Introducing

Deep Sparse

Deep Learning Model Optimization Tools & CPU Inference Engine

Sparsify and quantize your machine learning models to minimize footprint & run on CPUs at GPU speeds.
See the Code
graph ResNet50 - milliseconds/image Neural Magic GPU CPU
Need a crash course on Neural Magic?
Learn about our community edition software components.
Save Your Seat
Sparsify is open sourced. Optimize your model today!
Learn More

ComponentsCommunity Edition

Sparsify
Open-source, easy-to-use interface to automatically sparsify and quantize deep learning models for CPUs & GPUs.
See the Code
Install
SparseML
Open-source libraries and optimization algorithms for CPUs & GPUs, enabling integration with a few lines of code.
See the Code
Install
SparseZoo
Open-source neural network model repository for highly sparse and sparse-quantized models with matching pruning recipes for CPUs and GPUs.
See Models
Install
DeepSparse Engine
Free CPU runtime that runs sparse models at GPU speeds.
Learn More
Install

Paths to Sparse Acceleration

path-to-acceleration Sparse Model Sparse Model DeepSparse Engine SparseZoo Transfer Learning SparseML Sparsify Your Dense Library Sparse Model Dense Model Sparse Model Dense Model DeepSparse Engine Your Training Process Your Dense Library Dense Model Dense Model DeepSparse Engine Your Training Process
A.) Original Dense Path
Take your dense model & run it in the DeepSparse Engine, without any changes.
B.) SparseZoo Path
Take a pre-optimized model & run it in the DeepSparse Engine, or transfer learn with your data.
C.) Sparsified Path
Sparsify and quantize your dense model with ease & run it in the DeepSparse Engine.

Benefits

Unprecedented Performance –– Run models on CPUs at GPU speeds. No special hardware required.
Reduce Costs –– Deploy and scale models on cheaper and more commonly available CPU servers.
Smaller Footprint –– Unlock unlimited production possibilities by reducing model footprint by 20x.
Reduce Power –– Save on energy at the edge with unique in-cache execution technology.
Run Anywhere –– Run on premise, in the cloud, or at the edge, with both CPUs and GPUs.
Easy to Use –– Eliminate trial and error. Analyze, sparsify, and fine-tune models with a few clicks.

Choose a Better way to Optimize & Run Deep Learning Models

See the Code