Your resource for web content, online publishing
and the distribution of digital products.
S M T W T F S
 
 
 
 
 
1
 
2
 
3
 
4
 
5
 
6
 
7
 
8
 
9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
 
24
 
25
 
26
 
27
 
28
 
29
 
30
 

Deductive Verification with Natural Programs: Case Studies

DATE POSTED:September 8, 2024

:::info Authors:

(1) Zhan Ling, UC San Diego and equal contribution;

(2) Yunhao Fang, UC San Diego and equal contribution;

(3) Xuanlin Li, UC San Diego;

(4) Zhiao Huang, UC San Diego;

(5) Mingu Lee, Qualcomm AI Research and Qualcomm AI Research

(6) Roland Memisevic, Qualcomm AI Research;

(7) Hao Su, UC San Diego.

:::

Table of Links

Abstract and Introduction

Related work

Motivation and Problem Formulation

Deductively Verifiable Chain-of-Thought Reasoning

Experiments

Limitations

Conclusion, Acknowledgements and References

\ A Deductive Verification with Vicuna Models

B More Discussion on Improvements of Deductive Verification Accuracy Versus Improvements on Final Answer Correctness

C More Details on Answer Extraction

D Prompts

E More Deductive Verification Examples

E More Deductive Verification Examples

In this section, we present more deductive verification examples using our Natural Program-based approach on single reasoning steps.

\ In Tab. 18, we demonstrate that the language model (ChatGPT) not only successfully identifies ungrounded information, but also identifies logical errors within the given solutions.

\ In Tab. 19, we illustrate a case where the language model fails to detect ungrounded premise numbers, mistakenly assuming that these numbers can be derived from grounded ones.

\ Lastly, in Tab. 20, we illustrate a case where the language model is sometimes unable to correctly identify grounded numbers.

\  Two-shot prompt for direct reasoning chain verification without Natural Program format.

\  One-shot Natural Program prompt for reasoning chain generation on math word problems.

\  One-shot Natural Program prompt for reasoning chain generation on math word problems with multiple choice.

\  Two-shot Natural Program prompt for reasoning chain generation on the Date dataset.

\  One-shot Natural Program prompt for reasoning chain generation on the Last Letters dataset.

\  One-shot prompt for deductive verification of a single reasoning step, following our Natural Program format and step-by-step reasoning chain decomposition.

\  our deductive verification approach successfully discovers ungrounded information and reasoning mistakes.

\  our deductive verification process fails to find out ungrounded information in the reasoning step. The number 240 in the reasoning step is ungrounded, but the model states that it can be calculated from grounded numbers.

\  our deductive verification process sometimes treats grounded information as if they were ungrounded. The number 120 is provided in the given information, but the model states that it is ungrounded.

\

:::info This paper is available on arxiv under CC BY 4.0 DEED license.

:::

\