reduce batch size

generate.py

@@ -30,7 +30,7 @@ if torch.backends.mps.is_available():
 else:
     device = "cuda"
     model_id = "google/gemma-2b-it"
-    batch_size = 20
+    batch_size = 10
 
 model = models.transformers(model_id, device=device)