https://t25556536.p.clickup-attachments.com/t25556536/7613b9b1-80ff-4763-b455-85a69229c31f/image.png

Preprint, Oct 17, 2023

Introduction


Motivation

Propose

https://t25556536.p.clickup-attachments.com/t25556536/a77d1138-831d-48e6-9fed-c8da58f83ab4/image.png

Related work


Retrieval-Augmented Generation

https://t25556536.p.clickup-attachments.com/t25556536/c0a8965b-943d-426e-8f4e-1e3b57bb49cf/image.png

Reinforcement Learning from Human Feedback (RLHF)

GPT-3 ➝ ChatGPT에서 큰 성능 변화를 이끌어낸, 사람을 이용한 LLM 학습 방식