PandaLM: An Automatic Evaluation Benchmark for LLM Instruction Tuning Optimization

TL;DR

PandaLM is a novel large language model-based evaluation benchmark designed to assess instruction-tuned LLMs by considering subjective factors like clarity and adherence, providing a cost-effective and privacy-preserving alternative to traditional evaluation methods.

Contribution

This paper introduces PandaLM, a judge LLM trained to evaluate other models based on subjective criteria, and demonstrates its effectiveness and reliability in LLM evaluation.

Findings

PandaLM-7B achieves 93.75% of GPT-3.5's evaluation ability.

PandaLM evaluation correlates well with human preferences.

Models tuned with PandaLM outperform those with default hyperparameters.

Abstract

Instruction tuning large language models (LLMs) remains a challenging task, owing to the complexity of hyperparameter selection and the difficulty involved in evaluating the tuned models. To determine the optimal hyperparameters, an automatic, robust, and reliable evaluation benchmark is essential. However, establishing such a benchmark is not a trivial task due to the challenges associated with evaluation accuracy and privacy protection. In response to these challenges, we introduce a judge large language model, named PandaLM, which is trained to distinguish the superior model given several LLMs. PandaLM's focus extends beyond just the objective correctness of responses, which is the main focus of traditional evaluation datasets. It addresses vital subjective factors such as relative conciseness, clarity, adherence to instructions, comprehensiveness, and formality. To ensure the reliability of PandaLM, we collect a diverse human-annotated test dataset, where all contexts are generated by humans and labels are aligned with human preferences. Our results indicate that PandaLM-7B achieves 93.75% of GPT-3.5's evaluation ability and 88.28% of GPT-4's in terms of F1-score on our test dataset. PandaLM enables the evaluation of LLM to be fairer but with less cost, evidenced by significant improvements achieved by models tuned through PandaLM compared to their counterparts trained with default Alpaca's hyperparameters. In addition, PandaLM does not depend on API-based evaluations, thus avoiding potential data leakage. All resources of PandaLM are released at https://github.com/WeOpenML/PandaLM.