Training an LLM-as-a-Judge Model: Pipeline, Insights, and Practical Lessons
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917913674383360 |
|---|---|
| author | Hu, Renjun Cheng, Yi Meng, Libin Xia, Jiaxin Zong, Yi Shi, Xing Lin, Wei |
| author_facet | Hu, Renjun Cheng, Yi Meng, Libin Xia, Jiaxin Zong, Yi Shi, Xing Lin, Wei |
| contents | The rapid advancement of large language models (LLMs) has opened new possibilities for their adoption as evaluative judges. This paper introduces Themis, a fine-tuned LLM judge that delivers sophisticated context-aware evaluations. We provide a comprehensive overview of the development pipeline for Themis, highlighting its scenario-dependent evaluation prompts and two novel methods for controlled instruction generation. These designs enable Themis to effectively distill evaluative skills from teacher models, while retaining flexibility for continuous development. We introduce two human-labeled benchmarks for meta-evaluation, demonstrating that Themis can achieve high alignment with human preferences in an economical manner. Additionally, we explore insights into the LLM-as-a-judge paradigm, revealing nuances in performance and the varied effects of reference answers. Notably, we observe that pure knowledge distillation from strong LLMs, though common, does not guarantee performance improvement through scaling. We propose a mitigation strategy based on instruction-following difficulty. Furthermore, we provide practical guidelines covering data balancing, prompt customization, multi-objective training, and metric aggregation. We aim for our method and findings, along with the fine-tuning data, benchmarks, and model checkpoints, to support future research and development in this area. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_02988 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Training an LLM-as-a-Judge Model: Pipeline, Insights, and Practical Lessons Hu, Renjun Cheng, Yi Meng, Libin Xia, Jiaxin Zong, Yi Shi, Xing Lin, Wei Computation and Language Artificial Intelligence Machine Learning The rapid advancement of large language models (LLMs) has opened new possibilities for their adoption as evaluative judges. This paper introduces Themis, a fine-tuned LLM judge that delivers sophisticated context-aware evaluations. We provide a comprehensive overview of the development pipeline for Themis, highlighting its scenario-dependent evaluation prompts and two novel methods for controlled instruction generation. These designs enable Themis to effectively distill evaluative skills from teacher models, while retaining flexibility for continuous development. We introduce two human-labeled benchmarks for meta-evaluation, demonstrating that Themis can achieve high alignment with human preferences in an economical manner. Additionally, we explore insights into the LLM-as-a-judge paradigm, revealing nuances in performance and the varied effects of reference answers. Notably, we observe that pure knowledge distillation from strong LLMs, though common, does not guarantee performance improvement through scaling. We propose a mitigation strategy based on instruction-following difficulty. Furthermore, we provide practical guidelines covering data balancing, prompt customization, multi-objective training, and metric aggregation. We aim for our method and findings, along with the fine-tuning data, benchmarks, and model checkpoints, to support future research and development in this area. |
| title | Training an LLM-as-a-Judge Model: Pipeline, Insights, and Practical Lessons |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2502.02988 |