Configuration

Development Configuration (config.py)

Assumed Folder Structure

When using config.py, the project expects a specific directory structure relative to the root directory (dev):

dev/
│── NeuralLib/              # Library source code (contains `config.py`)
│── data/                   # Directory where datasets are stored
│── results/                # Stores model training logs, evaluation results, and checkpoints
│── hugging_prodmodels/     # Stores trained models locally before uploading to HF

Defined Paths

• DATA_BASE_DIR: Points to dev/data/, where datasets are stored.
• RESULTS_BASE_DIR: Points to dev/results/, where model training outputs are saved. If it does not exist, it is automatically created at runtime.
• HUGGING_MODELS_BASE_DIR: Points to dev/hugging_prodmodels/, where production models are stored before uploading.