Searching...
16 results for “local ai models”
“Why founders are ditching ChatGPT for local AI models”
...you have models of different sizes and different memory and hardware requirements. The reason people are running things locally is because they keep running out of Claude tokens or or OpenAI tokens.
15:47 / 16:26
market insight15:47 – 16:26
...the models in question here, we're not talking about just, you know, western closed source models, but also a a deluge, if you will, of of really high quality open source models from China. There's Mistral over in France. Where else are we seeing, ne
31:28 / 32:30
from transcript31:28 – 32:30
...models, which doesn't really have a, like, a Chinese company that is has an analog. While I'm talking, I'll say that the Chinese open language models tend to be much bigger, and that gives them this higher peak performance as MOEs where a lot of thes
44:13 / 45:26
from transcript44:13 – 45:26
...models are available throughout our API, except for administer seven three b. But the for the premium model, they have a special license, menstrual, research license. You can use it for free for exploration, but if you want to use it for enterprise,
35:15 / 36:20
from transcript35:15 – 36:20
...models and the growing post training and mid training stack with hot takes on everything from constitutional AI to DPO to rejection sampling, and also previewed the sea change coming to the Allen Institute and to InterConnex, his incredible substack
0:49 / 2:04
from transcript0:49 – 2:04
...own language models with their own data to just not educate other LLMs and to keep their data private. So I think the future is gonna be, you know, a lot of people running their own deep seek instance, you know, over at a Google Cloud or Azure, etcet
8:15 / 9:28
from transcript8:15 – 9:28
...clearly care a great deal about, open source models are pretty good on math instruction following and adversarial robustness. The llama model is amongst the top three of of evaluated models. Included the agenting tool use here just to point out that
5:31 / 6:32
from transcript5:31 – 6:32
...models. And essentially, what you want is for our AI models that centers on human speaking, we want them to see lots of different scenarios, lots of different types of people interacting in many, many different ways. And that requires lots of trainin
47:15 / 48:21
from transcript47:15 – 48:21
...models, which doesn't really have a, like, a Chinese company that is has an analog. While I'm talking, I'll say that the Chinese open language models tend to be much bigger, and that gives them this higher peak performance as MOEs where a lot of thes
44:58 / 46:11
from transcript44:58 – 46:11
...They do. Yeah. So they're in the talking avatars category for the model layer. So their model is like a talking avatar model. And then they also host a bunch of the image and video models so that you can generate other stuff on their platform too. Fl
26:30 / 28:06
from transcript26:30 – 28:06
...models. Models that are only they're only trained on licensed data.
26:20 / 27:25
from transcript26:20 – 27:25
...are now API available or open source. If you think of some of those properties too, some of them have built really deep and interesting workflows around creating content even if they don't have their own model. Yeah. Some of them are also starting to
14:15 / 15:23
from transcript14:15 – 15:23
...language models, AI running locally? Obviously, you you brought up this incredible example of, ASML wanting to not have, you know, all their important, you know, information and innovations on somebody else's LLM, you know, that that's gonna eventual
40:27 / 41:31
from transcript40:27 – 41:31
...source models are from China these days. China has made a really big push on open source. Obviously, DeepSeek is an open source Chinese model. That was the first big one. Kimmy is one. Quinn from Alibaba. And so I think that if you want The US to win
49:26 / 50:33
from transcript49:26 – 50:33
...models distill, like, this will end up in an oligopoly. But I I mean, I don't know. That's just my guess. To what extent do you think the large model providers in ten years' time have already been created, or are they yet to be founded? I think that
8:16 / 9:18
from transcript8:16 – 9:18
...server. What local models can I run, let's say, on my MacBook Pro here, to use in reply to handle the offline processing? Is it just stuff from Meta, or have you guys opened the aperture to also, you know, models from moonshot and so forth? So we're
8:09 / 9:37
from transcript8:09 – 9:37
Have a podcast?
Run your podcast through Clypt
Get ranked clips, hooks, and ready-to-post copy from your own episodes. Free to try.