Unleashing the Power of Data Tsunami: A Comprehensive Survey on Data Assessment and Selection for Instruction Tuning of Language Models
Abstract: Instruction tuning plays a critical role in aligning large language models (LLMs) with human preference. Despite the vast amount of open instruction datasets, naively training a LLM on all existing instructions may not be optimal and practical. To pinpoint the most beneficial datapoints, data assessment and selection methods have been proposed in the fields of natural language processing (NLP) and deep learning. However, under the context of instruction tuning, there still exists a gap in knowledge on what kind of data evaluation metrics can be employed and how they can be integrated into the selection mechanism. To bridge this gap, we present a comprehensive review on existing literature of data assessment and selection especially for instruction tuning of LLMs. We systematically categorize all applicable methods into quality-based, diversity-based, and importance-based ones where a unified, fine-grained taxonomy is structured. For each category, representative methods are elaborated to describe the landscape of relevant research. In addition, comparison between the latest methods is conducted on their officially reported results to provide in-depth discussions on their limitations. Finally, we summarize the open challenges and propose the promosing avenues for future studies. All related contents are available at https://github.com/yuleiqin/fantastic-data-engineering.
Submission Length: Long submission (more than 12 pages of main content)
Changes Since Last Submission: All the comments from the reviewers and editor have been addressed point-by-point.
Accordingly, the manuscript has been modified including but not limited to the followings:
* explanations on the survey scope
* explanations on the differences between existing surveys
* unifiication and modification of mathematic equations
* improvement on the illustration of preprocessing (tokenization) of instruction datasets
* added discussions on the hybrid techniques of data selection
* added discussions on the connections and distinctions between selection methods that prioritize different data aspects
* added discussions on the future study and research directions
* added discussions on the bias and fairness concerns in data selection
* corrections on typos and layout of latex formatting
Code: https://github.com/yuleiqin/fantastic-data-engineering
Certifications: Survey Certification
Assigned Action Editor: ~Xingyou_Song1
Submission Number: 3172
Loading