Inference as a Service¶

This guide will walk you through the various services available, focusing on services about model inference. Below you will find details on each function categorized for ease of understanding.

Overview¶

Inference as a Service offers a collection of tools that make it easy to use pre-trained models for real-time predictions. Users can explore models, select those that fit their needs, and seamlessly deploy them into production.

This category consists of the following services:

Playground¶

The Playground offers a user-friendly interface for experimenting with various pre-trained models.

With the Playground, users can:

• Interactively test different models on their own data.
• Change Model Parameters in real-time in the UI.
• Get started without coding, making it accessible for non-developers and AI enthusiasts.

Model Market¶

The Model Market is a curated collection of pre-trained or public domain models available for use privately.

Users can:

• Browse and select provided models.
• Deploy a provided model on new resources.

Model Service¶

The Model Service allows users to deploy either public models or private models as endpoints for production.

With the Model Service, you can:

• Deploy models as APIs for real-time inference.
• Scale inference based on demand with GPU-backed infrastructure.
• Monitor model performance and view usage metrics to ensure reliability and efficiency.

Next Step¶

The GPU Cloud Platform provides powerful tools for both inference and training. Whether you're looking to get started with pre-trained models or train your own, the platform offers the infrastructure and services you need for AI development.