TY  - RPRT
TI  - Constructing Synthetic Instruction Datasets for Improving Reasoning in Domain-Specific LLMs: A Case Study in the Japanese Financial Domain
AU  - Yuma Okochi
AU  - Fabio Milentiansen Sim
AU  - Tomoyasu Okada
PY  - 2026
UR  - https://arxiv.org/abs/2603.01353
ID  - 2603.01353
ER  -