Update app.py

app.py

@@ -39,7 +39,7 @@ def load_model():
     model = AutoModelForCausalLM.from_pretrained(
         MODEL_NAME,
         # quantization_config=nf4_config, # add config
-        torch_dtype=torch.bfloat16, # save memory using float16
+        # torch_dtype=torch.bfloat16, # save memory using float16
         # low_cpu_mem_usage=True,
         token=st.secrets["HF_TOKEN"],
     ).to("cuda")