Update utils.py

utils.py

@@ -87,10 +87,11 @@ def load_tokenizer_and_model(base_model,load_8bit=False):
             load_in_8bit=load_8bit,
             torch_dtype=torch.float16,
             device_map="auto",
+            trust_remote_code=True
         )
     else:
         model = AutoModelForCausalLM.from_pretrained(
-            base_model, device_map={"": device}, low_cpu_mem_usage=True
+            base_model, device_map={"": device}, low_cpu_mem_usage=True, trust_remote_code=True
         )
 
     #if not load_8bit: