Update app.py

app.py

@@ -156,8 +156,8 @@ def load_models_and_documents():
             trust_remote_code=True,
             enforce_eager=True,
             quantization='gptq',
-            gpu_memory_utilization=0.5,
-            load_format='npcache'
+            gpu_memory_utilization=0.5
+            #load_format='npcache'
         )
         model = models.VLLM(llm)