Update app.py

app.py

@@ -2,11 +2,13 @@ import os
 os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "0"
 
 import torch
-import gradio as gr
+from fastapi import FastAPI
+from pydantic import BaseModel
 from transformers import AutoTokenizer, AutoModelForCausalLM
+import gradio as gr
 
+# --- Модель ---
 model_id = "sberbank-ai/rugpt3medium_based_on_gpt2"
-
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(model_id)
 
@@ -33,28 +35,28 @@ def respond(message, history=None):
             pad_token_id=tokenizer.eos_token_id
         )
 
-    full_output = tokenizer.decode(output_ids[0], skip_special_tokens=True)
-
-    if "Ответ:" in full_output:
-        answer = full_output.split("Ответ:")[-1].strip()
+    output = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+    if "Ответ:" in output:
+        answer = output.split("Ответ:")[-1].strip()
     else:
-        answer = full_output[len(prompt):].strip()
-
+        answer = output[len(prompt):].strip()
     return answer
 
-### API маршрут
-@gr.route("/ask", methods=["POST"])
-def ask_api(question: str):
-    return {"answer": respond(question)}
-
-### Интерфейс с согласованным type="messages"
-chat = gr.ChatInterface(
-    fn=respond,
-    chatbot=gr.Chatbot(label="Диалог", type="messages"),
-    textbox=gr.Textbox(placeholder="Задай вопрос...", label="Твой вопрос"),
-    title="Бот об Университете Иннополис (на русском)",
-    type="messages"  # <-- добавлено!
-)
+# --- Gradio UI ---
+chat = gr.ChatInterface(fn=respond, title="Иннополис Бот")
+
+# --- FastAPI для API доступа ---
+app = FastAPI()
+
+class QuestionRequest(BaseModel):
+    question: str
+
+@app.post("/ask")
+def ask(request: QuestionRequest):
+    answer = respond(request.question)
+    return {"answer": answer}
+
+# --- Подключаем FastAPI к Gradio ---
+gr.mount_gradio_app(app, chat, path="/")
 
-if __name__ == "__main__":
-    chat.launch()
+# --- Не нужно писать iface.launch()! Hugging Face сам запустит Uvicorn ---