Skip to content
master
Switch branches/tags
Code

Tutorials for the Natural language processing course (UL FRI)

This repository is a part of Natural Language Processing course at the University of Ljubljana, Faculty for computer and information science. Please contact slavko.zitnik@fri.uni-lj.si for any comments.

If you have an NVIDIA GPU, make sure NVIDIA drivers, CUDA and cuDNN are installed on your system and that versions match with PyTorch and Tensorflow..

Anaconda installation (OPTION A)

Conda environment management and usage:

# Creation of an environment (first time only)
conda create -n nlp-course-fri python=3.6

# Activation of an environment (before running examples)
source activate nlp-course-fri

# Dependencies installation (one time only)
conda install nb_conda==2.2.1 nltk==3.6.1 matplotlib==3.3.4 bs4==4.9.3 pandas==1.1.5 mpld3==0.5.2 python-crfsuite==0.9.7 h5py==2.10.0 pydot==1.4.1 graphviz==2.40.1 gensim==3.8.3 seaborn==0.11.1 
conda install -c huggingface transformers==4.4.2
conda install -c conda-forge ipywidgets==7.6.3
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit==11.0.221 -c pytorch
pip install tqdm seqeval tensorflow==2.4.1 keras==2.4.3
conda install -c anaconda scikit-learn==0.24.1

# Separately install crfsuite if above does not work
# Download from https://pypi.org/project/sklearn-crfsuite/#files or via pip if works
# This library is not only CRFSuite wrapper but also includes CRFSuite binaries
pip install sklearn_crfsuite-0.3.6-py2.py3-none-any.whl

# Explore and run notebooks
jupyter notebook 

# Or install and run jupyter lab
conda install -c conda-forge jupyterlab
jupyter lab

# Close environment
source deactivate

Show existing environments:

conda info --envs

Anaconda installation (OPTION B)

Create a new conda environment based on the provided environment.yml file:

# Creation of an environment (first time only)
conda env create -f environment.yml

# Activate environment
conda activate nlp-course-fri

The environment was successfully used within the following system: Ubuntu 20.04, CUDA 11.3, cuDNN 8.1.1.33-1+cuda11.2.

non-Anaconda environment

I propose that you us libraries listed above using Python 3 and virtualenv.

About

No description, website, or topics provided.

Resources

Releases

No releases published

Packages

No packages published