TMLS Workshop 2023

This repository contains the necesary files for the workshop.

If you wish to run locally, you will need:

• Pytorch >=2.0.0
• Nvidia Apex package (https://github.com/NVIDIA/apex)
• Deepview Profile (https://pypi.org/project/deepview-profile/)
• Deepview Explore (https://marketplace.visualstudio.com/items?itemName=CentML.deepview-explore)

Follow these steps:

1. Create a python virtual environment and activate it (recommended)
   - python3 -m virtualenv ./env
   - source env/bin/activate
2. pip install torch deepview-profile numpy packaging
3. For Apex:
    - git clone https://github.com/NVIDIA/apex
    - cd apex
    - pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./

Alternative installation

1. clone this repository
2. cd tmls-workshop-2023
3. bash setup.sh

Notes:

Independently of the installation method, for Apex you will need to install local cuda libraries. You can follow these instructions