Software Snack Bites cover image

#12 - Phillip Carter (Principal PM @ Honeycomb) - All The Hard Stuff when Building Products with LLMs, Actual Results from Leveraging AI

Software Snack Bites

CHAPTER

How to Use Few Shot Prompting to Count Tokens

How does that token counting actually even work? Like, can you go to open AI's documentation or something? And they have everything listed out? Are you writing some sort of script where you're just running the prompt output and then saying, okay, this is what we would estimate? I'm just curious how do you even get to the point of like token counting? So they use a pretty standard by parent coding algorithm that they do actually make available through open source, but only through Python. We then have a port of that and go because our backend is written go and produces the same results as what their thing is. However, we do the counting as we go rather than counting the whole

00:00
Transcript
Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner