AI4Finance-Foundation/FinGPT

Built on lightweight LoRA fine-tuning of open-source base models (Llama 2, ChatGLM), FinGPT enables rapid, cost-effective adaptation to financial datasets—retraining in hours for under $300 versus weeks and millions for full model training. The framework incorporates reinforcement learning from human feedback (RLHF) to capture individual investor preferences and risk profiles for personalized robo-advisory. Applications span sentiment analysis, stock forecasting, and multi-task financial reasoning across automatically curated internet-scale data pipelines.

18,815 stars and 185 monthly downloads. Actively maintained with 18 commits in the last 30 days. Available on PyPI.