Update app.py

app.py

@@ -112,12 +112,13 @@ def load_model_and_tokenizer():
 # Create FastAPI app
 app = FastAPI(title="DeepSeek QA with KV Cache API")
 
-# Global variables to store the cache, origin length, and model/tokenizer
-cache_store = {}
 
+os.environ["HF_HOME"] = "/tmp/hf_cache"
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/hf_cache"
 # Initialize model and tokenizer at startup
 model, tokenizer = load_model_and_tokenizer()
-
+# Global variables to store the cache, origin length, and model/tokenizer
+cache_store = {}
 class QueryRequest(BaseModel):
     query: str
     max_new_tokens: Optional[int] = 150