Verifiable Format Control for Large Language Model Generations
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929702136971264 |
|---|---|
| author | Wang, Zhaoyang Jiang, Jinqi Zhou, Huichi Zheng, Wenhao Zhang, Xuchao Bansal, Chetan Yao, Huaxiu |
| author_facet | Wang, Zhaoyang Jiang, Jinqi Zhou, Huichi Zheng, Wenhao Zhang, Xuchao Bansal, Chetan Yao, Huaxiu |
| contents | Recent Large Language Models (LLMs) have demonstrated satisfying general instruction following ability. However, small LLMs with about 7B parameters still struggle fine-grained format following (e.g., JSON format), which seriously hinder the advancements of their applications. Most existing methods focus on benchmarking general instruction following while overlook how to improve the specific format following ability for small LLMs. Besides, these methods often rely on evaluations based on advanced LLMs (e.g., GPT-4), which can introduce the intrinsic bias of LLMs and be costly due to the API calls. In this paper, we first curate a fully verifiable format following dataset VFF. In contrast to existing works often adopting external LLMs for instruction-following validations, every sample of VFF can be easily validated with a Python function. Further, we propose to leverage this verifiable feature to synthesize massive data for progressively training small LLMs, in order to improve their format following abilities. Experimental results highlight the prevalent limitations in the format following capabilities of 7B level open-source LLMs and demonstrate the effectiveness of our method in enhancing this essential ability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_04498 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Verifiable Format Control for Large Language Model Generations Wang, Zhaoyang Jiang, Jinqi Zhou, Huichi Zheng, Wenhao Zhang, Xuchao Bansal, Chetan Yao, Huaxiu Computation and Language Recent Large Language Models (LLMs) have demonstrated satisfying general instruction following ability. However, small LLMs with about 7B parameters still struggle fine-grained format following (e.g., JSON format), which seriously hinder the advancements of their applications. Most existing methods focus on benchmarking general instruction following while overlook how to improve the specific format following ability for small LLMs. Besides, these methods often rely on evaluations based on advanced LLMs (e.g., GPT-4), which can introduce the intrinsic bias of LLMs and be costly due to the API calls. In this paper, we first curate a fully verifiable format following dataset VFF. In contrast to existing works often adopting external LLMs for instruction-following validations, every sample of VFF can be easily validated with a Python function. Further, we propose to leverage this verifiable feature to synthesize massive data for progressively training small LLMs, in order to improve their format following abilities. Experimental results highlight the prevalent limitations in the format following capabilities of 7B level open-source LLMs and demonstrate the effectiveness of our method in enhancing this essential ability. |
| title | Verifiable Format Control for Large Language Model Generations |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2502.04498 |