Deeplite, a provider of AI software designed to make other AI models faster, is introducing Deeplite Neutrino Community version, offering a hands-on introduction while also enabling new connections and knowledge exchange among community members from different commercial, research, and academic environments.
“The network edge is critical—it’s where users interact with devices and applications, businesses connect with customers, and the data to drive strategy and operations is generated. And while businesses want to push their AI software to the edge, the resource limitations of edge devices are holding them back,” said Nick Romano, CEO and co-founder at Deeplite. “Rather than spending time and money to try building optimization software on their own, our free Neutrino Community version lets deep learning engineers and researchers download our software to immediately test optimizing their models. This will accelerate more AI on the edge and move our mission of ‘AI for Everyday Life’ forward.”
Community users will benefit from sharing feedback and ideas using Deeplite Neutrino on Github, while optimizing their deep learning models for memory, compute, power, and other resource constraints on cameras, drones, smartphones, and other network edge devices.
Deeplite created Neutrino, an intelligent optimization engine for Deep Neural Networks (DNNs) deployed on cloud servers, where increased throughput can save money, and edge devices where size, speed, and power are often major challenges.
With Neutrino, AI experts automatically optimize high-performance DNN models for these resource constraints. Neutrino inputs large, initial DNN models that have been trained for a specific use case and understands the edge device constraints to deliver smaller, more efficient, and accurate models.
Neutrino Community is designed for deep learning engineers and teams at both startups and larger corporations, as well as industry researchers and academics—anyone looking to test the benefits of optimization on their AI models.
The Community will allow members to accelerate and enhance their current AI research and development projects by testing out the impact of optimizing their AI models.
Deeplite provides various examples of how to optimize different applications and use cases for deep learning models such as image classification, object detection and semantic segmentation.
The Community Version of Deeplite Neutrino also includes access to two key resources:
- Deeplite Torch Zoo – a collection of popular DNN model architectures and benchmark datasets for PyTorch framework. Its pre-trained models can be used as a starting point for optimizing model architectures using Neutrino.
- Deeplite Profiler – a tool to measure the performance of a deep learning model easily and effectively in both PyTorch and TensorFlow 1.x frameworks. Members can utilize existing metrics or create their own custom metrics. In addition, performance can be compared between two deep learning models – for example, a teacher and student model.
For more information about this news, visit www.deeplite.ai.