Update handler.py

handler.py

@@ -5,13 +5,12 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, AutoConfig
 
 
 class EndpointHandler:
-    def __init__(self, path="", unused=""):
-        print('===> __init__', path, unused)
+    def __init__(self, path=""):
         # load model and tokenizer from path
         self.tokenizer = AutoTokenizer.from_pretrained(path) # AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b")
 
         config = AutoConfig.from_pretrained(path, trust_remote_code=True)
-        config.attn_config['attn_impl'] = 'triton'
+        # config.attn_config['attn_impl'] = 'triton'
         config.init_device = 'cuda:0' # For fast initialization directly on GPU!
         config.max_seq_len = 4096 # (input + output) tokens can now be up to 4096