Skip to main content

Deep neural network optimizer to make them faster, smaller, and energy-efficient from cloud to edge computing.

Project description

Build Status codecov

Neutrino Engine

Neutrino is a deep learning library for optimizing and accelerating deep neural networks to make them faster, smaller and more energy-efficient. Neural network designers can specify a variety of pre-trained models, datasets and target computation constraints and ask the engine to optimize the network. High-level APIs are provided to make the optimization process easy and transparent to the user. Neutrino can be biased to concentrate on compression (relative to disk size taken by the model) or latency (forward call’s execution time) optimization.

Community Release

Our community edition provides all the important features to experience the power and usability of model optimization with Neutrino. With the community version, engineers and researchers can verify the seamless integration of Neutrino into standard AI processes, test compatibility with existing model development and explore the benefits of optimization to various products. Feel free to use it as you please! The aim of the community edition is multifold, with examples such as:

  • Provide hands-on experience with automated model architecture optimization and see first-hand the possibilities with Deeplite Neutrino
  • Compare and complement the results obtained using Deeplite Neutrino with other open-source and industry model architecture optimization frameworks
  • Export an optimized model to test integration with endpoint applications
  • Verify the integration of Deeplite Neutrino within industry and production pipelines
  • Utilize Deeplite Neutrino to accelerate academic research, expedite results and share your achievements in research papers
  • Have fun! Users can play around with Deeplite Neutrino and enjoy the advantages of model architecture optimization in various use-cases

For detailed comparison of features on our community and production editions, refer to the documentation

Get Your Free Community License

The community license key is completely free-to-obtain and free-to-use. Fill out this simple form to obtain the license key for the Community Version of Deeplite Neutrino™.

Installation

Use pip to install neutrino-engine from PyPi repository. We recommend creating a new python virtualenv, then pip install using the following commands.

    pip install --upgrade pip
    pip install neutrino-engine
    pip install neutrino-torch

For other methods of installation and detailed instructions, refer to the documentation

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

neutrino_engine-5.3.3-cp39-cp39-manylinux2010_x86_64.whl (13.7 MB view hashes)

Uploaded CPython 3.9 manylinux: glibc 2.12+ x86-64

neutrino_engine-5.3.3-cp39-cp39-manylinux1_x86_64.whl (13.7 MB view hashes)

Uploaded CPython 3.9

neutrino_engine-5.3.3-cp38-cp38-manylinux2010_x86_64.whl (15.4 MB view hashes)

Uploaded CPython 3.8 manylinux: glibc 2.12+ x86-64

neutrino_engine-5.3.3-cp38-cp38-manylinux1_x86_64.whl (15.4 MB view hashes)

Uploaded CPython 3.8

neutrino_engine-5.3.3-cp37-cp37m-manylinux2010_x86_64.whl (12.3 MB view hashes)

Uploaded CPython 3.7m manylinux: glibc 2.12+ x86-64

neutrino_engine-5.3.3-cp37-cp37m-manylinux1_x86_64.whl (12.3 MB view hashes)

Uploaded CPython 3.7m

neutrino_engine-5.3.3-cp36-cp36m-manylinux2010_x86_64.whl (12.4 MB view hashes)

Uploaded CPython 3.6m manylinux: glibc 2.12+ x86-64

neutrino_engine-5.3.3-cp36-cp36m-manylinux1_x86_64.whl (12.4 MB view hashes)

Uploaded CPython 3.6m

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page