Nota is releasing the beta version of 'Model Search', a hardware-aware AutoML tool for ahead-of-the-curve developers and engineers, on April 20th. This new module will be made available under the startup’s flagship product NetsPresso, a hardware-aware AI optimization platform. NetsPresso consists of Model Search and two other modules, Compression Toolkit and Model Packaging, both of which are in the final stages of development at this time.

Overview of NetsPresso and the sub-modules:
Model Search, Compression Toolkit, and Model Packaging

Model Search (HW-Aware AutoML)

The module automatically finds the optimum model for a target device with only input of a dataset. It is easy-to use and has no coding required, where one can choose from many models according to their purpose such as accuracy vs latency trade offs etc., developing close production ready code in significantly less time.

Compression Toolkit (Ready-to-use toolkit)

Nota introduced the beta version of Compression Toolkit in late 2021 for expert deep learning engineers and AI developers with a requirement for compressing their AI models. It is a ready-to-use toolkit which supports all CNN models. The ready-to use toolkit takes only minutes to upload and compress any AI model based on CNN architectures, without the user having to worry about configurations or ratios. With this toolbox, users may easily configure what and how to compress, or they can rely on the Compression Toolkit to produce an appropriate design and do the task in seconds.

Model Packaging

The company is also planning to introduce its third module, Model Packaging, in July. The module enables engineers to accelerate and deploy a model on the hardware of their choice in seconds.

How AI developers and ML engineers with different needs may utilize NetsPresso

 

 
Find more information on Nota AI’s proprietary hardware-aware AI optimization platform,  NetsPresso, and its modules which are in beta testing and recruiting testers Model Search and Compression Toolkit.

If you have any questions, please feel free to reach out to the NetsPresso team at netspresso@nota.ai.