clerencemathonsi/semantic-relatedness-african-languages

# COS802 Semantic Relatedness This repository contains code and resources for experiments on semantic relatedness, including data loading, model training, evaluation, and analysis. ## Project Structure - `analyze_results.py` — Analyze experiment results - `check_data.py` — Data validation and checks - `run_experiment.py` — Main script to run experiments - `test_setup.py` — Test environment setup - `requirements.txt` — Python dependencies - `configs/` — Configuration files - `notebooks/` — Jupyter notebooks for exploration - `results/` — Experiment results and model checkpoints - `SemRel2024/` — Data for various languages - `src/` — Source code - `data_loader.py` — Data loading utilities - `evaluation.py` — Evaluation metrics and logic - `models.py` — Model definitions - `training.py` — Training routines ## Getting Started 1. **Install dependencies:** ```bash pip install -r requirements.txt ``` 2. **Run experiments:** ```bash python run_experiment.py ``` 3. **Analyze results:** ```bash python analyze_results.py ``` ## Data The `SemRel2024/` folder contains datasets for multiple languages. Each subfolder corresponds to a language (e.g., `eng`, `hau`, `kin`). ## Results Model checkpoints and results are stored in the `results/` directory, organized by experiment and checkpoint. ## Notebooks Exploratory analysis and visualizations can be found in `notebooks/exploration.ipynb`. ## License This project is for academic use. Please cite appropriately if used in research.