chunked/nltk_chunking/model_doc_openai-gpt/chunk_5.txt

In
contrast to previous approaches, we make use of task-aware input transformations during fine-tuning to achieve
effective transfer while requiring minimal changes to the model architecture.