Update app.py

app.py

@@ -3,6 +3,7 @@ import os, json, torch, asyncio
 from fastapi import FastAPI, WebSocket, WebSocketDisconnect
 from huggingface_hub import login
 from transformers import AutoTokenizer, AutoModelForCausalLM, LogitsProcessor
+from transformers.generation.utils import Cache
 from snac import SNAC
 
 # 0) Login + Device ---------------------------------------------------