audio
audioduration (s)
3
27.3
transcription
stringlengths
8
259
and to make everything deterministic we're going to use a generator object in
extracting the noise so what to extract is the noise uh predicted noise to
that it's going to take in two parameters the first is going to be text
the fourth third second and first so it's only so information only flows from
need the collection we don't need to add something just create create it in this
This code that we are going to write will be available in the notes and I will walk us through it
in the same directory as your wav files right this is important create the file
then inspect this list we're going to see that it ordered our
vicinity of what we expect to achieve but before we achieved it by counting
to why we'd want to remotely control objects. This cute little robot, for instance, has a remote API.
celta w and x and in python zip takes two iterators
And it will show you this as well. As we said, the responses come back
my name is Christine how are you doing today we're gonna hit enter uh let it do
and let's call it the capital n because it's going to represent sort of the
we are interested in this row we've created this
But typically this is what you do is you write a server that then the server keeps track
that's it and then we have b grad is again dl by
I want to define the create LM create LM chain function then create a memory or
We wanna look at all the numbers that were was sent to, right?
is i would like to take you through building of micrograd now micrograd is
basically a clip model which is fine tuned on text image pairs this model is
to this Discover section and kind of get inspired by what's available, right?
I don't think I am. Oh, oh no. The images with bicycles. That one's got a bike.
ones way begins as all zero okay so if I just print way in the
So we'll be sending a text messages using that trial. Dollar dollar bills, y'all. So, here we go.
begin uh with zero and you can think of this as like an interaction strength or
external external Text data
and because we're squaring uh there will be no signs anymore um
fine-tuning stage is to actually align it to be an assistant and uh this is the
so let me actually come back we use the decay learning rate of 0.01
told you already it will just take 20 or 30 minutes only so it is really um
pd.pdf document from the PDF bytes as PDF file
these capital letters and these uh like punctuations we will just import like
So this is the to here, is what we want. Now this 2B here just happens to be a percent
pass in the template that we just created
piece by piece by following the loss and that's why we also obtain the same
these characters like this uh given some context of characters in the past the
requirements and you can see we're using streamlit streamlit will be our front
uploader here let's make St sidebar because we want to have it at the
a and this is how much of every row we want to average up and it's going to be
you'll see that there's these different headings. We've got this Discover, I think this is great
I'm gonna put this back so that we have things working. So this was Auth token
a bit more complex and uh useful example we are going to back propagate through a
means for us is there's a very simple recipe for deriving what we want which
we're going to copy paste this code that we previously used for counting
of elements from 0 to 17 and the shape of this
our data set into a train and a validation split so in particular we're
which I hope that you have installed. And so I'm gonna do code dot, and if this doesn't work
concluded that the actual TTS package from Coco has been installed properly
GPT okay okay so now I wanted to do a very brief walkthrough of nanog GPT
ignored because um we're not we're not using that and we have no targets that
and then of course i actually invite you to read this paper it is 19 pages but at
on that string to get all of our words as a python list
will be zero and then when all the probabilities are lower this will grow
optimization object so here we are using the optimizer ATM W um now in a make
dimensional as it as a standard and then there's going to be six layers of that
That way you won't have to come back and ask again." Or, come back and bother me, is what he really means.
amazing but at least you see that we can see like is here this now grief syn like
I need you to add to my capabilities. And these are the things I need." The administrator gives him those and then also says, "You know what?
There we go. And then I'm gonna call this gallery. So again, I have an array.
the password to my luggage 1, 2, 3, 4, 5, 6. So that is not my account SID that is going in there.
and that's going to tell us the sl the the sign of the derivative
That is, we won't have any single point of control. In fact, what we're trying to do is force collusion by two bad actors-- or more than two bad actors --in order to compromise the system.
previous videos on the same YouTube channel in particular my make more
so fundamentally what's happening here again is
example we have we had 130 ,000 if you want to uh use more uh sorry but we just
so this is way too much code so actually there's a way to simplify this which is
replace these special characters special the Turkish characters how to do so we
large number of character level language models in terms of the neural networks
So you will not be able to have it. So I've got this token, it's filled out here.
training split then we're going to look at train data otherwise at valid data
x uh up to T and including T and the target is always the teth character but
using dot node and then we also create edges using dot
So one of the things, one of the photos that I want to do, I love to get photos of burritos from around town.
access these probabilities not just listing them out in a tuple
this list method didn't actually finish working. So one of the things you always wanna make sure that you do
or a send button is sent therefore we need to encapsulate it here only if the
And as long as we match the same format in what we return, that is, as long as our API maintains the same contract,
0 is dot so we've reversed that here
do log 10 i believe that's a function of the loss
these tensors but none of the math changes this is done purely for
possible vocabulary elements we can actually guess at what the loss should
value node so we actually want its data um
broadcasting the way it works this vector of 27 suddenly becomes a row
here in C well what is this number well this is the first row times the First
we basically want to do p p p divide p that sum
character of the bigram and the columns are going to be the second character and
that I am actually going to put this thing right here inside the STD
useful to be dealing with tensors from scratch uh and i think and that's why i
and you know that Beyonce's is, the songs are gonna come back very high in the danceability thing there.
in a bit but first let me just show you what this code does
understands how the Play button works. She coded the button so that when it gets clicked,
instead of one slash it's gonna be two slashes now it wants an output folder so
what we'd like is a loss function and a loss function has the semantics that low
and now we're going to recurse our way backwards again
backward on and so pytorch can be a lot more efficient with its memory use
Now, quick word of warning, if JavaScript is new to you, there is some stuff here that might be a little confusing.
0.5 0 negative 1.5 and 1. so if we just run this
is going to be a little complicated now we want to return line break join and
that we're about ready to have to do. You can do things a little bit more cleanly.