Welcome to LEIP

Tip

Reading this page entirely from top to bottom is recommended.

Hello! Welcome to the Latent AI Edge Inference Platform (LEIP).

This home page will serve to help you intuitively understand how to navigate and leverage LEIP to rapidly create highly performant AI models for the edge continuum.

If you would like to schedule an evaluation of LEIP, please contact us at support@latentai.com.

Before You Begin…

To keep things simple, LEIP is broken down into 3 products. You can use each of these products based on what tools you need to deploy your Machine Learning (ML)model.

1. Design
   • Easily create and train an ML model by Bringing Your Own Data (BYOD). This can be done by selecting from our curated Golden Recipes, or create a recipe through Recipe Designer.
2. Optimize
   • Easily apply software and hardware optimizations to your model for a chosen hardware deployment. No specific hardware optimization knowledge is needed.
   • Users can take advantage of LEIP Optimize whether they created a model from LEIP Design, or if they have created one outside of LEIP (this path is also known as Bring Your Own Model (BYOM)).
3. Deploy
   • Install our Latent Runtime Engine (LRE) to manage and run your models created from LEIP Optimize.
   • Utilize LRE Services to enable your model to report metrics and enhance security.

Getting Started: Identifying Your Path

There are 2 paths to enter into LEIP. Your path will depend on which LEIP Products you will use to create your AI solution. The steps for each path are listed in order below.

If you’re more of a visual person, click this graph to see a high level overview of the paths.

Bring Your Own Data (BYOD)

1. Pre-requisites
   • A labeled dataset in COCO or PASCAL VOC format.
   • Don’t know how to create a labeled dataset? Click here.
2. Design
   • Create and train a model here, with the end result being a trained (and traced PyTorch) model.
3. Optimize
   • Use the trained model from Design in the optimization phase to quantize and target deployment hardware of your choosing.
4. Deploy
   • Install our Latent Runtime Engine (LRE) to manage and run your models created from LEIP Optimize.

Bring Your Own Model (BYOM)

1. Pre-requisites
   • A trained model from a popular ML framework (e.g., TensorFlow, Keras, PyTorch, etc.).
2. Optimize
   • Use your trained model in the optimization phase to quantize and target deployment hardware of your choosing.
3. Deploy
   • Install our Latent Runtime Engine (LRE) to manage and run your models created from LEIP Optimize.

Installing LEIP SDK & LRE

Note

The LEIP SDK consists of Optimize and Deploy. This is installed on your development machine.

For deployment, you will install the LRE on your deployment hardware. The LRE is responsible for running the optimized model created from the SDK.

Once you’ve identified your LEIP path, check to see if you have all the prerequisites to run LEIP by visiting Installing Prerequisites.

Once you have satisfied all prerequisites, refer to Installing LEIP SDK.

After installation, you can browse and use our tutorials below.

Once you have a model you’re ready to deploy with, you can reference Installing LRE to install the runtime and deploy your model on your target hardware.

Tutorials

Tutorials will be labeled withBYOM orBYOD depending on which path you choose.

• Getting Started with BYOD to Deployment