nltk
datasets
transformers
torch
tokenizers
evaluate
rouge_score
sentencepiece
huggingface_hub