Update app.py

app.py

@@ -324,7 +324,7 @@ title_markdown = """
 # 🌋 LLaVA: Large Language and Vision Assistant
 [[Project Page]](https://llava-vl.github.io) [[Paper]](https://arxiv.org/abs/2304.08485) [[Code]](https://github.com/haotian-liu/LLaVA) [[Model]](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md)
 
-ONLY WORKS WITH GPU!
+ONLY WORKS WITH GPU! 8-bit is used by default to load 13B model in A10G-Large (24G). Set environment `bits` to 4 to load it in T4-medium (15G).
 """
 
 tos_markdown = """
@@ -591,8 +591,7 @@ if __name__ == "__main__":
     logger.info(f"args: {args}")
 
     model_path = "liuhaotian/llava-v1.5-13b"
-    bits = 4
-    # set bits=4 for T4, bits=8 for A10G (24G), and bits=16 for A100 (40G)
+    bits = os.getenv("bits", 8)
 
     preload_models(model_path)