Overview

Overview

The AI Playground Chat App is an interactive tool that allows users to engage with AI models and pipelines using text prompts and image uploads. This feature enables seamless testing and refinement of AI workflows within the Dataloop platform.

Key Features

• AI Integration – Connects effortlessly with Dataloop’s AI models and pipelines.
• Real-Time Streaming – Provides instant AI-generated responses via Server-Sent Events (SSE).
• File Upload Support – Enhances interactions by allowing image uploads.
• Session Management – Stores chat history and user sessions within Dataloop datasets.

Accessing AI Playground

To access AI Playground:

1. Click AI Playground from the left-side menu. The empty state page will be displayed.
2. Click Explore Marketplace to install AI models and pipelines.

Explore Marketplace

The Marketplace provides access to:

• Models – AI models designed for interactive responses to text, images, and audio.
• Pipelines – Pre-built AI workflows for managing prompt-based tasks.

To use models and pipelines effectively:

• Models must be deployed before interaction.
• Pipelines must be active to execute workflows.

Refer to the following articles for installation guidance:

• Install Model
• Install a Pipeline from a template

Recommended Pipelines and Models

To maximize AI Playground performance, Dataloop recommends using RAG-based LLM Pipelines and Prompt-based models.

The AI Playground supports a diverse range of Vision-Language Models (VLMs) and Large Language Models (LLMs), enabling seamless integration of any AI model that processes prompt-based inputs, whether text, images, or a combination of both.

Ask Dataloop AI

Once a model is deployed or a pipeline is activated, the Ask Dataloop AI chat interface becomes available for interaction.

Using a pipeline for AI responses

1. In AI Playground, select a Pipeline from the list.

3. In the Ask Dataloop AI chat box, enter your query or upload an image.
4. Press Enter or click the up-arrow icon to submit. The Pipeline Execution Status will be displayed.
5. Once completed, an inline AI response will appear.

Using a model for AI responses

1. In AI Playground, select a Model from the list.

2. By default, Pipeline Mode is selected; switch to Model Mode if needed.
3. Enter your query in the Ask Dataloop AI chat box or upload an image.
4. Press Enter or click the up-arrow icon to submit. The Model Execution Status will be displayed.
5. Once completed, an inline AI response will appear.

Viewing Chat History & Sessions

All chat interactions are stored in Dataloop datasets (e.g., ai-playground-history).

Chat history is available in JSON format for further analysis.