Initial commit: Simple Transformer implementation

example.py

@@ -0,0 +1,52 @@
+import numpy as np
+from transformer import SimpleTransformer, create_padding_mask, create_look_ahead_mask
+
+
+def main():
+    vocab_size = 1000
+    d_model = 512
+    num_heads = 8
+    num_layers = 6
+    d_ff = 2048
+    max_seq_len = 100
+    
+    model = SimpleTransformer(vocab_size, d_model, num_heads, num_layers, d_ff, max_seq_len)
+    
+    batch_size = 2
+    seq_len = 10
+    x = np.random.randint(0, vocab_size, (batch_size, seq_len))
+    
+    print("=== Simple Transformer Example ===")
+    print(f"Vocabulary size: {vocab_size}")
+    print(f"Model dimension: {d_model}")
+    print(f"Number of heads: {num_heads}")
+    print(f"Number of layers: {num_layers}")
+    print(f"Feed-forward dimension: {d_ff}")
+    print(f"Max sequence length: {max_seq_len}")
+    print()
+    
+    print(f"Input shape: {x.shape}")
+    print(f"Input sample: {x[0]}")
+    print()
+    
+    output = model.forward(x)
+    
+    print(f"Output shape: {output.shape}")
+    print(f"Output sample (first 5 values): {output[0, 0, :5]}")
+    print()
+    
+    total_params = model.count_parameters()
+    print(f"Total parameters: {total_params:,}")
+    print()
+    
+    print("=== Attention Mask Examples ===")
+    padding_mask = create_padding_mask(x)
+    print(f"Padding mask shape: {padding_mask.shape}")
+    
+    look_ahead_mask = create_look_ahead_mask(seq_len)
+    print(f"Look-ahead mask shape: {look_ahead_mask.shape}")
+    print(f"Look-ahead mask sample:\n{look_ahead_mask[:5, :5]}")
+
+
+if __name__ == "__main__":
+    main()