Abstract
From the background information supplied by the inventors, news correspondents obtained the followingquote: “Large language models (LLMs) such as OpenAl’s ChatGPT can exhibit human-level abilities toanswer questions and perform tasks. One such important application is answering a question based oninformation contained in a potentially very large set of documents. Unfortunately, LLMs are very expensiveto run both in terms of computational time and clock time. For extremely large corpora of documents thatinclude, for instance, hundreds of thousands of documents, the clock time and compute time required toanswer a question becomes intractable. For instance, answering a simple question based on such a largecorpus of documents may require months of time using a cluster of state-of-the-art GPUs. Accordingly,improved techniques for natural language processing are desired.”
NETL
NSTL