Efficiently Building a Domain-Specific Large Language Model from Scratch: A Case Study of a Classical Chinese Large Language Model
This addresses the need for efficient domain-specific models in fields like ancient text collation and dictionary editing, though it is incremental as it builds on existing methods for a niche application.
The study tackled the problem of general-purpose large language models performing poorly in specific domains like Classical Chinese by developing a domain-specific model, AI Taiyan, which achieved results close to or surpassing human baselines in tasks such as punctuation and translation with only 1.8 billion parameters.
General-purpose large language models demonstrate notable capabilities in language comprehension and generation, achieving results that are comparable to, or even surpass, human performance in many natural language processing tasks. Nevertheless, when general models are applied to some specific domains, e.g., Classical Chinese texts, their effectiveness is often unsatisfactory, and fine-tuning open-source foundational models similarly struggles to adequately incorporate domain-specific knowledge. To address this challenge, this study developed a large language model, AI Taiyan, specifically designed for understanding and generating Classical Chinese. Experiments show that with a reasonable model design, data processing, foundational training, and fine-tuning, satisfactory results can be achieved with only 1.8 billion parameters. In key tasks related to language processing of Classical Chinese such as punctuation, identification of allusions, explanation of word meanings, and translation between ancient and modern Chinese, this model exhibits a clear advantage over both general-purpose large models and domain-specific traditional models, achieving levels close to or surpassing human baselines. This research provides a reference for the efficient construction of specialized domain-specific large language models. Furthermore, the paper discusses the application of this model in fields such as the collation of ancient texts, dictionary editing, and language research, combined with case studies.