Update utils.py

utils.py

@@ -123,6 +123,28 @@ def load_tokenizer_and_model_gpt2(base_model,load_8bit=False):
 
 
 
+    
+
+
+
+def load_tokenizer_and_model_bloke_gpt(base_model, model_basename):
+    if torch.cuda.is_available():
+        device = "cuda"
+    else:
+        device = "cpu"
+
+    tokenizer = AutoTokenizer.from_pretrained(base_model, use_fast=True)
+    model = AutoGPTQForCausalLM.from_quantized(base_model,
+        model_basename=model_basename,
+        use_safetensors=True,
+        trust_remote_code=False,
+        device="cuda:0",
+        use_triton=use_triton,
+        quantize_config=None)
+    
+    return tokenizer,model,device
+
+
 # Greedy Search
 def greedy_search(input_ids: torch.Tensor,
                   model: torch.nn.Module,