Welcome to TheStage AI documentation!

Hello and welcome to TheStage AI documentation! We are providing infrenence acceration stack for AI models. Its based on our mathematical framework which allows to control models’ accuracy and performance trade-off with a slider movement. Resulted models can be compiled for execution on Nvidia GPUs and Apple Silicon devices.

We are providing different components to build your own AI stack:

• TheStage AI Platform:

  • A platform to run your AI models on self-hosted instances.

  • Provides a web interface and CLI to manage your projects, instances, and models.

• TheStage CLI:

  • A command-line interface to interact with TheStage AI Platform.

• Elastic Models:

  • A library with pre-compiled models with 4 perfomance tiers: XL, L, M, S.

  • Allows to use models with different performance and accuracy trade-offs.

  • Allows to serve desired models in a single line of code on your machine or in the cloud.

• Qlip:

  • A full-stack AI framework to build, train, and deploy AI models on-prem.

  • Compile models for execution on Nvidia GPUs and Apple Silicon devices.

  • Automatically apply quantization and pruning algorithms to optimize models for performance and control that with a slider.

  • Serve resulted models with a convenient API.