GPT Index and Weaviate with Jerry Liu and Bob van Luijt - Weaviate Podcast #37
Weaviate Podcast
The Cost and Latency of Defining Complex Graph Structures
Every call to GPT is gonna take around like, two to three seconds of you just give it like, 3,000, 4,000 tokens. So there are just some practical constraints in making this graph overly complicated. For a lot of tree structures, practically speaking, what has worked well are just slightly more shallow structures. But I think generally speaking, you know, as these models get better and cost and lanes does come down, this notion of repeatedly querying or recursively querying the L M for more information has already proven to be pretty useful in a lot of cases.
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.
