Financial domain-specific model built upon Baichuan4-Turbo using continual pre-training with a novel domain self-constraint training strategy. Integrates over 100 billion pieces of bilingual financial knowledge, developed in collaboration with Renmin University. Surpasses GPT-4o in financial task accuracy after SFT and RLHF/RLAIF.

Outputs 2

Baichuan4-Finance Technical Report

paper

Technical report describing the domain self-constraint training strategy, financial data pipeline, and evaluation results against competitive baselines.

arXiv: 2412.15270

Baichuan4-Finance (model)

model

Finance-optimized proprietary model with base and chat variants for financial certification questions and real-world scenario applications.

proprietaryfinancenlp