What is Hugging Face? - Models, Datasets & Spaces

Good morning everyone, how's it going today? Today we're going to be taking a look at an introduction to Hug and Face. And the idea right here is that by the end of the video you will have a much better understanding of everything that is possible within the Hug and Face ecosystem, which is likely much more than you think.

So what we're going to do is we're going to be splitting the whole thing into three parts. The first part is going to talk about Hug and Face models. Now we're going to see how to use them, how to search for them, how to understand the model page, and how to run it either locally or in the cloud.

After that, we're going to take a look at datasets, how to explore them, how to find the right one, what they are useful for, and we're also going to take a look at how to explore them using Data Studio, which essentially allows you to chat with your data. And finally, we're going to talk about spaces, which are tiny little web applications that allow you to explore the capabilities of existing models in Hugging Face. After that I'm going to show you how to deploy your own applications to Spaces so that you can deploy not only Gradio applications but also you can deploy your own MCP servers or your own custom web applications as long as they fit into a Docker container.

Now the ecosystem of Hugging Face is huge so I'm not going to be able to put absolutely everything into a single video but by the end you will have a much better idea of everything that is possible right here and I'm going to try to keep everything simple and not go into the technical details but if you have any questions just feel free to drop them in the comments and I'll be there to answer them okay so without any further ado let's get right into it all right so let's start off with the models part what are the models in Hugging Face and how can you actually use them so when you create an account you will be greeted with this welcome page right here and what you can do is go right here into models. And from here, you will be able to actually explore all of the models that are available in Hugging Face. As you can see, we have over 2.

6 million models available for you. On the left-hand side, you have a bunch of filters, such as if you want to use the cloud, inference providers, if you want to run them locally, which apps available they are. If you want to run it with a library, you have this available.

You can also filter by size, so the number of parameters, and by task. Let me show you a couple of models. I'm going to show you, first of all, a text generation model.

I'm going to sort them by trending. And right here, we can take a look at Minimax. So Minimax is a very famous and very powerful model that actually came out just a few weeks ago.

And as you can see right here in the left-hand side, we have their own description of the model. And in their description, as you can see, they have added some benchmarks. As you can see, this is a very, very good model.

It is comparable to Anthropics, Gemini's, and OpenAI's models, which is pretty cool. And on the right-hand side, you have the more technical specs of the model. So you can take a look at the number of downloads that this model had last month, some technical specs on safe tensors.

I'm going to tell you a little bit more about inference providers in a moment. But just to be clear, this means that the model is available to be run in the cloud, so you don't have to run it locally. this is of course very useful because this model is like 229 billion parameters which will definitely not run in any consumer hardware and here you have the spaces that are using this model i'm going to tell you a little bit more about that in a moment evaluation results collections including it etc so all right let's suppose that you have found this model and you want to use it.

So how do you use it? Now, there is this button right here to deploy it. And that is, of course, possible for very sophisticated setups.

If you really want to have your own model in your own cloud, you can use HF inference endpoints and we will deploy the model to the cloud of your choosing. However, the easiest way to do this is probably with inference providers, which means that we have already set the model available on inference providers like Cerebras, like Noveda, etc and you are able to just call it like you would call an open ai api actually and the models are available with the open ai sdk so if in your code you have you're calling an open ai model all you're going to have to do to change that for this model is just change the base url in the open ai initialization to the router from hug and face paste the hug and face api key or the token i'm I'm going to show you in a moment how to get that and you will be able to just get that just call the model like that. Very, very easy.

Super straightforward. You can also of course use the Hug and Face Hub library if you prefer, but it's also compatible with the OpenAI SDK and that for inference providers But if you want to run the model locally you can also do that You can also just click right here on transformers and you will have the exact script that you have to use to run this model locally. So I can just copy this, paste it in a colab notebook, and I will be able to use it.

However, there is kind of a tricky thing with this particular model because as you can see, this is a huge model, so this will not run in my computer. Let's suppose that I actually want a model that can run in my computer. So I'm going to go back to models right here.

I'm going to filter by size. So I'm going to say smaller than 9 billion parameters. And I can, instead of sort them by trending, I can sort them by downloads.

That means that I'm going to only see the models that actually people are downloading, which means that they're probably using them locally. And right here, I can filter by text generation. And here I find one.

So QAN 3 0. 6B. So that's a way smaller model that I can actually use.

So in order to use it, all that I have to do is just click right here and I can either copy the code from transformers or I can just open it directly using Colab like this. And it will open the notebook that I have to use and just run the code and it will automatically run the model for me. Now be sure to have added your Hugging Face token right here to your secrets as hftoken and then you will be able to just run this code okay so there we go now let's suppose that you want to use a model like this one but instead of using it on inference providers you just want to test it in a chat interface now we can also do that I'm just going to go back right here to minimax and I can click right here and use this model and right here you can see we have HuggingChat, which is the chat GPT kind of interface from HuggingFace.

And this one right here allows you to just run any model in inference providers in a chat GPT-like interface. So what I'm going to do right here, as you can see, the minimax model is already selected because I opened it straight from the model card and I can just query it right here. So who are you?

and there you go it finished thinking and entering my message it said i am an ai assistant i can help you jar jar jar and that's basically me already chatting with this ai model super easy and super straightforward and here of course you also have a very nice little chat window that is basically the same thing as running it on chat on hugging chat okay so there you go that is how you can run these models. But let's suppose that you don't want to use a text generation model. Let's suppose that you actually want a text to image model.

So let's go right here, click on text to image in tasks. And let's take a look at the most popular ones. I'm going to use Z image turbo.

I like this model quite a bit. And here I can see that it's also available on inference providers. I can also run it as before, but I'm going to run it right here.

So I'm going to say generate generate an image of a duck playing the guitar in cartoon style so I'm just going to click on generate and let's take a look at it and there you go so we have the duck playing the guitar in cartoon style and that was very easy and very straightforward and as you can see it's even faster than opening eyes image 1. 5 so and it's of course much more affordable so there you go that this, the models, how to use them. And of course, we're going to be delving a little bit more deeply into this in a moment, but I hope that this is clear.

Now let's take a look at datasets. All right, so now let's talk about Hug and Face datasets. What are they and how can you access them?

So in order to access them, you're going to go right here to datasets. And if you've ever done a little machine learning, you probably know that this is one of the most important parts of the hub. If you're used to datasets from, for example, say Kaggle, this is essentially the same thing, although the datasets on Hug and Face are a little bit more geared towards training, fine-tuning, and evils rather than data visualization, which is more of the data science Kaggle side.

You can upload and search for datasets here and apply filters just like with the models. Let me actually show you a few datasets that are pretty interesting. So I have VibraBox right here.

So this specific data set is a data set that includes a bunch of different versions of the same voice recording. So in this example, for example, let me play it for you. As you can see it is the same recording but over different interfaces and different kind of microphones So here a soft in microphone forehead accelerometer headset microphone So that this is going to be very useful to train a ASR model or automatic speech recognition model.

Um, another, um, data set that I want to show you is for example, if you're training a classification model. So we're going to go right here. I'm just going to search for this.

Just going to paste the URL. actually. This is an IMDB dataset that includes a bunch of different reviews for movies, and they are labeled whether they are negative or positive.

And as you can see, this is also going to be very useful if you're trying to train, for example, a classification model. Or for example, let's take a look at this other dataset, which is the CodeForces cuts, which is a dataset that includes a lot of coding problems alongside the solution generated by a very powerful language model. You can see the total size of the data set.

And actually something very cool that you can do with data sets is that you can visualize them in Data Studio. If you are a pro or enterprise user, you can also use Data Studio for your private data sets. So here we go.

Now Data Studio is loaded. and what we are going to do is we're going to load this the contents of this data set into an agent so that we can directly talk to the data set or chat with the data set so i'm going to select a small subset of it in order to actually load it so that it's not too heavy so i'm going to load the train subset and there we go so it has finished loading the data for me and now i can just talk with it. So I can just ask it, what is this data set useful for?

And as you can see, it's using GPT OSS 120B to answer my question. And here we have the answer. So for example, it has just a very quick explanation of the data in it.

What kind of model can I train with it? And there you go. We have a bunch of different examples of models that I could train with it.

So for example, a problem to code generation, a prompt to code generation, retrieval augmented generation, something like code correctness, a small transformer or gradient boosted trees on engineered features. So there you go. I mean, these are some pretty useful hints right here.

So basically the most useful thing right here for you is that you can chat with your data within Data Studio. And you can also do this, of course, with your own data sets. And so there you go.

That is for, that's it for data sets. Let's now take a look at Spaces. All right.

So now it's time to see one of the most fun and exciting parts of the entire Hugging Face ecosystem, which is Spaces. Now, Spaces are where you can see the models in action. They are basically just web applications that are mostly used to showcase the models.

However, you can upload pretty much any application that you want, and it is free hosting, so feel free to take a look at it. So let me just show you some of the most common and useful spaces that are here. So in order to open spaces, all you're going to have to do is click right here on spaces, and you're going to go right here, and you're going to be able to explore all the spaces available.

Now, I'm going to show you a few. I can show you, for example, an image generation space. And we can just click on it.

And as you can see, this is basically just a regular web app. And this one right here has some examples of prompts that we can use. We can click on generate image.

And here we can just take a look at the image that is being generated. We're trying to generate a young Chinese woman in Red Hanfu, intricate embroidery, blah, blah, blah. Let's take a look at it.

And there we go. So here we have the result. We can also generate another image.

So for example, a coffee, coffee shop interior with warm lightning, etc. And here we have the result. There we go.

And this one right here is, of course, Z Image Turbo. If you want to take a look at the model, it is linked right here. And there we go.

And something that I wanted to show you before when we were talking about models is that within the model card or the model page, whenever you have a model that you're interested in, you can go all the way to the bottom right here. and you can find the spaces that actually make this model available. So here you can see that we have the space from Tongi MAI.

We can have this space that I just showed you, which is this one right here. And this is a good way to just test the models right away without having to use a notebook or trying to actually run the code You can just find a space that just showcases the model for you It very very straightforward Let take a look at other spaces that are pretty fun to use Some of my favorite spaces are the arenas. These are spaces that give you a clear comparison of two models and they don't really tell you which generation comes from which model so you can vote for the generation that you prefer.

And that way we generate a ranking of the models that people prefer based on ELO scores. So here, for example, we can, I'm just going to search for Arena. Here you can see, for example, we can find an image Arena dashboard, leaderboard.

So here we have the prompt. So change the label to say organic oat milk in green font. And here you can see the two results.

I actually feel like both are okay. Oh no, this one actually messed up the text. This one's a little bit better, but still not great.

I'm just going to vote for this one. You can see that this is a very, very fun kind of space and you can actually take a look at the leaderboard. As you can see right here, OpenAI is winning.

Then we have Nanobanana 2 and Nanobanana Pro also right here in the top results. And you can also have the API pricings right here. You can see that OpenAI is definitely among the most expensive ones alongside Nano Banana.

But you have also these open source models, for example, Quen Image Max, which is much more affordable. And there you go. Another space that I would like to show you, for example, if we're looking for Arena, is for example, the Music Arena, where you can just compare music generation models and see which one you like better.

So for example, you can just write a random prompt, black metal with shrieking vocals and tremolo guitars. We can just generate it from two models that are available right here and just compare which one we like the most. And there we are.

We have the two generations right here. So with shrieking vocals and tremolo guitars, let's see. well I definitely did not like that one well I think it's clear that B is better right here I can just vote and there you go and I can also take a look at the leaderboard and as you can see right here we have the most the models that have beaten the most models in the arena so we have refusion flaws we have Sonodo and 11Labs Music.

So that's it. There you go. That is this another very cool space.

And what I told you before is that you can create your own spaces. And that is also very straightforward and very simple. All that you have to do is we're going to go back to Hugging Face.

We're going to click right here. Sorry, here. And you're going to click on new space and this basically works like creating a github repository just create the space name add a short description add a license and select which kind of application you're deploying gradio is probably the most appropriate because it's the the best integrated with spaces you can also if you use gradio to showcase your models you can just deploy them right here and they will automatically be taken as MCP servers if you configure that correctly.

But something really cool is that you can basically deploy any application as long as it fits into a Docker container, which basically means that you can deploy any application to a space or also static content right here. So very, very fun. It is free hosting.

So definitely take, I mean, take advantage of it. If you are on a pro account, you can use a zero GPU for your Gradio applications. And a quick tip about this is that you can also just create your NCP servers and deploy them right here for free.

It's a super fun and super quick way to just deploy your agent tools. So there you go. That is it for spaces on Hugging Face.

There we go. So we have effectively covered everything about models, datasets, and spaces, and a little bit about the product offerings around them that can make your experience even better. Now, I try not to make things too technical, but in case you have any questions, feel free to just let me know.

You can just send a comment right here in the comments and I'll be there to help you out. So thanks a lot for watching and I will see you in the next one. Good morning everyone, how's it going today?

Today we're going to be taking a look at an introduction to Hug and Face. And the idea right here is that by the end of the video you will have a much better understanding of everything that is possible within the Hug and Face ecosystem, which is likely much more than you think. So what we're going to do is we're going to be splitting the whole thing into three parts.

The first part is going to talk about Hug and Face models. Now we're going to see how to use them, how to search for them, how to understand the model page, and how to run it either locally or in the cloud. After that, we're going to take a look at datasets, how to explore them, how to find the right one, what they are useful for, and we're also going to take a look at how to explore them using Data Studio, which essentially allows you to chat with your data.

And finally, we're going to talk about spaces, which are tiny little web applications that allow you to explore the capabilities of existing models in Hugging Face. After that I'm going to show you how to deploy your own applications to Spaces so that you can deploy not only Gradio applications but also you can deploy your own MCP servers or your own custom web applications as long as they fit into a Docker container. Now the ecosystem of Hugging Face is huge so I'm not going to be able to put absolutely everything into a single video but by the end you will have a much better idea of everything that is possible right here and I'm going to try to keep everything simple and not go into the technical details but if you have any questions just feel free to drop them in the comments and I'll be there to answer them okay so without any further ado let's get right into it all right so let's start off with the models part what are the models in Hugging Face and how can you actually use them so when you create an account you will be greeted with this welcome page right here and what you can do is go right here into models.

And from here, you will be able to actually explore all of the models that are available in Hugging Face. As you can see, we have over 2. 6 million models available for you.

On the left-hand side, you have a bunch of filters, such as if you want to use the cloud, inference providers, if you want to run them locally, which apps available they are. If you want to run it with a library, you have this available. You can also filter by size, so the number of parameters, and by task.

Let me show you a couple of models. I'm going to show you, first of all, a text generation model. I'm going to sort them by trending.

And right here, we can take a look at Minimax. So Minimax is a very famous and very powerful model that actually came out just a few weeks ago. And as you can see right here in the left-hand side, we have their own description of the model.

And in their description, as you can see, they have added some benchmarks. As you can see, this is a very, very good model. It is comparable to Anthropics, Gemini's, and OpenAI's models, which is pretty cool.

And on the right-hand side, you have the more technical specs of the model. So you can take a look at the number of downloads that this model had last month, some technical specs on safe tensors. I'm going to tell you a little bit more about inference providers in a moment.

But just to be clear, this means that the model is available to be run in the cloud, so you don't have to run it locally. this is of course very useful because this model is like 229 billion parameters which will definitely not run in any consumer hardware and here you have the spaces that are using this model i'm going to tell you a little bit more about that in a moment evaluation results collections including it etc so all right let's suppose that you have found this model and you want to use it. So how do you use it?

Now, there is this button right here to deploy it. And that is, of course, possible for very sophisticated setups. If you really want to have your own model in your own cloud, you can use HF inference endpoints and we will deploy the model to the cloud of your choosing.

However, the easiest way to do this is probably with inference providers, which means that we have already set the model available on inference providers like Cerebras, like Noveda, etc and you are able to just call it like you would call an open ai api actually and the models are available with the open ai sdk so if in your code you have you're calling an open ai model all you're going to have to do to change that for this model is just change the base url in the open ai initialization to the router from hug and face paste the hug and face api key or the token i'm I'm going to show you in a moment how to get that and you will be able to just get that just call the model like that. Very, very easy. Super straightforward.

You can also of course use the Hug and Face Hub library if you prefer, but it's also compatible with the OpenAI SDK and that for inference providers But if you want to run the model locally you can also do that You can also just click right here on transformers and you will have the exact script that you have to use to run this model locally. So I can just copy this, paste it in a colab notebook, and I will be able to use it. However, there is kind of a tricky thing with this particular model because as you can see, this is a huge model, so this will not run in my computer.

Let's suppose that I actually want a model that can run in my computer. So I'm going to go back to models right here. I'm going to filter by size.

So I'm going to say smaller than 9 billion parameters. And I can, instead of sort them by trending, I can sort them by downloads. That means that I'm going to only see the models that actually people are downloading, which means that they're probably using them locally.

And right here, I can filter by text generation. And here I find one. So QAN 3 0.

6B. So that's a way smaller model that I can actually use. So in order to use it, all that I have to do is just click right here and I can either copy the code from transformers or I can just open it directly using Colab like this.

And it will open the notebook that I have to use and just run the code and it will automatically run the model for me. Now be sure to have added your Hugging Face token right here to your secrets as hftoken and then you will be able to just run this code okay so there we go now let's suppose that you want to use a model like this one but instead of using it on inference providers you just want to test it in a chat interface now we can also do that I'm just going to go back right here to minimax and I can click right here and use this model and right here you can see we have HuggingChat, which is the chat GPT kind of interface from HuggingFace. And this one right here allows you to just run any model in inference providers in a chat GPT-like interface.

So what I'm going to do right here, as you can see, the minimax model is already selected because I opened it straight from the model card and I can just query it right here. So who are you? and there you go it finished thinking and entering my message it said i am an ai assistant i can help you jar jar jar and that's basically me already chatting with this ai model super easy and super straightforward and here of course you also have a very nice little chat window that is basically the same thing as running it on chat on hugging chat okay so there you go that is how you can run these models.

But let's suppose that you don't want to use a text generation model. Let's suppose that you actually want a text to image model. So let's go right here, click on text to image in tasks.

And let's take a look at the most popular ones. I'm going to use Z image turbo. I like this model quite a bit.

And here I can see that it's also available on inference providers. I can also run it as before, but I'm going to run it right here. So I'm going to say generate generate an image of a duck playing the guitar in cartoon style so I'm just going to click on generate and let's take a look at it and there you go so we have the duck playing the guitar in cartoon style and that was very easy and very straightforward and as you can see it's even faster than opening eyes image 1.

5 so and it's of course much more affordable so there you go that this, the models, how to use them. And of course, we're going to be delving a little bit more deeply into this in a moment, but I hope that this is clear. Now let's take a look at datasets.

All right, so now let's talk about Hug and Face datasets. What are they and how can you access them? So in order to access them, you're going to go right here to datasets.

And if you've ever done a little machine learning, you probably know that this is one of the most important parts of the hub. If you're used to datasets from, for example, say Kaggle, this is essentially the same thing, although the datasets on Hug and Face are a little bit more geared towards training, fine-tuning, and evils rather than data visualization, which is more of the data science Kaggle side. You can upload and search for datasets here and apply filters just like with the models.

Let me actually show you a few datasets that are pretty interesting. So I have VibraBox right here. So this specific data set is a data set that includes a bunch of different versions of the same voice recording.

So in this example, for example, let me play it for you. As you can see it is the same recording but over different interfaces and different kind of microphones So here a soft in microphone forehead accelerometer headset microphone So that this is going to be very useful to train a ASR model or automatic speech recognition model. Um, another, um, data set that I want to show you is for example, if you're training a classification model.

So we're going to go right here. I'm just going to search for this. Just going to paste the URL.

actually. This is an IMDB dataset that includes a bunch of different reviews for movies, and they are labeled whether they are negative or positive. And as you can see, this is also going to be very useful if you're trying to train, for example, a classification model.

Or for example, let's take a look at this other dataset, which is the CodeForces cuts, which is a dataset that includes a lot of coding problems alongside the solution generated by a very powerful language model. You can see the total size of the data set. And actually something very cool that you can do with data sets is that you can visualize them in Data Studio.

If you are a pro or enterprise user, you can also use Data Studio for your private data sets. So here we go. Now Data Studio is loaded.

and what we are going to do is we're going to load this the contents of this data set into an agent so that we can directly talk to the data set or chat with the data set so i'm going to select a small subset of it in order to actually load it so that it's not too heavy so i'm going to load the train subset and there we go so it has finished loading the data for me and now i can just talk with it. So I can just ask it, what is this data set useful for? And as you can see, it's using GPT OSS 120B to answer my question.

And here we have the answer. So for example, it has just a very quick explanation of the data in it. What kind of model can I train with it?

And there you go. We have a bunch of different examples of models that I could train with it. So for example, a problem to code generation, a prompt to code generation, retrieval augmented generation, something like code correctness, a small transformer or gradient boosted trees on engineered features.

So there you go. I mean, these are some pretty useful hints right here. So basically the most useful thing right here for you is that you can chat with your data within Data Studio.

And you can also do this, of course, with your own data sets. And so there you go. That is for, that's it for data sets.

Let's now take a look at Spaces. All right. So now it's time to see one of the most fun and exciting parts of the entire Hugging Face ecosystem, which is Spaces.

Now, Spaces are where you can see the models in action. They are basically just web applications that are mostly used to showcase the models. However, you can upload pretty much any application that you want, and it is free hosting, so feel free to take a look at it.

So let me just show you some of the most common and useful spaces that are here. So in order to open spaces, all you're going to have to do is click right here on spaces, and you're going to go right here, and you're going to be able to explore all the spaces available. Now, I'm going to show you a few.

I can show you, for example, an image generation space. And we can just click on it. And as you can see, this is basically just a regular web app.

And this one right here has some examples of prompts that we can use. We can click on generate image. And here we can just take a look at the image that is being generated.

We're trying to generate a young Chinese woman in Red Hanfu, intricate embroidery, blah, blah, blah. Let's take a look at it. And there we go.

So here we have the result. We can also generate another image. So for example, a coffee, coffee shop interior with warm lightning, etc.

And here we have the result. There we go. And this one right here is, of course, Z Image Turbo.

If you want to take a look at the model, it is linked right here. And there we go. And something that I wanted to show you before when we were talking about models is that within the model card or the model page, whenever you have a model that you're interested in, you can go all the way to the bottom right here.

and you can find the spaces that actually make this model available. So here you can see that we have the space from Tongi MAI. We can have this space that I just showed you, which is this one right here.

And this is a good way to just test the models right away without having to use a notebook or trying to actually run the code You can just find a space that just showcases the model for you It very very straightforward Let take a look at other spaces that are pretty fun to use Some of my favorite spaces are the arenas. These are spaces that give you a clear comparison of two models and they don't really tell you which generation comes from which model so you can vote for the generation that you prefer. And that way we generate a ranking of the models that people prefer based on ELO scores.

So here, for example, we can, I'm just going to search for Arena. Here you can see, for example, we can find an image Arena dashboard, leaderboard. So here we have the prompt.

So change the label to say organic oat milk in green font. And here you can see the two results. I actually feel like both are okay.

Oh no, this one actually messed up the text. This one's a little bit better, but still not great. I'm just going to vote for this one.

You can see that this is a very, very fun kind of space and you can actually take a look at the leaderboard. As you can see right here, OpenAI is winning. Then we have Nanobanana 2 and Nanobanana Pro also right here in the top results.

And you can also have the API pricings right here. You can see that OpenAI is definitely among the most expensive ones alongside Nano Banana. But you have also these open source models, for example, Quen Image Max, which is much more affordable.

And there you go. Another space that I would like to show you, for example, if we're looking for Arena, is for example, the Music Arena, where you can just compare music generation models and see which one you like better. So for example, you can just write a random prompt, black metal with shrieking vocals and tremolo guitars.

We can just generate it from two models that are available right here and just compare which one we like the most. And there we are. We have the two generations right here.

So with shrieking vocals and tremolo guitars, let's see. well I definitely did not like that one well I think it's clear that B is better right here I can just vote and there you go and I can also take a look at the leaderboard and as you can see right here we have the most the models that have beaten the most models in the arena so we have refusion flaws we have Sonodo and 11Labs Music. So that's it.

There you go. That is this another very cool space. And what I told you before is that you can create your own spaces.

And that is also very straightforward and very simple. All that you have to do is we're going to go back to Hugging Face. We're going to click right here.

Sorry, here. And you're going to click on new space and this basically works like creating a github repository just create the space name add a short description add a license and select which kind of application you're deploying gradio is probably the most appropriate because it's the the best integrated with spaces you can also if you use gradio to showcase your models you can just deploy them right here and they will automatically be taken as MCP servers if you configure that correctly. But something really cool is that you can basically deploy any application as long as it fits into a Docker container, which basically means that you can deploy any application to a space or also static content right here.

So very, very fun. It is free hosting. So definitely take, I mean, take advantage of it.

If you are on a pro account, you can use a zero GPU for your Gradio applications. And a quick tip about this is that you can also just create your NCP servers and deploy them right here for free. It's a super fun and super quick way to just deploy your agent tools.

So there you go. That is it for spaces on Hugging Face. There we go.

So we have effectively covered everything about models, datasets, and spaces, and a little bit about the product offerings around them that can make your experience even better. Now, I try not to make things too technical, but in case you have any questions, feel free to just let me know. You can just send a comment right here in the comments and I'll be there to help you out.

So thanks a lot for watching and I will see you in the next one.