See Edge MLOps in Action Request a Demo
See Edge MLOps in Action Request a Demo
LEIP is an MLOps Software Development Kit (SDK) that can optimize and secure neural network runtimes for specific hardware targets. LEIP creates ultra-efficient and portable models optimized for compute, energy, and memory without requiring any changes to existing AI/ML infrastructure and frameworks. Customizable templates called Recipes are pre-qualified for a wide range of hardware targets and enable rapid prototyping of new models.
LEIP compresses conventional AI models below 8-bit integer precision but without losing result accuracy, creating up to a 10x reduction in size and a 3x improvement in inference. And with Latent AI’s Adaptive AI technology, the model can self-regulate its computational needs, only firing the parts of the neural network necessary to get the job done.
With LEIP, ML engineers develop optimized neural network runtimes for heterogeneous low power hardware (CPU/GPU/DSP). Our Latent-AI-Runtime-Engine (LRE) offers a modular micro-service software stack for secured inference processing, with support for CI/CD and asset tracking across the entire model life-cycle.
Latent AI takes the hard work out of design and deployment by exploring thousands of candidate recipes that are pre-configured for your hardware. With hundreds of possible target models and hardware already in place, LEIP Recipes makes it easy to find the best optimized configuration that meets your design requirements.
