Lamdastack Quick Introduction/ FAQ

post by sidop6 on Sep 6, 2018

Hi,

just started using your Lambda stack. Seems like a good way to start training our models. However I have certain queries:

1. is there an internal guide for getting started.

2. Also which all packages have been installed and there install locations. Need to know as it will help avoid installing redundant packages. The information on the web page is too sparse.

3. Removing undesired packages or the entire stack.

Thanks

Sid

post by sabalaba on Sep 7, 2018

Hi Sid,

glad you’re enjoying Lambda Stack.

1. There currently is not a guide to using Lambda Stack. To get started, just boot up a python REPL and go:

$ python

You’ll note that with a Lambda Stack install, TensorFlow, Torch, and Caffe are all installed with full GPU support.

TensorFlow

>>> import tensorflow as tf
>>> s = tf.Session()

PyTorch

>>> import torch

Caffe

>>> import caffe

2. To list the installed packages you can use the apt-cache depends command. Unless you feel very confident in your linux system administration abilities, I would recommend you keep all of the installed packages. That said, here’s a semi-complete list. You’ll definitely want to keep all of the hard dependencies of each package you want. Simply type sudo apt-cache depends <package_name>. You’ll then know what packages you can’t remove.

$ sudo apt-cache depends lambda-stack-cuda
lambda-stack-cuda
 |Depends: libnvidia-compute-390
  Depends: <libcuda-9.0-1>
    libnvidia-compute-390
  Depends: python-torchvision
  Depends: python3-torchvision
  Depends: python-keras
  Depends: python3-keras
  Depends: python-theano
  Depends: python3-theano
  Depends: python-sklearn
  Depends: python3-sklearn
  Depends: python-pandas
  Depends: python3-pandas
  Depends: python-opencv
  Depends: python3-opencv
  Depends: tensorboard
  Depends: nvidia-cuda-toolkit
  Depends: libcudnn-dev
  Depends: caffe-cuda
  Depends: python-caffe2-cuda
  Depends: python3-caffe2-cuda
  Depends: python-torch-cuda
  Depends: python3-torch-cuda
  Depends: python-tensorflow-cuda
  Depends: python3-tensorflow-cuda
  Depends: tensorflow-tools-cuda
  Depends: python-pygpu
  Depends: python3-pygpu
  Depends: python-pycuda
  Depends: python3-pycuda
  Depends: python-skcuda
  Depends: python3-skcuda
  Recommends: lambda-desktop
  Recommends: nvidia-driver-390
$ sudo apt-cache depends lambda-desktop
lambda-desktop
  Depends: git
    git:i386
  Depends: vim
    vim-athena
    vim-gtk
    vim-gtk3
    vim-nox
  Depends: curl
    curl:i386
  Depends: emacs
  Depends: htop
  Depends: tmux
  Depends: screen
  Depends: valgrind
  Depends: python-all
  Depends: python3-all
  Depends: python-pip
  Depends: python3-pip
  Depends: virtualenv
  Depends: ipython
  Depends: ipython3
  Depends: jupyter
  Depends: build-essential
  Depends: cmake
    cmake:i386
  Depends: gfortran
  Depends: flake8
  Depends: fasd
  Depends: lambda-wallpapers
  Depends: lambda-settings
  Depends: chromium-browser
  Depends: grace
  Depends: feh
  Depends: grub-theme-lambda
  Recommends: ubuntu-desktop

Hopefully that helps you out.