ai_dataset/main_qa_data/resource/tokenizer/enDict