add proper generation wrapper

app.py

@@ -1,5 +1,9 @@
 import gradio as gr
 
+from transformers import GPT2LMHeadModel, GPT2Tokenizer
+tokenizer = GPT2Tokenizer.from_pretrained("sberbank-ai/mGPT")
+model = GPT2LMHeadModel.from_pretrained("sberbank-ai/mGPT")
+
 description = "Multilingual generation with mGPT"
 title = "Generate your own example"
 
@@ -11,12 +15,29 @@ article = (
     "</p>"
 )
 
+def generate(prompt: str):
+            input_ids = tokenizer.encode(prompt, return_tensors="pt").cuda(device)
+            out = model.generate(
+                    input_ids, 
+                    min_length=100, 
+                    max_length=200, 
+                    top_p=0.8, 
+                    top_k=0 
+                    no_repeat_ngram_size=5
+            )
+            generated_text = list(map(tokenizer.decode, out))[0]
+            return generated_text
+            
+
 interface = gr.Interface.load("huggingface/sberbank-ai/mGPT",
             description=description,
             examples=examples, 
+            fn=generate,
+            inputs="text",
+            outputs='text',
             thumbnail = 'https://habrastorage.org/r/w1560/getpro/habr/upload_files/26a/fa1/3e1/26afa13e1d1a56f54c7b0356761af7b8.png',
             theme = "peach",
             article = article
 )
 
-interface.launch()
+interface.launch(enable_queue=True)