Update app.py

app.py

@@ -38,10 +38,10 @@ def load_model():
 
     model = AutoModelForCausalLM.from_pretrained(
         MODEL_NAME,
-        quantization_config=nf4_config, # add config
+        # quantization_config=nf4_config, # add config
         torch_dtype=torch.bfloat16, # save memory using float16
         # low_cpu_mem_usage=True,
-        token=get_hg_token(),
+        token=st.secrets["HF_TOKEN"],
     ).to("cuda")
 
     tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)