tokens = tokenize(text)

x = embedding(tokens)

x = x + positional_encoding

attention_output = multi_head_attention(x)

x = layer_norm(x + attention_output)

ffn_output = feed_forward(x)

output = layer_norm(x + ffn_output)