Summary of A Human-inspired Reading Agent with Gist Memory Of Very Long Contexts, by Kuang-huei Lee et al.
A Human-Inspired Reading Agent with Gist Memory of Very Long Contexts
by Kuang-Huei Lee, Xinyun Chen, Hiroki Furuta, John Canny, Ian Fischer
First submitted to arxiv on: 15 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed ReadAgent system is an Large Language Model (LLM) agent that can process long inputs with increased effective context length, up to 20 times more than current LLMs. It uses a simple prompting system inspired by human reading habits, consisting of three components: deciding what content to store together in a memory episode, compressing those episodes into gist memories, and taking actions to look up passages if needed. ReadAgent outperforms baselines on long-document reading comprehension tasks such as QuALITY, NarrativeQA, and QMSum. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ReadAgent is a new way for computers to understand really long texts. Right now, computer models can only handle short pieces of text. But humans can read long documents easily because we have ways of organizing what we’ve read in our minds. ReadAgent uses this idea to help computer models read longer texts by breaking them down into smaller chunks and storing important bits in “memory episodes”. It also has a way to quickly look up information if it needs to remind itself of something important. This new system is better than old methods at understanding long texts, and it can even process texts that are 3.5 to 20 times longer than before. |
Keywords
» Artificial intelligence » Context length » Large language model » Prompting
