upd

app.py

@@ -38,8 +38,9 @@ def predict(message, history):
         num_beams=1,
         stopping_criteria=StoppingCriteriaList([stop])
         )
-    t = Thread(target=model.generate, kwargs=generate_kwargs)
-    t.start()
+    #t = Thread(target=model.generate, kwargs=generate_kwargs)
+    #t.start()
+    model.generate(**generate_kwargs)
 
     #Initialize an empty string to store the generated text
     partial_message  = ""