1 of 27

Zero-Shot Reasoning by LLM: Math Word Problem Solving and Next-Item Recommendation

Lei Wang

Homepage: https://demoleiwang.github.io/HomePage/

2 of 27

Overview

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023.

​

​

Zero-Shot Next-Item Recommendation using Large Pretrained Language Models.

​

​

Other Recent Works.

3 of 27

Overview

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023.

​

​

Zero-Shot Next-Item Recommendation using Large Pretrained Language Models.

4 of 27

Chain-of-Thought (CoT) Prompting

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. NeurIPS 2022.

Demos

Test

Instance

5 of 27

Chain-of-Thought (CoT) Prompting

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. NeurIPS 2022.

6 of 27

Zero-Shot CoT

Large Language Models are Zero-Shot Reasoners. NeurIPS 2022.

7 of 27

Zero-Shot CoT

Large Language Models are Zero-Shot Reasoners. NeurIPS 2022.

8 of 27

Zero-Shot CoT

Large Language Models are Zero-Shot Reasoners. NeurIPS 2022.

9 of 27

Least-to-Most Prompting

Least-to-Most Prompting Enables Complex Reasoning in Large Language Models. ICLR 2023.

10 of 27

Motivation

Error Analysis of Zero-Shot-CoT

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023

11 of 27

PS Prompting

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023

12 of 27

PS+ Prompting

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023

13 of 27

Results

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023

14 of 27

Overview

Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models. ACL 2023.

​

​

Zero-Shot Next-Item Recommendation using Large Pretrained Language Models.

15 of 27

Motivation

Domain gap.

​

Very large recommendation space.

​

Modeling user preference.

16 of 27

Example

17 of 27

Three-Step Prompting with Candidate Set

18 of 27

Results

19 of 27

Overview

Other Recent Works.

  • LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models. Manuscript 2023.

​

  • ICL-D3IE: In-context learning with diverse demonstrations updating for document information extraction. Manuscript 2023.

​

  • T-SciQ: Teaching Multimodal Chain-of-Thought Reasoning via Large Language Model Signals for Science Question Answering. Manuscript 2023.

​

  • Read, Diagnose and Chat: Towards Explainable and Interactive LLMs-Augmented Depression Detection in Social Media. Manuscript 2023.

20 of 27

LLM-Adapters

LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models. Manuscript 2023.

GitHub: https://github.com/AGI-Edgerunners/LLM-Adapters

21 of 27

LLM-Adapters

GitHub: https://github.com/AGI-Edgerunners/LLM-Adapters

22 of 27

ICL-D3IE

ICL-D3IE: In-context learning with diverse demonstrations updating for document information extraction. Manuscript 2023.

23 of 27

ICL-D3IE

ICL-D3IE: In-context learning with diverse demonstrations updating for document information extraction. Manuscript 2023.

24 of 27

T-SciQ

T-SciQ: Teaching Multimodal Chain-of-Thought Reasoning via Large Language Model Signals for Science Question Answering. Manuscript 2023.

25 of 27

T-SciQ

T-SciQ: Teaching Multimodal Chain-of-Thought Reasoning via Large Language Model Signals for Science Question Answering. Manuscript 2023.

26 of 27

Read, Diagnose and Chat

Read, Diagnose and Chat: Towards Explainable and Interactive LLMs-Augmented Depression Detection in Social Media. Manuscript 2023.

27 of 27

Thanks!