CritiQ: Mining Data Quality Criteria from Human Preferences

TL;DR

CritiQ is a novel data selection method that automatically mines human preference-based quality criteria, enabling efficient and interpretable data filtering for language models with minimal human annotation.

Contribution

It introduces CritiQ Flow, a system that evolves quality criteria from limited human preferences and leverages a knowledge base for improved data selection in language model training.

Findings

Achieves high accuracy in data quality assessment across multiple domains.

Improves downstream task performance of Llama 3.1 models with selected data.

Demonstrates the interpretability and reusability of verbal quality criteria.

Abstract

Language model heavily depends on high-quality data for optimal performance. Existing approaches rely on manually designed heuristics, the perplexity of existing models, training classifiers, or careful prompt engineering, which require significant expert experience and human annotation effort while introduce biases. We introduce CritiQ, a novel data selection method that automatically mines criteria from human preferences for data quality with only ~30 human-annotated pairs and performs efficient data selection. The main component, CritiQ Flow, employs a manager agent to evolve quality criteria and worker agents to make pairwise judgments. We build a knowledge base that extracts quality criteria from previous work to boost CritiQ Flow. Compared to perplexity- and classifier- based methods, verbal criteria are more interpretable and possess reusable value. After deriving the criteria, we train the CritiQ Scorer to give quality scores and perform efficient data selection. We demonstrate the effectiveness of our method in the code, math, and logic domains, achieving high accuracy on human-annotated test sets. To validate the quality of the selected data, we continually train Llama 3.1 models and observe improved performance on downstream tasks compared to uniform sampling. Ablation studies validate the benefits of the knowledge base and the reflection process. We analyze how criteria evolve and the effectiveness of majority voting.