Update handler.py

handler.py

@@ -17,8 +17,8 @@ alpaca_prompt = """Below is an instruction that describes a task, paired with an
 class EndpointHandler:
     def __init__(self, path=""):
         # load model and processor from path
-        self.model = AutoModelForCausalLM.from_pretrained(path, device_map="auto")
-        self.tokenizer = AutoTokenizer.from_pretrained(path)
+        self.model = AutoModelForCausalLM.from_pretrained(path, gguf_file='./unsloth.Q4_K_M.gguf')
+        self.tokenizer = AutoTokenizer.from_pretrained(path, gguf_file='./unsloth.Q4_K_M.gguf')
 
         with open(f"{path}/zero_shot_cot_prompt.txt", 'r') as file:
             self.instruction_prompt = file.read()