Skip to content

Latest commit

 

History

History
92 lines (68 loc) · 2.94 KB

cmds.md

File metadata and controls

92 lines (68 loc) · 2.94 KB
Raw

DammK ML / AI CMD pack

From stratch. Tested for recent course tasks ( CS5489 / CS5491 / CS6493 )

  • Must activate environment first!
  • Python 3.10 still too new
  • May need linux subsystem (WSL2) which may collides with Windows environment
  • Do not use install CMDs in jupyter environment! You will crash your computer!
  • Only use pip for last resort (e.g. mujoco-py)
  • Be familiar with minor software version difference (should be fine for educational use)

Computation environment

  • Development mode: (Notebook) ROG GM501GM, Win 10 21H2, WSL2, 6c12t i7-8750, 32GB DDR4 2666, GTX 1060 mobile, 500GB SATA SSD + 16GB M2 Optane
  • Production mode: (Workstation) Huananzhi X99-F8D, Win 10 21H2, no linux subsystem, 24c48t 2678v3 x2, 64GB DDR4 2133 ECC, GTX 1080 Ti x2, 500GB SATA SSD + 2TB SATA HDD

Video / IG links

Here's the CMDs

#!/bin/bash

# python=3.10 will lead to almost no avaliable packages!
conda create -n sklearn-env -c conda-forge scikit-learn python=3.9
# As in 220628, no need to open cmd manually
conda init powershell
# Activate before calling jupyter
conda env list
conda activate sklearn-env

# May be fine?
conda install -c conda-forge jupyterlab

# cs6493 tutorial 4
conda install -c conda-forge jupyterlab_widgets
conda install -c conda-forge ipywidgets

# Start jupyter
jupyter lab

# Prefer conda-forge with auto version (older version is not preferred)

# conda-forge failed on 3.9
# supressed by torchtext: https://github.com/pytorch/text
conda install -c pytorch pytorch
conda install -c pytorch torchvision
conda install -c pytorch torchaudio

# cs6493 tutorial 2
conda install -c pytorch torchtext
conda install -c conda-forge ray-tune

# 2.7.0 will have serious issue (cudnn issue)
# 2.6.0 has some weird bug (e.g. keras not found)
conda install -c conda-forge tensorflow-gpu=2.5.0

# cs5483 project
conda install -c esri tensorflow-addons

# Should be fine with almost no risk
conda install -c conda-forge pandas
conda install -c conda-forge matplotlib
conda install -c conda-forge scikit-image
conda install -c conda-forge scipy
conda install -c conda-forge networkx

# Optional: CS6493 asg1 (Testing only)
# transformers includes tokenizers
# conda install -c conda-forge tokenizers

# cs6493 tutorial 3
conda install -c conda-forge datasets

# cs6493 tutorial 4
# Overrided by QG task which requires exact 3.0.0
# conda install -c conda-forge transformers

# cs6493 tutorial 6
conda install -c conda-forge spacy==3.0.0

# cs6493 project (QG task)
# 3.0.2 has incapability with Python 3.9. Hence transformers must be newer then 4.0
conda install -c conda-forge transformers
conda install -c conda-forge nltk
pip install git+https://github.com/Maluuba/nlg-eval.git@master

# Extra (projects out of educational)
conda install -c conda-forge statsmodels