Update app.py

app.py

@@ -3,7 +3,7 @@ import spaces
 
 import os
 import gradio as gr
-from transformers import LlamaForCausalLM, LlamaTokenizer
+from transformers import AutoTokenizer, AutoModelForCausalLM
 from huggingface_hub import hf_hub_download, snapshot_download
 
 # Load the model and tokenizer from Hugging Face
@@ -11,8 +11,8 @@ model_path = snapshot_download(
     repo_id=os.environ.get("REPO_ID", "SimpleBerry/LLaMA-O1-Supervised-1129")
 )
 
-tokenizer = LlamaTokenizer.from_pretrained(model_path)
-model = LlamaForCausalLM.from_pretrained(model_path,device_map='auto')
+tokenizer = AutoTokenizer.from_pretrained(model_path)
+model = AutoModelForCausalLM.from_pretrained(model_path,device_map='auto')
 
 DESCRIPTION = '''
 # SimpleBerry/LLaMA-O1-Supervised-1129 | Duplicate the space and set it to private for faster & personal inference for free.