Transcripts
1. Introduction: What if I told you
you can harness the power of artificial
intelligence, specifically Deepsek
locally on your computer or your device offline without
the need to go online. Literally, you have
the ability to take artificial intelligence
applications with you wherever you go, on your computer, on your
laptop, and not just that. You have the ability to
maintain the data privacy without sending
information across servers between
different locations. Everything is going to be
contained within your device. In other words, you got your own personal artificial intelligence
assistant on the go. In this current class,
I'll be teaching you how to actually
localize Deep Seek install it on your computer
and harness the power of artificial intelligence
locally within your device, maintaining your
own ecosystem of data without compromising
data sharing, data leaks, transfer of
data between servers. Everything will be kept within your device for your own
purposes and applications, at the same time,
harnessing the power of artificial
intelligence locally.
2. Your Project: Your project for the
class revolves around the localized implementation
of Deep Seek in which you are going to follow the
lessons taught in order to incorporate Deep Seek
locally within your device, after which you
are going to have a demonstration on a basic
prompt based on which you are going to confirm
your understanding in terms of the
process that you have followed to localize
the usage of Deepsik, after which you're
going to be sharing your output with the rest of
the community for feedback.
3. Setting Up Deepseek Locally: Now I'm going to
show you how to use Deepsek locally
on your computer. Yes, locally, because
sometimes when you're using the artificial
intelligence model, your information
that you're sharing, the commands are being accessed through servers
across the globe. Now, in order to ensure that you're using this the
way that you'd like to use it in the most
convenient way possible, without even the need to actually try to connect
to these servers is simply by having Deepsk get installed on your local device, literally on your computer, your desktop, it will be stored like a software
that you can use it where you have the language
learning model accessible without the whole privacy
policy considerations and the data security issues
that might arise with the usage of artificial
intelligence model since you dare to be
transferred between servers. So how do you go about this? There's an easy way
to go about this. I'm going to show you a couple
steps you need to follow as simply navigate to
the following website. Just click in momstudio.ai. You click on this. Now, this is the homepage for LMStudio. Now you might have heard
about Lama and LM Studio. Both of them, they
get the job done. You should find such
a similar logo, and you can just simply
type on Google or the actual search bar mtudio AI, and you would be prompt
to the following page. On the following page,
you need to download the select the option to download it to the actual
computer that you're using. You have for Mac. You got the Microsoft computer
desktop Windows. It's up to you. Just
simply select any one of them and then download it.
That's step number one.
4. Deepseek localized Configuration: Come back. So once
you've downloaded the LM Studio and you have activated it
onto your computer, once you click on it,
this is basically the window that you're going
to be seeing LM studio. Simply think about it this way. You download a model, you select a model, then you have the ability to use the model on your device. So how does that work?
Click simply over here on the search icon. Click on it. This is where you discover various models which
are being used. You can just simply download them on your computer.
Type in DeepC. Okay. Now, as you notice, Deep Sik has multiple options. These are the distilled models. Think about them this way. These are the lightweight
models which will get the job done on your computer without
being quite too heavy. Now, as you scroll down, you might find that you have different options for deep Sk. Some of them are really heavy, like it says here, the
file is too large. If you take a look at
the option over here, 347.45 GB the parameters. This is like the full deep
seek on your computer. So it's going to be
demanding on your machine. Or basic applications, which
will get the job done, by the way, these are
the go to options. They will help you
get the job done because you don't need
all of the parameters of Deep seeks like
having the full on artificial intelligence
model on your computer. You're not going
to be using Ovid. Select either one of them,
the quen or dilemma. Let's go for d ilamaF
example, over here. You'll notice that
this is 4.32 GBs, and this is about 4.68. Now, here it says 7
billion parameters. Here it says 8
billion parameters. The previous option that
we have went through, let's take a look
at the options. Well, the file is too large. If you take a look at
it, we scroll down. We got 47 GBs, the amount that we need to actually download
on the computer, the storage required
because this will be about the full on parameters for the number of parameters
for Deep Sik as a whole. So the distilled model,
8 billion parameters, 4.3 GBs, or seven GBs, 4.68 GBs to download it. Both of them will
work perfectly fine. Just simply click on
Download. Here we go. And you are going
to be downloading this onto your computer locally. Clearly, you are going to get the model for Deepsk the
LLM, the language model, and just simply download it on your computer or your desktop, whatever the device that
you're dealing with, but preferably something strong. You have a computer or
desktop that you have the ability to
actually handle this. But Lemma Lama or Quin, the options that
you have are most probably downloadable on
your computer directly if you have an average
PC or something which is for consumer usage,
that's perfectly fine. You're able to get the job done, and the whole purpose
of this is you have the ability to download the
artificial intelligence using the full on contextual length of 128 K. You have the ability
to access the R one model. You have the ability to
use it on your computer without having your data transmitted over
servers and offline, by the way. This
is very powerful. So if you'd like to use
some sort of documentation, you have a certain
tool on your computer, certain sheets that
you would like to use, and you would like
to have the help of artificial intelligence within
those sheets, for example, tool analysis can
just simply use the LLM model without
accessing the network. That way, you're able to contain the data and keep it
localized on your computer. So we have about 1
minute left in order to download Deepsek
onto the computer, and then we are going to
actually get to use it. You're able to track the
downloading process over here. We're about 60% downloaded. Now you might be
wondering why not to use it directly on the server? Sometimes, I've
noticed when I'm using Deep Seek on the actual server, like the actual Deepsek website, sometimes the server is busy. Literally, there are a
lot of traffic going through the website,
since it's quite free. And for basic
tasks, for example, why would you go on the server and just wait till the server is clear on the website and
it might take you some time. At the same time, you
might face the issue of privacy and
policy constraints. As apparently it seems that the data that you're
uploading on Deepsek, there are certain
data security issues that many users are
kind of highlighting. So in order to avoid
the whole hassle in the first place and still use Deep seek for your
own applications without the need to go online, without compromising data
within your organization, for example, you can
just simply download it. Either you got the first
option or the second option. And if your company,
for example, provides really powerful or
you got really powerful PC, you have the ability to
select either one of them, the other options,
and then you're able to see if it's
downloadable or not. Now we have reached almost 100% on the download. So it
seems, here you go. Download completed.
The model is ready. You can just simply
click on Load Model. Here we go. So these are
the downloaded models. How would you go about this?
Take a look at the top part. Right? Once you click on the
drop down menu, you notice, what are the models that I have available as part
of the LM Studio. By the way, you could
install other models. I doesn't have just
simply to be Deep Sk. Take a look at other models. Feel free to try them
out, test them out. But we're trying to
help you save time, save effort loaded
onto your computer. You can eject it as well. So think about LM Studio, like a hub, you
connect to it a model. You can eject the model or
just connect the model. So over here, we have
the LM studio connected. These are the current models, which is Deep Seek
R O. Click on it. So this is the right
model that we have. Load the model. Here we go. Now the model is being loaded. We have downloaded it.
Now, in the next lesson, we're going to show you how
you could actually use it.
5. Examples on Local Applications: Have installed it locally.
Now, think about it this way. Now you've got Deep Seek
on your local device. I got for you a couple
examples award of notice. When you are using
Deep Seek locally, make sure that you're
closing any other tools or applications or anything
which will be right. We've installed it
for 4.82 GBs, zero, two GBs in this current case, and the CPU usage is 0%
at this current state. Then by itself will be
taxing on your system. So I got for you a couple
examples over here. Now, the first example is using deep seek to solve
mathematical equations. This is one of the powerhouse
features of Deep Seek, by the way, calculations
and coding. So if you installed it locally, you're using its
power in the best way possible to help you with
your day to day applications, think about analysis,
think about cation quadratic equation X square plus two x plus four
equals to zero. Let's try to solve it using
Deep Seek and you're able to see once you click on here
as if it's words with text, but it's not very clear
for mathematical purposes, calculation purposes,
analysis purposes. So what you do is navigate to appearance click on
view mode, Mark Down. This will give you the
result the same way you would see it on the
website for Deep Seek. Then once it says 2
minutes and 31 seconds, it took long because
basically I had multiple applications tools and models running at the same time, which is taxing on the system, which would slow
down the process. But if you close all the other artificial intelligence models applications that you're using, this will take seconds. Equation, you can go through
it to get the final answer, and the final answer is
correct, by the way. So think about it this way. It's like a personal assistant that you could use it
to help you do stuff. Not just simply calculations,
summarizing emails. Is the limit. So you
can get creative. You can get innovative with
the approach. Not just that. I got another example for you, where we need to
write the code for a bouncing ball in HTML. So as we can see here, it took about 7 minutes to go about the process, but
like I've mentioned, make sure that when
you are using it, that your device is
strong enough or you're not using other applications
or language models. This should take
a second or two, but because I have other
language models being used, I've installed them,
I've used them. At the same time, simultaneously,
this took a while. So here's the code by default. So if you're into
coding, for example, for me as an
engineer, we do code, various application purposes within the engineering domain, but it takes time to do so. So in order to
minimize the process, we could just simply use Deep Seek in this current
case to help us with the code generation process
on the spot locally. And this current case is an
example for a bouncing ball, and it could be used in HTML. You can put it to CSS. Also it incorporates
JavaScript, so it's up to you. If you are coder,
professional, developer, whatever application that you're at, you can use it locally. You can just simply
then copy paste the code into your
own application, and that's mainly
it simply by using proper prompt engineering
practices that we have went through as we are going through
the application. So that's very important to keep into consideration when
you're using deepsk. You have the ability to use it locally. I've
showed you the steps. Feel free to think it, to experiment with it, to try to find different
areas of application, and feel free to share the insights that
you're coming across. Let's say you've downloaded it locally and you
found some sort of creative way to use deeps colocally for a
certain application, feel free to share it with us with the rest
of the community. That way we have a pool of knowledge that you could share, that you
could learn from, to help everyone develop
by sharing your questions, by sharing your feedback, by sharing your snapshots or
screenshot of your results. That way you were able
to learn together and grow together in such a very, very important and
powerful domain in today's industry
and in today's world.
6. Wrapping Up: What do you think? I
truly hope they found the class helpful if
it helped you leverage the power of Deep Seek on artificial intelligence
locally within your device. It's a job well done, and I look forward to
receiving your feedback on the current class and
make sure they follow my profile for the latest
releases on updates on crucial industry expertise
that will surely help you stand out and excel and I'll see you
in the next class.
Engr. Hussein Attié, Entrepreneur I Engineer I Educator
