Update utils.py

utils.py

@@ -82,6 +82,20 @@ def load_tokenizer_and_model(base_model, load_8bit=False):
     return tokenizer,model, device
 
 
+def load_tokenizer_and_model_Blaize(base_model, load_8bit=True):
+    if torch.cuda.is_available():
+        device = "cuda"
+    else:
+        device = "cpu"
+    
+    
+    tokenizer = LlamaTokenizer.from_pretrained(base_model, add_eos_token=True, use_auth_token=True)
+    model = LlamaForCausalLM.from_pretrained(base_model, load_in_8bit=True, device_map="auto")
+    #model.eval()
+    return tokenizer,model, device
+
+
+
 def load_model(base_model, load_8bit=False):
     if torch.cuda.is_available():
         device = "cuda"