Update app.py

app.py

@@ -39,10 +39,10 @@ class ChatModel:
             
             model = AutoModelForCausalLM.from_pretrained(
                 config["model_name"],
-                quantization_config=bnb_config,
                 device_map="auto",
-                attn_implementation="flash_attention_2" if "phi-3" in model_name else None,
-                torch_dtype=torch.float16
+                torch_dtype=torch.float16,
+                attn_implementation="flash_attention_2" if "phi-3" in model_name else "eager",
+                low_cpu_mem_usage=True
             )
             
             self.models[model_name] = model