Get Started With Our Platform
A quick-start guide on the Clarifai Community Platform
Clarifai Community is a low-code, no-code platform that provides everything you need to build AI-powered apps. You can use it to discover, test, create, and share a wide range of state-of-the-art AI resources, including:
-
Models — Allow you to convert inputs to outputs by generating predictions based on the patterns extracted from the input data;
-
Workflows — Allow you to create robust AI solutions by applying multiple models in a single flow;
-
Apps — Act as self-contained projects for storing, handling, and managing the data and resources related to the tasks you want to accomplish on the platform;
-
Modules — Allow you to create performant AI solutions by combining task-specific UI modules into your applications to do more.
Our platform offers powerful functionalities for computer vision, natural language processing, and audio recognition. With the full-lifecycle deep learning platform, you can transform unstructured images, videos, texts, and audio data into structured data at a significantly faster and more accurate rate than humans could do on their own.
-
We provide a seamless, end-to-end AI workspace that you can use to manage your data, annotate your inputs, and build custom models.
-
Instead of starting from scratch, you can find many pre-trained models and workflows in the platform. You can jumpstart your AI solutions with pre-built models from Clarifai, leading AI providers like Hugging Face, Meta, and Google, and other top contributors.
We have everything you need to unleash your full AI potential!
Explore the Community
No login is required to explore our Clarifai platform.
You can go to the Community's main page and search for what you need to build state-of-the-art AI-powered functionalities.
You can explore our huge collection of published resources and discover the one(s) best suitable for your use case.
For example, to search for models, click the Models tab on the menu bar. Next, use the search box to look for the model you require — by its ID, description, or user.
You can use the various filter options on the left side of the page to quickly search for the specific model you require. You can also use the sorting options on the right side of the page to arrange the listed items according to your preference.
Get Started
Once you log in to the Community platform, you’ll find a horizontal navigation bar that helps you to effortlessly go to the sections of your interest.
My Apps
The My Apps navigation element lets you easily access the apps and templates you’ve created on the Community platform.
If you click the element, you’ll be redirected to a page that lists the apps and templates you own. This allows you to search and access them quickly.
On the same page, you can click other tabs to access other resources you own — such as the Models tab or the Workflows tab to access the models and workflows you own, respectively.
Community
The Community navigation element lets you easily access the resources publicly available on the Community platform.
If you click the element, you’ll be redirected to a page that lets you explore the resources available on the Community platform, including the featured ones.
On the same Community explore page, you can click other tabs to access other resources available publicly — such as the Models tab or the Workflows tab to search and access the models and workflows available for everyone to use.
The Featured tab lets you explore the resources that stand out on the platform.
Question Mark
The Question Mark navigation element lets you easily access other resources that could help you make the most of the Community platform.
If you click the element, a drop-down list will appear with items that point to various external resources you can access, such as Contact Us, Documentation, API Status, and Report a Bug.
Create
The Create navigation element lets you easily create an application on the Community platform.
If you click the element, a small window will pop up, asking you to choose the method you want to use to create a new application.
There are two ways to create an application:
-
Start with a blank app — Begin from scratch, building your app from the ground up. Learn more about it here.
-
Use a template — Use pre-built blueprints to create apps quickly. Learn more about it here.
Let's demonstrate how you can use the first option to create an app.
After selecting the Start with a Blank App option, a window will appear requiring you to provide the required information to create a new application.
Simply provide a unique app ID and go with the other default options.
Finally, click the Create App button at the upper-right corner of the page to complete the creation of your app.
Profile Icon
The Profile Icon navigation element lets you easily access the details related to your account on the Community platform. The icon is a small circle that contains either the initials of your name or a photo you've uploaded.
If you click the element, a drop-down list will appear with items that point to various pages within the Community platform.
- Organization profile — This allows you to start using the Community platform with your organization profile. If you select it, you’ll begin interacting with the platform as an organization — just like a normal user would, albeit with some other added capabilities. You can also add an organization from there, making it easier to collaborate and manage projects.
- My Apps — This allows you to go to the apps’ listing page and access your own resources, such as apps, models, and workflows.
- Collaborations — This allows you to access the apps that you share with other team members and work on together.
- Account — This allows you to access and edit the details you provided when registering for an account. If you select it, you’ll be redirected to a page that lets you accomplish various tasks, such as updating your account information, changing your user ID, adding an email address, updating your user profile visibility to either public or private, or deleting your account.
- Billing — This allows you to access the billing information associated with your account. If you select it, you’ll be redirected to a page that lets you change your current billing plan, add credit card details, or access the invoices associated with your account.
- Security — This allows you to access the security details associated with your account. If you select it, you’ll be redirected to a page that lets you update your password, create a Personal Access Token (PAT), or activate two-factor authentication.
- Usage — This allows you to access the usage dashboard that monitors how you consume Clarifai’s resources.
- Sign Out — This allows you to log out of the Community platform. If you select it, you’ll be signed out of the platform.
How to Add Inputs
An input is any data you provide to the Community platform. You can upload your inputs — such as images, videos, audio, or texts — to be used for data labeling, training new models, performing searches, or making predictions.
To upload data, such as images, to an app you've created, navigate to the app’s individual page and select the Inputs option on the collapsible left sidebar. We previously demonstrated how to create an app.
Next, click the Upload Inputs button and upload the inputs you want to add to the app. A small window will pop up, allowing you to upload your inputs either directly from your local device or by providing a publicly accessible URL.
For this illustration, let's select the URL tab and provide the following images of dogs and cats:
https://samples.clarifai.com/dog1.jpeg
https://samples.clarifai.com/dog2.jpeg
https://samples.clarifai.com/dog3.jpeg
https://samples.clarifai.com/cat1.jpeg
https://samples.clarifai.com/cat2.jpeg
https://samples.clarifai.com/cat3.jpeg
Finally, click the Upload inputs button at the bottom of the pop-up window to complete uploading your inputs.
Click here to learn more about how to add and make use of your inputs on the Clarifai platform.
How to Label Your Inputs
Labeling (also known as annotating) refers to the process of adding one or more relevant tags, or keywords — usually referred to as concepts — that best describe the state of your inputs.
For example, annotations might indicate whether an image contains a dog or a cat, which words were spoken in a recorded audio file, or if a concrete block contains cracks.
Successfully labeling data is a key ingredient to any custom AI solution. You can use a concept to annotate an input if that input has that entity. That’s how you prepare training data to teach your models to recognize new concepts. After training your model using the annotated concepts, the model can learn to recognize them when it encounters data without those tags.
To label your inputs on the Clarifai platform, navigate to your app’s individual page and select the Inputs option on the collapsible left sidebar.
On the Input-Manager page, select the input(s) you like to label. For this illustration, let's start by selecting the images of dogs we added previously.
Next, click the Label as … button that appears on the bottom section of the page.
The small window that pops up allows you to annotate the selected input(s) with concepts. Let's enter the dog concept in the Concepts field and click the Add new concept button.
Finally, click the Add to inputs button at the bottom of the pop-up window to complete labeling your images of dogs.
Similarly, annotate your cat images with the cat concept.
After annotating the inputs, you'll find the annotations listed in the Labels section of the Input-Manager page.
How to Create and Train a Custom Model
The Clarifai Community platform makes it unbelievably easy to create and train new models. With a custom model, you can create a solution that best addresses your specific use case.
To create a custom model, start by navigating to your app and uploading data, such as images of dogs and cats, and annotating them — as earlier described. The input data, labeled with concepts, is what will be used for training your model.
After adding inputs and labeling them with the concepts you want your model to see, you can now proceed to create your own custom model.
Navigate to your app's individual page and select the Models option on the collapsible left sidebar. On the ensuing page, click the Add Model button at the upper-right corner of the page.
In the window that pops up, select the Build a Custom Model option and click the Continue button.
Next, choose the type of model you want to create. For this example, we want to create a simple classification model that can identify the images of cats and dogs.
So, let’s choose the Transfer Learn model type, which uses the transfer learning technique to leverage the knowledge gained from a pre-trained model to facilitate the learning process of a new model for a related problem. This lets us train new models in seconds, not minutes or hours.
On the ensuing page, provide a unique, memorable ID for your model.
Click the Continue to Configure Model button to create your model.
Next, set up the details required to train your model.
-
Dataset — You can select a dataset to use to train the model. Since we don't have a dataset for this example, we'll use the default dataset automatically generated with all the compatible labeled inputs in the app.
-
Base Embed Model — You can select the base embed model to use for embeddings, which has to be one of the embed models in the app workflow. This allows you to specify the specific model in case the default workflow of your app has multiple embedding models present. For this example, let's go with the default option.
-
Concepts — Select the concepts that you want the model to predict. For this example, let's choose the concepts of
dogandcat,which we previously used to label our images. -
Concepts Mutually Exclusive — Let's turn the button on to indicate no overlap between any of the model concepts.
-
Enrich Dataset — If it's enabled (set to Automatic), it lets you enrich the model with supplemental data from pre-built datasets of negative embeddings, which improves the model's accuracy. Otherwise, setting it to Disabled lets you not use the negative embeddings, whether they are available or not. For this example, let's go with the default option.
-
Inference Settings — Optionally, you can configure the inference settings for your model. For this example, let's go with the default option.
-
Train Model — After configuring the settings, click the Train Model button to begin training your model.
Once the model is trained, you can put it to work, such as use it to make predictions. We'll illustrate how to make predictions below.
As mentioned earlier, training helps your model learn from the annotated concepts in your inputs so that it can recognize them.
Click here to learn more about how to create and train models on the Clarifai platform.
How to Create a Workflow
Workflows is a useful Clarifai's feature that allows you to combine multiple models and carry out different operations on our platform. With workflows, you can create a powerful multi-model system that meets various use cases in a single call — instead of relying only on a single model.
You can create workflows out of any public models available on the Community platform or custom models that you have created for your app. The inputs and outputs supported by your custom models will depend on the inputs and outputs supported by the public models, or model templates that you have used to build them.
In this example, we'll create a simple custom workflow that first extracts text from an image and then translates the extracted text to Spanish.
We'll connect the following two models to achieve our objective:
- The ocr-scene-english-paddleocr model, which detects and recognizes English texts in images;
- The text-translation-english-spanish model, which translates texts from English to Spanish.
To start with, navigate to your app’s individual page and select the Workflows option on the collapsible left sidebar. Next, click the Create Workflow button at the upper-right corner of the page.
You'll be redirected to a simple no-code, drag-and-drop interface that allows you to connect your models together. You'll need to connect the input nodes in your workflow. This allows you to link models to any models that precede them in the visual graph.
In the Filter model type search box on the left side of the page, search successively for an “optical-character-recognizer” node and a “text-to-text” node.
After finding each of the nodes, drag and drop each of them on the workspace pane. Also, draw a line that connects the two nodes, which shows the flow of information from one node to another.
Next, click the optical-character-recognizer button. In the search box that appears on the right side of the page, search for and select the ocr-scene-english-paddleocr model as the one you’ll use for optical character recognition. Its version will be automatically selected for you.
Similarly, click the text-to-text button and search for and select the text-translation-english-spanish model as the one you'll use for translating the extracted text from English to Spanish.
Lastly, click the Save Workflow button to save the workflow. That will save the state of your workflow. Now you are ready to put it to work, such as use it to make predictions. We'll illustrate how to make predictions below.
Here is a video tutorial that showcases how to create two different multi-model workflows in the Community platform.
Click here to learn more about how to create and use workflows on the Clarifai platform.
How to Make Predictions
You can use your own custom models or workflows to make predictions on the Community platform. You can also get started right away by using one of the pre-optimized models or workflows published on the Community platform.
When you upload an input to a model or workflow, you'll see its predictions. The platform analyzes the input and identifies its contents, returning a list of concepts along with the corresponding probabilities of each concept present in the input.
For example, you can use the pre-built Clarifai's face-detection model to detect the location of human faces in images. If you go to its main page, you’ll notice that there are some default photos that show how the model works. There are bounding boxes around the detected faces. On the right side of the page, there are previews of the detected faces alongside the probability value of their correctness.
To make your own predictions using the model, click the blue + (plus) button. A modal will pop up, providing a convenient interface for adding input data and examining predictions.
Select the Try your own image or video option. A small window will appear, allowing you to upload the image you want to analyze and see its face predictions.
For this example, let's use this image of a crowd of seated monks. After uploading the image, the model will analyze it and return a list of faces identified on the right side of the page.
Click here to learn more about how to make predictions on the Clarifai platform.
How to Use Modules
Modules are custom plugins that provide UI elements as well as the ability to perform some computation around our API. Modules enable you to extend the Clarifai UIs, augment backend processing capabilities, and build customized solutions quickly and easily.
You can create your own modules or use the ones already available in the Community platform.
In this example, we'll demonstrate how to install and use the LLM Battleground module, which allows you to evaluate and compare the performance of multiple Large Language Models (LLMs) simultaneously.
To install the LLM Battleground module, go to its individual page and click the Install Module button in the upper-right corner.
Next, on the small window that pops up, select a desired user and destination app for installing the module.
Click the Confirm button.
On the ensuing page, choose a destination URL for installing the module, select the module visibility, and click the Install to this App! button.
After the module has been successfully installed, click the Click to use installed module version button to start using it.
You can also refresh the page so that your installed module appears in the left sidebar under the Installed Modules category.
If you click the button, you’ll be redirected to a page that requires you to authorize the module to access data on your behalf. You will see the permissions the module requires.
Click the Authorize button.
You’ll be redirected to the installed module page, where you can start using it for comparing the performance of LLMs.
Click here to learn more about how to create and use modules on the Clarifai platform.