[TOC]
- Title: Towards Uncertainty Aware Language Agent
- Author: Jiuzhou Han et. al.
- Publish Year: 30 May 2024
- Review Date: Thu, Jun 20, 2024
- url: arXiv:2401.14016v3
Summary of paper
Motivation
- The existing approaches neglect the notion of uncertainty during these interactions
Contribution
Some key terms
Related work 1: lang agent
- the author define what is language agent and discuss it – the prominent work of ReAct propose a general language agent framework to combine reasoning and acting with LLMs for solving diverse language reasoning tasks.
- continue the track of ReAct – introducing Reflexion, use the history failure trials as input to ask for reflection and can gain better results
- FireAct – add more diverse fine-tuning data to improve the performance
Later the author mention Toolformer, Gorilla and other lang agent that is not start from ReAct
Related work 2: uncertainty in generation with LLMs
- author mentioned some common practice in leveraging uncertainty during lang generation – via sampling or decoding which do not measure uncertainty directly, but rather they rely on the stochasticity over prediction space along with a form of aggregation approach such as majority voting.
- then the author talked about Consistency – a sampling method which takes majority voting over multiple sampling outputs. then they refer readers to a paper For a comprehensive review of sampling and decoding methods in NLG.
- the author stated that “We focus on leveraging uncertainty estimation in free-form QA with short answers. Uncertainty estimation in free-form NLG remains a challenge for LLMs due to the diversity of the outputs.”
- then the author give the category of how to do uncertainty estimation in NLG:
- logits or entropy based methods
- prompt based methods
- all the sections are provided with detailed info about previous work on these methods of uncertainty estimation
Results
