Searching...
8 results for “ai discovery”
You just And then you build up some intuition like this this mixture of expert works, and then this implementation of MLA works. Key hyperparameters like learning rate and regularization and things like this, and you find the regime that works for yo
56:55 / 58:01
from transcript56:55 – 58:01
And the final stage is much newer and will link to what is done in r one and these reasoning models is, I think, OpenAI's name for this. They had this new API in the fall, which they called the reinforcement fine tuning API. This is the idea that you
27:37 / 28:47
from transcript27:37 – 28:47
I'd say the techniques for, like, vanilla SAEs are pretty well established now where you take your model that you're interpreting, run a whole bunch of data through it, gather activations, and then, yeah, pretty straightforward pipeline to train an S
25:31 / 26:45
from transcript25:31 – 26:45
whenever there is a proposed solution for solving the problem, you're able to tell how good this solution is. So you basically define what makes a good solution. For discovering an algorithm for scheduling jobs on a data center, this evaluation funct
11:59 / 13:08
from transcript11:59 – 13:08
and then AlphaEvolve fills in the how. So that's the job of, of our system and you can do it in two fairly different ways. One is you'll tell AlphaEvolve, I have no idea how to solve this problem. Let's start it completely from scratch and let's try
12:48 / 13:51
from transcript12:48 – 13:51
maybe you could just think like one level beyond steering as as it is today. Like, imagine if you could adapt your model to be, you know, an expert legal reasoner, like in almost real time, like very efficiently using human feedback or using, like, y
33:06 / 34:08
from transcript33:06 – 34:08
so I've done a bunch of experiments. I've done hyperparameter tuning. I've done all the things I'm very used to and I've done for two decades. Yeah. And I've gotten to a certain point, and I thought it was, like, fairly well tuned. And then I let aut
18:35 / 19:54
from transcript18:35 – 19:54
method, but it's something like you just sort of, like, look at the data point gradient and see if it points in the direction and weight space of the fine tune, and then you take, like, the top data set. There are some tricks to it, but it's basicall
1:04:30 / 1:05:39
from transcript1:04:30 – 1:05:39
Have a podcast?
Run your podcast through Clypt
Get ranked clips, hooks, and ready-to-post copy from your own episodes. Free to try.