Fix.

rag.py

@@ -61,9 +61,9 @@ def load_index(index_file: str) -> faiss.Index:
 
 
 def load_reader(model_path: str, device: str) -> TextGenerationPipeline:
-    model = AutoModelForCausalLM.from_pretrained(model_path)
+    model = AutoModelForCausalLM.from_pretrained(model_path, token=HF_TOKEN)
 
-    tokenizer = AutoTokenizer.from_pretrained(model_path)
+    tokenizer = AutoTokenizer.from_pretrained(model_path, token=HF_TOKEN)
     tokenizer.pad_token = tokenizer.eos_token
 
     reader = pipeline(