FinGPT provides open-source financial large language models and benchmarks that can be fine-tuned at low cost. It addresses the prohibitive expense of training financial models from scratch and the inaccessibility of institutional financial data.
Project overview
Fine-tuning costs fall under $300, offering an accessible alternative to training financial models from scratch which can cost millions.
Project type
RAG · Model Development · Model Runtime · Data Processing
Use cases
Chat Assistants · Knowledge Q&A · Data Analysis
Deployment
Refer to project documentation
License
MIT
Best for
AI engineers, data teams, and researchers developing or fine-tuning models for financial tasks such as relation extraction, headline classification, and sentiment analysis.
Key capabilities
Analyzes market news and basics to predict next week's stock price movement for a given ticker.
Provides models for financial relation extraction, headline classification, and named entity recognition.
Enables lightweight adaptation and fine-tuning of base open-source LLMs for new data at low cost.
A framework designed for financial sentiment analysis, optimizing information depth through external knowledge retrieval.
Limitations and risks
The project and its outputs are not financial advice and do not constitute a recommendation to trade real money.
Getting started
Clone the repository, install dependencies using pip install -r requirements.txt and pip install -e ., configure the environment, and run Python scripts.
Evidence and sources
GitHub project description: FinGPT: Open-Source Financial Large Language Models! Revolutionize 🔥 We release the trained model on HuggingFace.
README: FinGPT can be fine-tuned swiftly to incorporate new data (the cost falls significantly, less than **$300 per fine-tuning**).
README: [Model Release] Nov, 2023: We release [FinGPT-Forecaster](https://github.com/AI4Finance-Foundation/FinGPT/tree/master/fingpt/FinGPT_Forecaster)! 🔥[Demo](https://huggingface.co/spa…
README: git clone https://github.com/AI4Finance-Foundation/FinGPT.git cd FinGPT
README: Best trainable and inferable FinGPT for sentiment analysis on a single RTX 3090, which is even better than GPT-4 and ChatGPT Finetuning.