chunked/nltk_chunking/model_doc_xglm/chunk_1.txt

The abstract from the paper is the following:
Large-scale autoregressive language models such as GPT-3 are few-shot learners that can perform a wide range of language 
tasks without fine-tuning.