import torch

# Data configuration
DATA_FILE=r'D:\Anudeep\Project\tenglish_dataset\telugu_asr\train-00000-of-00001-20a6e56929a59ecf.parquet'
TELUGUISH_COL = 'translit'
TELUGU_COL = 'text'

# Model hyperparameters
EMBEDDING_DIM = 256
HIDDEN_DIM = 512
GNN_LAYERS = 3
DROPOUT = 0.5
DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'

# Training configuration
LEARNING_RATE = 0.001
BATCH_SIZE = 16
NUM_EPOCHS = 5
MODEL_PATH = 'gnn_transliteration.pth'

# Special tokens
PAD_TOKEN = 0
SOS_TOKEN = 1 # Start of Sequence
EOS_TOKEN = 2 # End of Sequence