Tiger shark - Platform that converts deep learning models into highly optimized binaries

 

Tiger shark is a new platform that converts deep learning models into highly optimized binaries that run faster on a wide range of processors.

  

Tiger shark is made by Inteon, an Intel emerging growth and incubation venture.

  

It is said that it achieves 25 to 400 % performance Improvement compared to the same models that were manually optimized by humans.

  

Thus, Tiger Shark reduces response time and decreases the cloud bill.

  

The input to tiger shark Is a trained deep learning model that includes the model inputs and outputs, a data set for validation, as well as the target processor or cloud instance you'd like to model to run on.

  

Tiger shark then performs automated steps and imports the model and converts it to an internal representation. Then it performs a deep analysis and generates code that is optimized for the particular Target processor.

  

Tiger shark is not limited to Intel processors. It works on a wide range of CPUs and GPUs. Is compatible with all leading Cloud providers and can also generate code for Edge devices.

  

This technology is said to be in its early stage and hopefully we will hear more about it in the upcoming days.