Preparatory Lab for NTUA Speech and Language Processing course.
The project requires Python 3.
You can use virtualenv
but we recommend that you use conda
.
Download the appropriate Miniconda version for your system. Then follow the installation instructions.
Follow the instructions from the PyTorch home page: https://pytorch.org/
pip install -r slp-lab3-prep/requirements.txt
In order to minimize the memory requirements you can use low dimensional word embeddings, such as the 50d Glove embeddings. However, if your computer has enough RAM you will get better results with higher dimensional embeddings.
-
Glove 6B: Generic english word embeddings - 50d, 100d, 200d, & 300d vectors.
-
Glove Twitter: Twitter specific word embeddings - 25d, 50d, 100d, & 200d vectors
-
fastText: Generic english word embeddings - only 300d vectors.
The project expects the file(s) to be in the /embeddings
folder.