arXiv 2307.06945
In-context Autoencoder for Context Compression in a Large Language Model
By Tao Ge, Jing Hu, et al.
Published 2023-07-13
Citation lineage
Review the prior work and downstream research connected to this paper.
We propose the In-context Autoencoder (ICAE), leveraging the power of a large language model (LLM) to compress a long context into short compact memory slots that can be directly conditioned on by the LLM for various purposes. ICAE is first pretrained using both autoencoding and language modeling objectives on massive text data, enabling it to generate memory slots that accurately and comprehensively represent the o…