Skip to content

Latest commit

 

History

History
41 lines (27 loc) · 939 Bytes

File metadata and controls

41 lines (27 loc) · 939 Bytes

Setup CUDA, CUDNN, Tensorrt

https://github.com/SuperSecureHuman/Resources/blob/main/AI-ML-DL/Nvidia-Setup.md

Note that CUDA, and CUDNN are needed for any sort of GPU accelerated compute.

Tensorrt is used for optimizing the model for inference.

Conda

Go to conda download site, and download miniconda (miniconda is stripped version of anaconda, and has lesser default packages)

Preffered python version - 3.8

Python Packages needed

Use the pip versions of all these packages even if you have conda installed

juypter
jupyterthemes # In case you hate the blinding white of jupyter notebook
pytorch (https://pytorch.org/get-started/locally/) (Download the GPU version, pip)
tensorflow-gpu
transformers
datasets
sentence-transformers

# These packages usually have the dependencies as the regurarly used stuff
numpy, pandas, matplotlib etcc....

Not currently used, but may use later

jax
numba
cupy
pycuda