Update handler.py

handler.py

@@ -50,7 +50,7 @@ class EndpointHandler():
 
         with torch.no_grad():
             out = self.model.generate(**processed, 
-                                     max_new_tokens=50,
+                                     max_new_tokens=20,
                                      temperature = 0.5,
                                      do_sample=True,
                                      top_k=50,