torch colorama APScheduler black click datasets gradio gradio_client huggingface-hub matplotlib numpy pandas plotly python-dateutil requests semantic-version tqdm wandb transformers>=4.36.0 tokenizers>=0.15.0 lm_eval[ifeval] @ git+https://github.com/EleutherAI/lm-evaluation-harness.git@v0.4.2 accelerate sentencepiece langdetect sacrebleu cchardet rouge_score bert-score evaluate spacy==3.7.4 selfcheckgpt immutabledict gputil bitsandbytes