Skip to content

Using the Playground

Invite-only access

Access to Cleura AI Services is currently invite-only.

To access the Playground, log into the Cleura AI Inference Dashboard with a User or an Admin API key. The dashboard layout differs by API key type. The following examples are from the User API key dashboard.

Chat

In the left-hand pane, click Chat. Then, in the right-hand pane, use the Model drop-down menu to select an available LLM.

Chatting with the selected model Chatting with the selected model

Before you start chatting, you might want to adjust parameters like System prompt, Temperature, and Effort.

To begin chatting, enter your prompt in the Ask anything text input box at the bottom of the main pane. When you’re ready, press Enter or click the Send button. The answer appears above, and at the top of the page you also see the prompt you just entered. Note that you may expand the Reasoning box and see what the model understood about your prompt and how it decided to produce an answer.

Comparing answers

For any prompt, you might want to compare the answers of two different models. In that case, first click the Compare button at the top of the main pane. Then, on the right-hand pane, use the PRIMARY and SECONDARY tabs to select the two models. In the example we provide here, we use the System prompt box to describe a common characteristic for each model.

Trying the same prompt on two different models Trying the same prompt on two different models

After you type your prompt and send it, the main pane splits vertically. In the left half, you get the answer from the PRIMARY model, and in the right half, the answer from the SECONDARY model.

Embeddings

To create Embeddings, click that label in the left-hand pane. In the right-hand pane, notice that a suitable model is selected automatically.

In the INPUT box of the main pane, type anything you wish to convert into a vector in a multidimensional space.

Turning a sentence into an element of a 4096-dimensional vector space Turning a sentence into an element of a 4096-dimensional vector space

Click Run embedding, and 8 floating-point numbers appear in the OUTPUT box. In the example above, those are only the first 8 coordinates of an element in a 4096-dimensional vector space.

Audio

You can select the Audio tool to transcribe audio input into text. First, go to the right-hand pane and select any model capable of transcription. Then, use the SOURCE box in the main pane to upload an audio file you wish to transcribe.

Box for uploading an audio file for transcription Box for uploading an audio file for transcription

In the example below, we have uploaded the second chapter of The Aurora Borealis in 1719 by Sidney Perley (1858-1928), from LibriVox.

Audio file is uploaded Audio file is uploaded

Click Run transcription to begin.

Audio file is transcribed Audio file is transcribed

The transcribed text appears in the TRANSCRIPTION box, below the SOURCE box.