Update app.py

app.py

@@ -7,6 +7,19 @@ import tempfile
 from PIL import Image, ImageDraw
 import re # Import thư viện regular expression
 
+# --- Monkey-Patch to Force CPU Execution ---
+# ADD THIS BLOCK HERE
+import torch
+print("Applying monkey-patch to force CPU execution...")
+original_cuda = torch.Tensor.cuda
+def force_cpu_cuda(self, device=None, non_blocking=False, **kwargs):
+    # print(f"Intercepted .cuda() call. Forcing tensor to remain on CPU.")
+    return self.to("cpu") # Always move/keep on CPU
+
+torch.Tensor.cuda = force_cpu_cuda
+# --- End Monkey-Patch ---
+
+
 # --- 1. Load Model and Tokenizer (Done only once at startup) ---
 print("Loading model and tokenizer...")
 model_name = "deepseek-ai/DeepSeek-OCR"