We propose DianJin-R1, a novel framework that enhances financial reasoning in LLMs through reasoning-augmented supervision and reinforcement learning. Central to our approach is DianJin-R1-Data, a high-quality dataset constructed from CFLUE, FinQA, and a proprietary compliance corpus (Chinese Compliance Check, CCC), combining diverse financial reasoning… See the full description on the dataset page: https://huggingface.co/datasets/DianJin/DianJin-R1-Data.
