anzeyimana/DeepKIN

DeepKIN -- A deep learning toolkit for Kinyarwanda NLP. # DeepKIN A deep learning toolkit for Kinyarwanda NLP. ## Getting started This toolkit depends on a Morphological analyzer/generator for Kinyarwanda and related models. In order to use the toolkit, you need to go through the following steps: 1. Download and install the morphological analyzer/generator 2. Get a free license for the morphological analyzer/generator 2. Clone and install DeepKIN toolkit and its dependencies 3. Run experimental code ## Downloading and installing Kinyarwanda morphological analyzer/generator The current release (version 0.1.0) of the morphological analyzer is only compatible with Linux x84_64 platform. In order to run the morphological analyzer/generator, the system must meet the following minimum requirements: - Operating system: Linux x86_64 (amd64), we have tested it with Ubuntu 64-bit, 18.04, 20.04 and 22.04 versions - Drive space: 45 GB, (64 GB recommended) - System memory (RAM): 40 GB (64 GB recommended) The morphological analyzer is available for download from the following Google Drive link: drive.google.com With the link, you can download it directly to your machine. In order to download it from a terminal (i.e. on a remote server), you need to use an OAuth token as in the following steps: 1. Go to OAuth 2.0 Playground developers.google.com 2. In the Select the Scope box, paste googleapis.com 3. Click Authorize APIs and then Exchange authorization code for tokens 4. Copy the Access token 5. Run the following command in terminal, where ACCESS_TOKEN is the access token copied above: ``` curl -H "Authorization: Bearer ACCESS_TOKEN" googleapis.com -o KINLP.tar.gz ``` The morphological analyzer/generator package installation directory needs to be refered as `KINLP_HOME` environmental variable or be installed in `/opt/KINLP` path as follow: ``` gunzip …