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. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.