Thai Financial Domain Adaptation of THaLLE -- Technical Report
This addresses the problem of domain-specific LLM adaptation for Thai financial professionals, though it is incremental as it applies existing methods to a new domain.
The paper tackled the lack of Thai financial domain support in large language models by developing a model using the Investment Consultant exam dataset, achieving scores of 72%, 72%, and 84% on exam levels P1, P2, and P3.
Large Language Models (LLMs) excel in general tasks but struggle with domain-specific challenges, such as specialized terminology and localized regulations. Existing financial LLMs, like FinGPT and BloombergGPT, lack support for the Thai financial domain. We developed a Thai Financial LLM using the Investment Consultant (IC) exam dataset from the Stock Exchange of Thailand. To address dataset limitations, we applied data augmentation, ReLoRA for efficient training, Continued Pretraining (CPT) for domain knowledge, and Rank-Stabilized LoRA (rsLoRA) for fine-tuning. Supervised Fine-Tuning (SFT) simulated exam scenarios, while Direct Preference Optimization (DPO) refined the model using feedback. The model achieved scores of 72%, 72%, and 84% on IC exam levels P1, P2, and P3, respectively, demonstrating its effectiveness in Thai financial advisory tasks and its potential for specialized applications.