chain of thought

Chain-of-thought

Summarize and classify related papers about CoT.

Original work on CoT:

• First release time: 28 Jan 2022
• Title: Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
• Conference: NeurIPS 2022

Content

• Related Surveys
• COT Enhancement
• Two main types of CoTs
• Why CoT works? - Analysis of CoT
• CoT Evaluation
• Other types of CoTs

Related Surveys

First release time	Title	Conference
27 Sep 2023	A Survey of Chain of Thought Reasoning: Advances, Frontiers and Future	arXiv
15 Feb 2023	Augmented Language Models: a Survey	arXiv
4 Jan 2023	Iterated Decomposition: Improving Science Q&A by Supervising Reasoning Processes	arXiv
20 Dec 2022	Towards Reasoning in Large Language Models: A Survey	ACL 23 Findings
19 Dec 2022	Reasoning with Language Model Prompting: A Survey	ACL 2023

COT Enhancement

First release time	Title	What Changes?
21 Mar 2022	Self-Consistency Improves Chain of Thought Reasoning in Language Models	naive greedy decoding => self-consistency

Two main types of CoTs

Zero-shot:

First release time	Title	Conference
6 May 2023	Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models	ACL 23
3 Nov 2022	Large Language Models Are Human-Level Prompt Engineers	ICLR 23
24 May 2022	Large Language Models are Zero-Shot Reasoners	NeurIPS2022

Few-shot: five sub-groups

• Related to code generation
• Auto CoT
• Iterative prompt CoT
• Involve question decomposition
• Mix

Related to code generation

First release time	Title	Conference
22 Nov 2022	Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks	arXiv
18 Nov 2022	PAL: Program-aided Language Models	ICML 2023
13 Oct 2022	Language models of code are few-shot commonsense learners	EMNLP 2022

Involve question decomposition

First release time	Title	Conference
8 Dec 2022	Successive Prompting for Decomposing Complex Questions	EMNLP 2022
7 Oct 2022	Measuring and Narrowing the Compositionality Gap in Language Models	EMNLP 2023 (Findings)
5 Oct 2022	Decomposed Prompting: A Modular Approach for Solving Complex Tasks	ICLR 2023
21 May 2022	Least-to-Most Prompting Enables Complex Reasoning in Large Language Models	ICLR 2023
19 May 2022	Selection-Inference: Exploiting Large Language Models for Interpretable Logical Reasoning	ICLR 2023
15 May 2022	SeqZero: Few-shot Compositional Semantic Parsing with Sequential Prompts and Zero-shot Models	Findings of NAACL 2022

Auto CoT

First release time	Title	Conference
24 Feb 2023	Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Data	arXiv
7 Oct 2022	Automatic Chain of Thought Prompting in Large Language Models	ICLR 23

Why CoT works? - Analysis of CoT

First release time	Title	Conference
20 Dec 2022	Towards Understanding Chain-of-Thought Prompting: An Empirical Study of What Matters	ACL 2023
25 Nov 2022	Complementary Explanations for Effective In-Context Learning	ACL 2023 (Findings)
3 Oct 2022	Language Models Are Greedy Reasoners: A Systematic Formal Analysis of Chain-of-Thought	ICLR 2023
16 Sep 2022	Text and Patterns: For Effective Chain of Thought, It Takes Two to Tango	Google Research

CoT Evaluation

Other types of CoTs

First release time	Name	Title	Conference
28 May 2023	Tab-CoT	Tab-CoT: Zero-shot Tabular Chain of Thought	ACL 23 Findings
17 May 2023	Tree of Thoughts	Tree of Thoughts: Deliberate Problem Solving with Large Language Models	arXiv
9 May 2023	Memory of Thoughts	MoT: Pre-thinking and Recalling Enable ChatGPT to Self-Improve with Memory-of-Thoughts	arXiv
22 Nov 2022	Program of Thoughts	Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks	arXiv