Update handler.py

handler.py

@@ -1,16 +1,16 @@
 from typing import Dict, Any
 from transformers import AutoModelForCausalLM, AutoTokenizer
-from peft import PeftConfig, PeftModel
+from peft import PeftConfig, PeftModel, AutoPeftModelForCausalLM
 import torch.cuda
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 class EndpointHandler():
     def __init__(self, path=""):
-        config = PeftConfig.from_pretrained(path)
-        model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, device_map="auto")#, load_in_8bit=True, device_map='auto')
-        self.tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
-        self.model = PeftModel.from_pretrained(model, path)
+        # config = PeftConfig.from_pretrained(path)
+        # model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, device_map="auto")#, load_in_8bit=True, device_map='auto')
+        self.tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-j-6b")
+        self.model = AutoPeftModelForCausalLM.from_pretrained("ksee/testgptj")
 
     def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
         """