Update app.py

app.py

@@ -428,7 +428,7 @@ def load_llm_model(model_path: str) -> Tuple[str, str]:
             state.llm_model = LLM(
                 model=model_path,
                 tensor_parallel_size=tp_size,
-                gpu_memory_utilization=0.60,
+                gpu_memory_utilization=0.80,
                 max_model_len=15000,
             )
             state.llm_tokenizer = state.llm_model.get_tokenizer()