh is a application for this uh embedding.

Fine tuning is uh a different game.

Yeah.

Not Bucalium is an excellent example actually.

Yeah.

It's uh it's a Google product, so it's not something uh some bu someone can replicate uh in uh in a lab.

Yeah, yeah.

Showing 221–240 of 420 · page 12 of 21