Fast Device Agnostic AI model tuning and inference

Transforming models to a faster version
Listing Category
Short Description
Machine learning models, especially deep neural networks are tend to be computationally heavy. To reduce these costs it is possible to optimize the performance of the networks. A deep learning compiler can provide over 10x acceleration.

AnyCores is a vertically stacked ML model accelerator. Its provisioned goal is to provide a unified solution to compile any ML model to any hardware, whether it is on the edge or on the server. The system utilizes AI itself to compile the neural networks, instead of writing several hundreds of kernel versions manually, which is unscalable.

As a proof-of-concept, a series of accelerated networks will be released in the next 3 months.
By the end of the year, a restricted access will be provided to our accelerator platform.

Scroll to Top