Installation
Through PyPi Repository
Use pip
to install Neutrino engine from our internal PyPi repository. We recommend creating a new python virtualenv,
then pip install using the following commands, entering your credentials when prompted.
Important
We currently only support pip installation for linux environment or Windows Subsystem for Linux (WSL). Windows/mac installation is currently not supported.
Install Neutrino Engine
To install Neutrino engine:
$ pip install --upgrade pip
$ pip install neutrino-engine
Minimal Dependencies
Python 3.6+
Ubuntu>=16.04
neutrino-profiler==1.0.0
scipy==1.4.1
numpy==1.18.5
tensorly==0.4.5
pyyaml==5.3.1
onnx==1.7.0
ordered_set==4.0.2
licensing==0.26
grpcio==1.29.0
google-cloud-logging==1.15.1
pyAesCrypt==0.4.3
cryptography==3.4.6
Install Torch Backend
To install torch backend:
$ pip install --upgrade pip
$ pip install neutrino-torch
Minimal Dependencies:
neutrino-engine>=5.3.2
torch==1.4.0
ptflops==0.6.2
Augmentor==0.2.8”
(optional) Install deeplite-torch-zoo
Install deeplite-torch-zoo
to get access to our entire list of models and pretrained models, as follows
$ pip install --upgrade pip
$ pip install deeplite-torch-zoo
See deeplite-torch-zoo for additional installation details.
How to Update
It is recommended to update your package regularly. To update the package run:
$ pip install --upgrade neutrino-engine
$ pip install --upgrade neutrino-torch
$ pip install --upgrade deeplite-torch-zoo
How to Uninstall
To uninstall the package run:
$ pip uninstall neutrino-engine
$ pip uninstall neutrino-torch
$ pip uninstall deeplite-torch-zoo
Note
To install/update the requirements manually use --no-dependencies
flag in the commands above.
Otherwise, they will be installed/upgraded automatically.