best_parameters.json

{
    "learning_rate": 0.001,
    "hidden_size": 256,
    "batch_size": 16,
    "num_epochs": 50
}