Repolode
Submit
AI4Finance-Foundation

AI4Finance-Foundation/FinGPT

Open-source large language models fine-tuned for financial tasks such as sentiment analysis, forecasting and financial NLP benchmarks.

Open-source alternative to:Bloomberg Terminal

View on GitHubWebsite
Stars
21,436
Forks
3,036
Open issues
54
Last commit
28 days ago

What is it?

FinGPT is an open-source project for building financial large language models. It adapts general-purpose open LLMs such as Llama 2, ChatGLM2 and Qwen to finance using LoRA-based fine-tuning on financial news, tweets and instruction datasets, with the trained models published on Hugging Face.

The project covers several areas: financial sentiment analysis models, a multi-task model for relation extraction, headline classification and named-entity recognition, the FinGPT-Forecaster robo-advisor component that predicts weekly stock moves for Dow 30 tickers, and FinGPT-RAG and FinGPT-Benchmark for retrieval-augmented sentiment analysis and instruction-tuning evaluation. The FinNLP data pipeline is maintained in a separate repository.

The project's stated approach is low-cost, frequent re-tuning on new data rather than costly full retraining. Its README states that the code is shared for academic purposes and is not financial advice.

Key features

  • 01Financial sentiment analysis models fine-tuned with LoRA on several base LLMs.
  • 02Multi-task model covering relation extraction, headline classification and named-entity recognition.
  • 03FinGPT-Forecaster generates stock movement analysis and next-week predictions for Dow 30 tickers.
  • 04Instruction-tuning datasets for financial tasks published on Hugging Face.
  • 05Benchmark code for comparing open-source base models on financial sentiment tasks.

Use cases

  • Researchers studying LLM fine-tuning on financial text.
  • Classifying sentiment in financial news or social media posts.
  • Prototyping robo-advisor style analysis on US large-cap stocks.
  • Comparing open-source base models on financial NLP benchmarks.

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/AI4Finance-Foundation/FinGPT.git
  2. 2

    Enter the project directory

    cd FinGPT
  3. 3

    Install dependencies

    pip install -r requirements.txt
  4. 4

    Install the package in editable mode

    pip install -e .

Things to know

  • Local model use requires a GPU, and the README points to SETUP.md for hardware details.
  • FinGPT-Forecaster is trained on US market data (DOW 30) and is not designed for other markets such as China.

Topics

#llm#finance#sentiment-analysis#fine-tuning#lora#nlp#machine-learning#chatgpt#fingpt#fintech

FAQ

What is FinGPT?
Open-source large language models fine-tuned for financial tasks such as sentiment analysis, forecasting and financial NLP benchmarks. FinGPT is an open-source project for building financial large language models. It adapts general-purpose open LLMs such as Llama 2, ChatGLM2 and Qwen to finance using LoRA-based fine-tuning on financial news, tweets and instruction datasets, with the trained models published on Hugging Face.
Is FinGPT free to use? What license does it have?
FinGPT is open source under the MIT license. Check the license terms before using it in commercial products.
Is FinGPT still maintained?
Actively maintained. The last commit to FinGPT was on Sep 23, 2026.
What can I use FinGPT for?
Researchers studying LLM fine-tuning on financial text. Classifying sentiment in financial news or social media posts. Prototyping robo-advisor style analysis on US large-cap stocks. Comparing open-source base models on financial NLP benchmarks.
How do I install FinGPT?
Clone the repository: git clone https://github.com/AI4Finance-Foundation/FinGPT.git Enter the project directory: cd FinGPT Install dependencies: pip install -r requirements.txt Install the package in editable mode: pip install -e .
Where can I get FinGPT?
The source code and installation instructions are on GitHub: https://github.com/AI4Finance-Foundation/FinGPT

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects