Latent Space / episode intelligence
Speakers in the public record
Claim mix
Evidence policy
Every row below preserves an exact excerpt. Identified speakers are linked; unresolved voices are labeled and excluded from people counts.
Claim ledger
28 published records
01 / prediction
“Yeah, so I think that's a very, very good question. So I don't think we are running out of data on Earth.”
02 / evaluation
“Yeah, 30 trillion tokens. So I think what's exciting about Red Pajama V2 is not only the number of tokens, but we start to kind of learn from Red Pajama V1.”
03 / observation
“So, but the one problem of Lama is the data recipe is being described in a pretty detailed way in the paper, but the data is actually not there. So, and our original thinking is how about we take the recipe and we try to do our best effort reproduction and try to put it out, such that we can learn from our mistakes in the reproduction together, right?”
04 / commitment
“I will have to read the paper to understand a little bit more. Because when you say things like, we have to know in advance what we were trying to do with the model, then we do importance resampling.”
05 / belief
“If you go back to the original Google paper of federated learning, I think that's very different from what people are talking about today when they say federated.”
06 / belief
“I would say right now, the top five models on our inference stack are probably all fine-tuned versions of open models.”
07 / belief
“I think I finally get the name of the company, like- Bring it together, yeah. Everything needs to be automated together.”
08 / belief
“You know, we are pushing to do that on training. And that is, you know, we think, if there was a sort of, you know, developer experience message, that's probably the big one is where you have enough flexibility.”
09 / belief
“I think maybe a year ago, people were like, oh, yeah, you can use this GPU computer that is going to be end-of-life.”
10 / belief
“We also have, I would say, over 20 big model builds happening on the platform, which are customer.”
11 / commitment
“You know, the value we get from doing specific optimization, even for, you know, what works well for a particular model on A100s with a particular bus versus H100s, it's a worthwhile investment for us. So we will go down fairly deep into a specific architecture and specific hardware.”
12 / belief
“Like this is a concept that I hadn't heard before reading about this. So I think most people's mental models, like transformers or something else, it’s not transformers AND something else.”
13 / uncertainty
“I just noticed that you saw 8002 was used to be at the top of the MTB chart, and then it's just like sliding down and down and down, and all the new models are coming out of China for some reason. And I'm like, I don't know what's going on there.”
14 / belief
“If you go to AWS, you do know which region you are in, right? So I think one thing that we are trying to do is you have this disaggregated cloud, not only about location or geographically where they are, but about this reliability and also this diversity of this infrastructure.”
15 / prediction
“I would say the market is very tight still, and it's likely going to be this way for a while, is my sense that the demand for AI computing is just kind of ramped up very, very quickly, and it will take a while for supply to catch up.”
16 / belief
“I think on the inference stack, there are open source inference stacks which are pretty good and definitely today, it gives us a competitive advantage to have the best one.”
17 / belief
“I was surprised to see, I think Jina or, yeah, there's Jina AI, and then there's another guy, Tengyu's Voyage.”
18 / belief
“I think you need to have some kind of a marketplace for figuring out how to get this, you know, data into models and have, I think we'll increasingly see more of that.”
19 / evaluation
“It's just very expensive. So I think what those like state-based models trying to do is try to push the size of the state, right?”
20 / prediction
“I think that's a pretty important question because that really gives us a roadmap of where this industry should go.”
21 / evaluation
“I think people come on your website today and they say, you raised a hundred million dollars Series A.”
22 / preference
“Okay, well, I mean, it looks like it's paying off, so. And then high level, I will confess or admit or mention for the listeners who are also similarly skeptical, I did not used to care about long contexts because I was like, you know, 30K is enough, 100K is enough, right?”
23 / prediction
“So, and how can you make the whole thing really, really fast? So I think for the next couple years, yeah, we will see a whole bunch of new embeddings maybe of different size and much, much faster than today.”
24 / evaluation
“I would say there is a lot more understanding around fine tuning now, like even the last six months, there are, you know, source tools, recipes, literature, podcasts, discord channels where people are figuring out and it really is in many ways, one of the successes of open source is you have small collectives of, you know, engineers who have created, who are now creating the top models on open source leaderboards.”
25 / evaluation
“I know we kind of binned the lightning round in the last few episodes, but I think for you two, one of the questions we used to ask is like, what's the most interesting unsolved question in AI?”
26 / evaluation
“You know, and from there, finding, you know, other folks in the network, I think there is generally a lot of excitement and philosophical alignment around what we are doing, which, you know, we publish papers, we publish open source libraries and code, we build open models.”
27 / evaluation
“I don't know, but I wouldn't take that for granted that they should be the same, right? So that's one of the hypothesis that, so we have no opinion on that because I think that's the result of the study, not the assumption.”
28 / commitment
“You know, we'll be sort of building this best-of-class hardware. So as there are other versions of these coming out later this year, we plan to have those in the fleet as well.”