JudgeLM: Fine-tuned Large Language Models are Scalable Judges
This addresses the problem of comprehensive LLM evaluation for researchers and practitioners, though it is incremental as it builds on existing fine-tuning and bias mitigation techniques.
The authors tackled the challenge of evaluating Large Language Models (LLMs) in open-ended scenarios by fine-tuning LLMs as scalable judges (JudgeLM), achieving state-of-the-art performance with over 90% agreement with teacher judges and high efficiency (e.g., 3 minutes for 5K samples on 8 GPUs).
Evaluating Large Language Models (LLMs) in open-ended scenarios is challenging because existing benchmarks and metrics can not measure them comprehensively. To address this problem, we propose to fine-tune LLMs as scalable judges (JudgeLM) to evaluate LLMs efficiently and effectively in open-ended benchmarks. We first propose a comprehensive, large-scale, high-quality dataset containing task seeds, LLMs-generated answers, and GPT-4-generated judgments for fine-tuning high-performance judges, as well as a new benchmark for evaluating the judges. We train JudgeLM at different scales from 7B, 13B, to 33B parameters, and conduct a systematic analysis of its capabilities and behaviors. We then analyze the key biases in fine-tuning LLM as a judge and consider them as position bias, knowledge bias, and format bias. To address these issues, JudgeLM introduces a bag of techniques including swap augmentation, reference support, and reference drop, which clearly enhance the judge's performance. JudgeLM obtains the state-of-the-art judge performance on both the existing PandaLM benchmark and our proposed new benchmark. Our JudgeLM is efficient and the JudgeLM-7B only needs 3 minutes to judge 5K samples with 8 A100 GPUs. JudgeLM obtains high agreement with the teacher judge, achieving an agreement exceeding 90% that even surpasses human-to-human agreement. JudgeLM also demonstrates extended capabilities in being judges of the single answer, multimodal models, multiple answers, multi-turn chat, etc. Code is available at https://github.com/baaivision/JudgeLM.