{
  "experiment_name": "baseline",
  "dataset": {
    "training_file_path": "./data/train_processed.pickle",
    "validation_file_path": "./data/val_processed.pickle",
    "test_file_path": "./data/test_processed.pickle",
    "vocabulary_threshold": 2,
    "batch_size": 64,
    "num_workers": 8,
    "test_batch_size": 1
  },
  "experiment": {
    "num_epochs": 2,
    "learning_rate": 5e-4
  },
  "model": {
    "hidden_size": 512,
    "num_layers": 2,
    "embedding_size": 300,
    "model_type": "baseline"
  },
  "transformer": {
    "num_encoder_layers": 3,
    "num_decoder_layers": 3,
    "nhead": 8,
    "dim_feedforward": 512,
    "dropout": 0.2
  },
  "generation": {
    "max_length": 20,
    "deterministic": false,
    "temperature": 0.1
  }
}