Searching...
14 results for “public data”
Company10 mentions across episodes
Public
...Yeah. Data. But for public code, I think those are the most common. Yeah. I think some interesting things about those that I just personally find surprising,
22:30 / 23:36
from transcript22:30 – 23:36
...data. Was it the Apple paper? The Apple paper, the Kimi paper has mentioned this, a bunch of others.
36:48 / 37:49
from transcript36:48 – 37:49
...was the dataset. Mostly information that you might get access to from
6:42 / 7:47
from transcript6:42 – 7:47
...that are in the public domain. They are most private companies have to disclose their data to regulators any place you look. Typically, a number of other companies have looked there before and haven't yet found anything. But this data is even when it
3:42 / 4:55
from transcript3:42 – 4:55
...public after a period of time. And there's an enormous amount of data in maps of various kinds. One of my favorite datasets that we use around the world, a set of, of maps from Zambia from, almost a 100 years ago. The originals are hand painted on li
4:35 / 5:40
from transcript4:35 – 5:40
...data, that's free. But you find something that just, like, it wasn't worth that much before because, like, what do you do with flight data? What do you do with who is record data on the Internet? I actually talked to an entrepreneur recently. He was
38:00 / 39:23
from transcript38:00 – 39:23
...data than public data. I mean, you all work for Bloomberg, so I'm I you're probably intimately familiar with this. But if you think about the different AI labs that exist, they actually now do kind of have different data related modes like XAI.
25:30 / 26:40
from transcript25:30 – 26:40
...dataset that was actually good to train these models on. It's, of course, so much smaller than final web. It's only 30,000,000,000 tokens, but that's the scale that Microsoft datasets was.
8:55 / 10:03
from transcript8:55 – 10:03
...all the data. No. The the first step is like, okay. What are the cool data sources, and datasets people have put together for language model for training? Or when it comes to, like, our post training pipeline, we one of, the steps is, you wanna do so
6:41 / 7:59
from transcript6:41 – 7:59
...Because the dataset that they trained on was not public, so people were not really sure if these models are really good or maybe there's just some data contamination. So it was really hard to check if you just have the weights of the models. And, at
6:40 / 7:49
from transcript6:40 – 7:49
...data, to do a replication of the bit net paper in the open. So they took just a really, like, the initial part of a pipeline, and then did the thing on top of it. It goes both ways. So for example, for the two model, a lot of our pretrained data for
16:23 / 17:34
from transcript16:23 – 17:34
...we can is the natural starting point. Is there something to be said, about the topic of data quality? Is there some low hanging fruit there still where the quality could be improved? It's like turning the crank. So I think historically in the open, t
1:24:40 / 1:25:46
from transcript1:24:40 – 1:25:46
...flight data? What do you do with who is record data on the Internet?
37:26 / 38:41
from transcript37:26 – 38:41
Have a podcast?
Run your podcast through Clypt
Get ranked clips, hooks, and ready-to-post copy from your own episodes. Free to try.