Skip to content

Latest commit

 

History

History
102 lines (72 loc) · 3.81 KB

README.md

File metadata and controls

102 lines (72 loc) · 3.81 KB

PyTorch Beginner Workshop

Introduction

These are the updated versions of notebooks used in the PyTorch Beginner Series YouTube playlist, created by Brad Heintz.

Outline

Title Video Notebook
1 Introduction to PyTorch + +
2 Introduction to PyTorch Tensors + +
3 The Fundamentals of Autograd + +
4 Building Models with PyTorch + +
5 PyTorch TensorBoard Support + +
6 Training with PyTorch + +
7 Model Understanding with Captum + +
8 Production Inference Deployment with PyTorch + +

Installation

Follow one of the methods below to set up everything and install all necessary dependencies.

Method 1: Manual Way

Installing Miniconda

  1. Install miniconda (or anaconda).
mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh

~/miniconda3/bin/conda init bash
conda config --set auto_activate_base false
  1. Change the solver to speed up the process of installing new packages.
conda update -n base conda
conda install -n base conda-libmamba-solver
conda config --set solver libmamba

conda config --add channels conda-forge

Setting up the Environment

  1. Create a new conda environment.
conda create -n ai -y
  1. Install essential packages.
conda install -c conda-forge -n ai jupyterlab numpy matplotlib -y
conda install -c pytorch -c nvidia -n ai pytorch torchvision torchaudio pytorch-cuda=12.1 -y

Replace 12.1 with your cuda version extracted from the nvidia-smi output.

  1. Install TensorBoard.
conda install -c conda-forge -n ai tensorboard -y

If you encounter any problems due to incompatibility with the latest NumPy version, run the following commands instead:

conda activate ai
pip3 install tb-nightly
  1. Install captum.
conda install -c pytorch -n ai captum -y
pip install --upgrade --quiet jupyter_client ipywidgets
conda install -c conda-forge -n ai flask flask-compress
  1. Export installed packages for further use.
conda activate ai
conda env export > environment.yml
pip3 freeze > requirements.txt

Method 2: Automatic Way

conda env create -f environment.yml
conda activate ai
pip3 install -r requirements.txt