Update app.py

app.py

@@ -7,45 +7,79 @@ logger = logging.getLogger(__name__)
 
 app = Flask(__name__)
 
-# Используем pipeline для простоты
-chatbot = pipeline(
-    "text-generation",
-    model="microsoft/DialoGPT-medium",
-    torch_dtype="auto",
-    device_map="auto"
-)
+# Инициализация pipeline с русскоязычной моделью
+try:
+    chatbot = pipeline(
+        "text-generation",
+        model="sberbank-ai/rugpt3medium_based_on_gpt2",  # Быстрая русская модель
+        torch_dtype="auto",
+        device_map="auto",
+        tokenizer="sberbank-ai/rugpt3medium_based_on_gpt2"
+    )
+    logger.info("Russian model loaded successfully")
+except Exception as e:
+    logger.error(f"Error loading Russian model: {e}")
+    # Fallback to English model
+    chatbot = pipeline(
+        "text-generation",
+        model="microsoft/DialoGPT-medium",
+        torch_dtype="auto",
+        device_map="auto"
+    )
+    logger.info("Fallback English model loaded")
+
+@app.route('/health', methods=['GET'])
+def health_check():
+    return jsonify({"status": "healthy", "message": "Service is running"})
 
 @app.route('/chat', methods=['POST'])
 def chat():
     try:
         data = request.get_json()
-        message = data.get('message', '')
         
-        if not message:
-            return jsonify({"error": "No message provided"}), 400
+        if not data or 'message' not in data:
+            return jsonify({"error": "Missing 'message' in request"}), 400
+        
+        user_message = data['message']
+        logger.info(f"Received message: {user_message}")
+        
+        # Генерация ответа с промптом на русском
+        prompt = f"Пользователь: {user_message}\nАссистент:"
         
-        # Генерация ответа
         response = chatbot(
-            message,
-            max_length=100,
-            temperature=0.7,
+            prompt,
+            max_length=150,
+            temperature=0.8,
             do_sample=True,
             top_p=0.9,
-            repetition_penalty=1.1
+            repetition_penalty=1.1,
+            num_return_sequences=1,
+            pad_token_id=chatbot.tokenizer.eos_token_id
         )
         
+        # Извлекаем только ответ ассистента
+        generated_text = response[0]['generated_text']
+        assistant_response = generated_text.replace(prompt, "").strip()
+        
+        # Очищаем ответ
+        if "Пользователь:" in assistant_response:
+            assistant_response = assistant_response.split("Пользователь:")[0].strip()
+        
+        logger.info(f"Generated response: {assistant_response}")
+        
         return jsonify({
-            "response": response[0]['generated_text'],
+            "response": assistant_response,
             "status": "success"
         })
         
     except Exception as e:
-        logger.error(f"Error: {e}")
+        logger.error(f"Error in /chat: {e}")
         return jsonify({"error": "Internal server error"}), 500
 
-@app.route('/health', methods=['GET'])
-def health():
-    return jsonify({"status": "healthy"})
+@app.route('/clear', methods=['POST'])
+def clear_history():
+    # Для pipeline очистка не требуется, но endpoint для совместимости
+    return jsonify({"status": "success", "message": "OK"})
 
 if __name__ == '__main__':
-    app.run(host='0.0.0.0', port=7860)
+    app.run(host='0.0.0.0', port=7860, debug=False)