fix runtime error

app.py

@@ -1,20 +1,22 @@
+import os
 from fastapi import FastAPI
 from pydantic import BaseModel
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 
-# Khởi tạo app FastAPI
-app = FastAPI()
+# Fix lỗi ghi cache
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/hf_cache"
 
-# Load model và tokenizer
-model_name = "bmd1905/BARTpho2-ViT5-text2text"
+# Load model
+model_name = "bmd1905/BARTpho2-ViT5-question-answering"  # model này public
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 
-# Định nghĩa schema cho request
+# FastAPI app
+app = FastAPI()
+
 class InputText(BaseModel):
     text: str
 
-# Endpoint POST /generate
 @app.post("/generate")
 def generate_text(item: InputText):
     inputs = tokenizer(item.text, return_tensors="pt", truncation=True, max_length=512)