Update Model.py

Model.py

@@ -103,7 +103,7 @@ def txt_stream(file_path, num_lines=None):
 
 # Dataset 생성 (예: 처음 10,000라인만)
 dataset = tf.data.Dataset.from_generator(
-    lambda: txt_stream(DATA_PATH, num_lines=10000),
+    lambda: txt_stream(DATA_PATH, num_lines=100000),
     output_signature=(
         tf.TensorSpec(shape=(max_len,), dtype=tf.int32),
         tf.TensorSpec(shape=(max_len,), dtype=tf.int32),
@@ -129,7 +129,7 @@ class SwiGLU(layers.Layer):
         return tf.cast(out, x.dtype)
 
 class SparseCausalAttention(tf.keras.layers.Layer):
-    def __init__(self, num_heads, head_dim, window_size=16, **kwargs):
+    def __init__(self, num_heads, head_dim, window_size=8, **kwargs):
         super().__init__(**kwargs)
         self.num_heads = num_heads
         self.head_dim = head_dim