diff --git a/chectus_net/run.py b/chectus_net/run.py
index 33d7d09..14fdab5 100644
--- a/chectus_net/run.py
+++ b/chectus_net/run.py
@@ -2,8 +2,10 @@ import torch
 from model import ChessPredictModelS
 from fen2vec import parse_fen
 
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
 def run(model_path='best_model.pth'):
-	model = ChessPredictModelS()
+	model = ChessPredictModelS().half().to(device)
 	model.load_state_dict(torch.load(model_path, weights_only=True))
 	model.eval()
 
@@ -13,12 +15,11 @@ def run(model_path='best_model.pth'):
 			break
 
 		try:
-			input_tensor = parse_fen(fen_str)
-			input_tensor = input_tensor.unsqueeze(0)  # add batch dim
+			input_tensor, _wdl = parse_fen(fen_str, 0)
+			input_tensor = input_tensor.unsqueeze(0).to(device)  # add batch dim
 
 			with torch.no_grad():
-				output = model(input_tensor)				
-				transformed_output = torch.atanh(output) * 10
+				output = model(input_tensor)
 				print(f"$= {transformed_output[0, 0]} {output[0, 0]}")
 		except Exception as e:
 			print(f"Error: {e}")
diff --git a/chectus_net/train.py b/chectus_net/train.py
index 3f91191..516855e 100644
--- a/chectus_net/train.py
+++ b/chectus_net/train.py
@@ -63,11 +63,11 @@ def train(model, criterion, optimizer, train_loader, val_loader, num_epochs=10,
 				break
 
 if __name__ == "__main__":
-	batch_size = 32
+	batch_size = 256
 	num_epochs = 50
 	learning_rate = 0.001
 	patience = 2
-	model_path = 'best_model_baby.pth'
+	model_path = 'best_model.pth'
 	weight_decay = 0
 
 	print('Loading Data')
@@ -75,7 +75,7 @@ if __name__ == "__main__":
 	val_loader = create_dataloader(table_name='test', batch_size=batch_size, shuffle=False, num_workers=3)
 	print('Loaded Data')
 
-	model = ChessPredictModelBaby().half().to(device)
+	model = ChessPredictModelS().half().to(device)
 	criterion = nn.SmoothL1Loss()
 	optimizer = optim.Adam(model.parameters(), lr=learning_rate, weight_decay=weight_decay)
 
diff --git a/lichessdb/prepare.sh b/lichessdb/prepare.sh
old mode 100644
new mode 100755