Update app.py

app.py

@@ -3,13 +3,16 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 import os
 
+# ✅ Force Hugging Face cache to /tmp (writable in Spaces)
+os.environ["HF_HOME"] = "/tmp"
+os.environ["TRANSFORMERS_CACHE"] = "/tmp"
 
-os.environ["TRANSFORMERS_CACHE"] = "/app/cache"
 
 model_id = "rabiyulfahim/qa_python_gpt2"
 
-tokenizer = AutoTokenizer.from_pretrained(model_id, cache_dir="/app/cache")
-model = AutoModelForCausalLM.from_pretrained(model_id, cache_dir="/app/cache")
+tokenizer = AutoTokenizer.from_pretrained(model_id, cache_dir="/tmp")
+model = AutoModelForCausalLM.from_pretrained(model_id, cache_dir="/tmp")
+
 
 app = FastAPI(title="QA GPT2 API", description="Serving HuggingFace model with FastAPI")