Update README.md

README.md

@@ -30,7 +30,7 @@ For batch processing, the model takes as input
 The model compresses the questions into context embeddings and answers the question based on the provided context embeddings.
 
 ```python
-from transformers import AutoModelForCausalLM
+from transformers import AutoModel
 
 model = AutoModel.from_pretrained('naver/cocom-v1-128-mistral-7b', trust_remote_code=True)
 model = model.to('cuda')