0

app.py

@@ -26,8 +26,8 @@ def load_pipeline():
             torch_dtype=torch.float16,
             variant="fp16",
             use_safetensors=True,
-            device_map="auto",
-			attn_implementation="eager"
+            device_map="balanced",
+	    attn_implementation="eager"
         )
         
         # Enable memory optimizations