High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / uncertainty

Published · transcript-backed

Shawn Wang: uncertainty

6 Sept 2023 Latent Space The Point of LangChain — with Harrison Chase of LangChain

“In terms of like, for you, it's one of the, you know, the model IO modules and that's it. But like, you seem very personally, very passionate about it, but I don't know what the Langchain specific angle for this is, for fine-tuning local models, basically.”

— Shawn Wang

Source trail

Everything needed to verify it.

Speaker
Shawn Wang
Attribution
Verified speaker
Claim type
uncertainty
Recorded
6 Sept 2023
Publisher
Latent Space

Transcript context

…No, thanks for it. I'm really excited about that stuff. I mean, that's something that like we've been, you know, because like, as I said, like the only thing we know is that the space is moving so fast and changing so rapidly. And like, local models are, have always been one of those things that people have been bullish on. And it seems like it's getting closer and closer to kind of like being viable. So I'm excited to see what we can do with some fine-tuning. Yeah. I have to confess, I did not know that you cared. It's not like a judgment on Langchain. I was just like, you know, you write an adapter for it and you're done, right? Like how much further does it go for Langchain? In terms of like, for you, it's one of the, you know, the model IO modules and that's it. But like, you seem very personally, very passionate about it, but I don't know what the Langchain specific angle for this is, for fine-tuning local models, basically. Like you're just passionate about local models and privacy and all that, right? And open source. Well, I think there's a few different things. Like one, like, you know, if we think about what it takes to build a really reliable, like context-aware reasoning application, there's probably a bunch of different nodes that are doing a bunch of different things. And I think it is like a really complex system. And so if you're relying on open AI for every part of that, like, I think that starts to get really expensive. Also like, probably just like not good to have that much reliability on any one thing. And so I do think that like, I'm hoping that for like, you know, specific parts at the end, you can like fine-tune a model and kind of have a more specific thing for a specific task. Also, to be clear, like, I think like, I also, at the same time, I think open AI is by far the easiest way to get started. And if I was building anything, I would absolutely start with open AI. So.…

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

Search evidence