Tuned Model Architecture

silverlightning926 · Jun 13, 2024 · e27d268 · e27d268
1 parent 66c0070
commit e27d268
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 7 deletions.
diff --git a/src/build_model.py b/src/build_model.py
@@ -6,16 +6,15 @@ def build_model():
     model = Sequential([
         Input(shape=(8, 8, 12)),
         Conv2D(64, (2, 2), activation='relu'),
-        MaxPooling2D((2, 2)),
         Conv2D(64, (2, 2), activation='relu'),
-        MaxPooling2D((2, 2)),
+        Conv2D(64, (2, 2), activation='relu'),
         Flatten(),
         Dense(128, activation='relu'),
         Dropout(0.2),
-        Dense(3, activation='softmax')
+        Dense(1, activation='tanh')
     ])
 
     model.compile(optimizer='adam',
-                  loss='sparse_categorical_crossentropy', metrics=['accuracy'])
+                  loss='mean_squared_error', metrics=['accuracy'])
 
     return model
diff --git a/src/load_dataset.py b/src/load_dataset.py
@@ -64,11 +64,11 @@ def _generateBoards():
     def processRow(row):
         winner = row['winner']
         if winner == 'white':
-            winner = 2
+            winner = 1
         elif winner == 'black':
-            winner = 0
+            winner = -1
         else:
-            winner = 1
+            winner = 0
 
         board = chess.Board()
         for move in row['moves'].split():