Table of Links Abstract and 1. Introduction Abstract and 1. Introduction Definition of Critique Ability


Construction of CriticBench
3.1 Data Generation
3.2 Data Selection


Properties of Critique Ability
4.1 Scaling Law
4.2 Self-Critique Ability
4.3 Correlation to Certainty


New Capacity with Critique: Self-Consistency with Self-Check


Conclusion, References, and Acknowledgments Definition of Critique Ability Definition of Critique Ability Definition of Critique Ability Construction of CriticBench
3.1 Data Generation
3.2 Data Selection Construction of CriticBench Construction of CriticBench 3.1 Data Generation 3.1 Data Generation 3.2 Data Selection 3.2 Data Selection Properties of Critique Ability
4.1 Scaling Law
4.2 Self-Critique Ability
4.3 Correlation to Certainty Properties of Critique Ability Properties of Critique Ability 4.1 Scaling Law 4.1 Scaling Law 4.2 Self-Critique Ability 4.2 Self-Critique Ability 4.3 Correlation to Certainty 4.3 Correlation to Certainty New Capacity with Critique: Self-Consistency with Self-Check New Capacity with Critique: Self-Consistency with Self-Check New Capacity with Critique: Self-Consistency with Self-Check Conclusion, References, and Acknowledgments Conclusion, References, and Acknowledgments Conclusion, References, and Acknowledgments A. Notations A. Notations B. CriticBench: Sources of Queries B. CriticBench: Sources of Queries C. CriticBench: Data Generation Details C. CriticBench: Data Generation Details D. CriticBench: Data Selection Details D. CriticBench: Data Selection Details E. CriticBench: Statistics and Examples E. CriticBench: Statistics and Examples F. Evaluation Settings F. Evaluation Settings D CRITICBENCH: DATA SELECTION DETAILS D.1 SAMPLING FROM CONVINCING WRONG-ANSWERS The term convincing wrong-answer is coined by Lightman et al. (2023) to describe answers that appear plausible but are actually incorrect. Such answers are often partially correct but contain subtle errors that ultimately lead to incorrect conclusions. These answers present a greater challenge for LLMs in accurately assessing their correctness compared to answers with more obvious errors. Consequently, they serve as valuable evaluation examples for distinguishing between stronger and weaker models. convincing wrong-answer In generating responses to queries from GSM8K and TruthfulQA, each response usually comprises an intermediate chain-of-thought and a final answer. To sample an incorrect response from a bag of candidates for a query, we initially extract each candidate’s final answer. Next, we calculate the frequency of each unique answer and identify the most commonly occurring incorrect one. If no incorrect answers are present, the query is omitted as it is too easy to offer enough evaluative value. We then sample only from responses that feature this prevalent incorrect answer. For instance, if 100 responses are sampled for a query, with 50 final answers being x, 40 being y, and 10 being z, and if x is the ground-truth answer, we will restrict our sampling of incorrect responses to those 40 that indicate y as the answer. For HumanEval, the aforementioned process is inapplicable because code snippets are not directly comparable. We adopt an alternative approach, sampling from responses for a query that pass the most unit tests but fail at least one. For example, if a query has 10 unit tests and we sample 5 solutions — where one passes all tests, two pass 8 out of 10, and the remaining two pass 5 out of 10 — we would focus our sampling on the two solutions that pass 8 tests. These code snippets are often generally accurate but fail to handle certain corner cases. D.2 COMPLEXITY-BASED SELECTION Fu et al. (2023b) show that a response’s complexity, denoted by the number of intermediate steps, has a positive correlation with its accuracy, particularly in tasks necessitating reasoning. To leverage this finding, we employ a complexity-based sampling strategy when selecting from either correct or commonly incorrect responses. Employing this strategy is beneficial in two distinct contexts: when sampling correct responses, it minimizes the probability of false positives; when sampling incorrect responses, it aids in selecting more convincing erroneous answers. D.3 FILTERING BY GENERATOR During development, we find that smaller models, specifically PaLM-2-XXS and PaLM-2-XS, yield responses of very low quality. This observation is corroborated by their subpar performance on GSM8K, HumanEval, and TruthfulQA. Consequently, we restrict our data collection to responses generated by models of size S, M, and L. D.4 CERTAINTY-BASED SELECTION E CRITICBENCH: STATISTICS AND EXAMPLES E.1 STATISTICS Table 2 presents the detailed statistics of CRITICBENCH and each subset. E.2 EXAMPLES Figure 8, 9 and 10 provide examples in CRITICBENCH. Authors:
(1) Liangchen Luo, Google Research (luolc@google.com);
(2) Zi Lin, UC San Diego;
(3) Yinxiao Liu, Google Research;
(4) Yun Zhu, Google Research;
(5) Jingbo Shang, UC San Diego;
(6) Lei Meng, Google Research (leimeng@google.com). Authors: Authors: (1) Liangchen Luo, Google Research (luolc@google.com); (2) Zi Lin, UC San Diego; (3) Yinxiao Liu, Google Research; (4) Yun Zhu, Google Research; (5) Jingbo Shang, UC San Diego; (6) Lei Meng, Google Research (leimeng@google.com). This paper is available on arxiv under CC BY 4.0 DEED license. This paper is available on arxiv under CC BY 4.0 DEED license. available on arxiv available on arxiv

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

Why “Almost Right” Answers Are the Hardest Test for AI

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Untitled Story

A Formalization of the SylloBio-NLI Resource Generation Process

CRITICBENCH: A Benchmark for Evaluating the Critique Abilities of LLMs

Constructing CRITICBENCH: Scalable, Generalizable, and High-Quality LLM Evaluation

Are Your AI Benchmarks Fooling You?

Why Even the Best AI Struggles at Critiquing Code

A Formalization of the SylloBio-NLI Resource Generation Process

CRITICBENCH: A Benchmark for Evaluating the Critique Abilities of LLMs

Constructing CRITICBENCH: Scalable, Generalizable, and High-Quality LLM Evaluation

Are Your AI Benchmarks Fooling You?

Why Even the Best AI Struggles at Critiquing Code

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps