Getting started with Anaconda Desktop

This guide is designed to walk you through the essential steps to begin using Anaconda Desktop. It provides guidance on signing in, launching applications, downloading models, chatting with a model, and loading models into the API server.

Before you start

Make sure you have Anaconda Desktop installed.

Launching Desktop

Open Desktop using the following instructions:

Windows

From the Start menu, search for “Anaconda Desktop” and select to open.

macOS

Open Launchpad, then select the Anaconda Desktop icon.

Signing in

Anaconda Desktop requires you to have an Anaconda account to sign in. If you’re part of an Anaconda Platform organization, sign in with your assigned credentials.

Understanding Desktop

The Dashboard

When Desktop opens, you’ll start at the Dashboard, where you’re provided a few ideas on how to get started. After you’ve interacted with Desktop by, for example, downloading a model or starting a chat, your most recent activities will appear on your dashboard.

1. Recents

   From the Recents table, you can return to a previous chat, start a new chat with a model (text-generation type models only), or start an API server. Hover over a table row and select Chat or Start Server. Select the Actions menu to Delete a table item.

   
2. Models

   The Models section will only display if you have eight rows or fewer in the Recents table.

   The Models section displays a few of Anaconda’s recommended models. Select the model’s tile to view the model details, select a file quantization level, then select Download.

   The downloaded model appears in the Recents table on your dashboard and the AI Models page.

3. Applications

   The Applications section shows your most recently launched applications. Select anywhere on an application’s tile to launch it from its default environment. Hover over the card and select the Actions menu to launch the application from a different environment.

4. Documentation

   The Documentation section provides helpful links to the Anaconda Desktop documentation.

Navigating Desktop

Let’s take a look at the Anaconda Desktop navigation menu.

1. New

   Quickly open a new chat or navigate to the Model Catalog page.

2. Dashboard

   See your most recent activity, view and download popular models, and access learning materials.

3. Environments

   View and manage your environments.

4. AI Models

   View your downloaded models.

5. Model Servers

   Start an API server to interact with your models.

6. Model Catalog

   Search, filter, and download applications and models.

7. Documentation

   Visit Anaconda’s Desktop documentation (you’re already here!).

8. Support

   Open a Support ticket to request help with a technical issue.

You can toggle Dark Mode on and off from the user dropdown menu.

Launching JupyterLab from Desktop

To launch JupyterLab from Desktop:

1. From the Dashboard, select the JupyterLab tile.

If JupyterLab is not installed in any of your existing environments, Anaconda Desktop creates a new environment, installs JupyterLab, and then launches the application from that environment.

You can also launch JupyterLab from the Applications tab in the Catalog.

For more information about JupyterLab and other available applications, see Application Catalog.

Downloading and interacting with models

A typical workflow in Desktop might start with browsing available models and downloading one, then chatting with the model to understand its capabilities, and finally using it in a development workflow with the API server.

Exploring and downloading models

If you haven’t downloaded a model yet, you can explore Anaconda’s curated collection:

1. Select Catalog from the left-hand navigation.
2. Select the Models tab.
3. Browse the available models or use the search and filter options to find one that suits your needs.
4. Select a model tile to view its details, including resource requirements and performance information.
5. Select a Quantization Method from the dropdown.
6. Select Download to download the model locally.

Models are listed alphabetically by name. Verify that you have the available disk space and RAM required before you download a model. Most models fall between 2-10 GB, but some require up to 150 GB of storage space.

For more information about models and quantization methods, see Model Catalog.

Chatting with a model

Once you have a model downloaded, the easiest way to interact with it is through the chat interface:

You can only chat with models labeled text-generation. Learn more about model types.

1. Select Dashboard from the left-hand navigation.
2. Locate your downloaded model in the Recents table.
3. Hover over the model and select Chat.
4. Enter a prompt to start chatting with the model!

Alternatively, you can navigate to AI Models, hover over the model, and select Chat.

For more information on using the chat feature, see Chat.

Starting an API server

If you want to test a model’s performance within applications you’re developing, load the model into Desktop’s built-in API server:

1. Select Model Servers from the left-hand navigation.
2. Choose your model from the Model dropdown.
3. Select your model file from the File dropdown.
4. Configure your server settings (address, port, API key) as needed.
5. Select Start to start the server.

Alternatively, you can navigate to AI Models, hover over the model, and select Start Server.

Your model is now accessible via API calls at the specified server address for use in your own applications.

For more information about the API server, see API server.

What’s next?

Now that you have Desktop set up and running:

• Install applications: Explore the Application Catalog to install and launch popular data science applications like JupyterLab, Spyder, and Jupyter Notebook directly from Desktop, with automatic environment setup and dependency management.
• Manage environments: Explore the Environments page to create, manage, and delete environments.
• Learn with tutorials: Get hands-on experience by building a chatbot application or text comparison tool with step-by-step guidance.
• Use the Desktop API: Explore the Desktop API to programmatically manage models, control servers, and integrate Desktop’s capabilities directly into your applications.
• Use the Anaconda AI CLI tool: Learn how to use the Anaconda AI CLI to manage your models, servers, and applications from the command line.