中文

FinGPT: Open-Source Financial Large Language Models

FinGPT is an open-source financial large language model project developed and maintained by the AI4Finance Foundation.

Verified

Core facts

Read by a script from the official API, not rewritten by a model.

License
MIT
Languages
Jupyter Notebook
Maintainer
AI4Finance-Foundation
Verified

What it is

FinGPT is an open-source financial large language model project developed and maintained by the AI4Finance Foundation. FinGPT embraces a full-stack framework for FinLLMs with five layers. FinGPT-RAG is a retrieval-augmented large language model framework designed for financial sentiment analysis. FinGPT-FinNLP provides full pipelines for LLM training and finetuning in the field of finance.

Main features

  1. The README says the data source layer assures comprehensive market coverage, addressing the temporal sensitivity of financial data through real-time information capture.

  2. The data engineering layer is primed for real-time NLP data processing and tackles the inherent challenges of high temporal sensitivity and low signal-to-noise ratio in financial data.

  3. The LLMs layer focuses on a range of fine-tuning methodologies such as LoRA.

  4. The task layer is responsible for executing fundamental tasks.

  5. The application layer showcases practical applications and demos, highlighting the potential capability of FinGPT in the financial sector.

  6. FinGPT-Benchmark introduces an Instruction Tuning paradigm optimized for open-source large language models in finance.

Getting started

  1. The README refers to detailed setup instructions for running FinGPT locally or on Replit, including hardware requirements and troubleshooting.

  2. The README points to a released FinGPT-Forecaster demo.

  3. FinGPT v3.3, v3.2 and v3.1 use llama2-13b, llama2-7b and chatglm2-6B as base models, respectively.

  4. A reader can finetune their own FinGPT v3 model with the LoRA method on only an RTX 3090.

  5. For the Finogrid platform, FinGPT supports multiple cloud LLM providers as alternatives to running local models.

From the README · bash
# Clone the repository
git clone https://github.com/AI4Finance-Foundation/FinGPT.git
cd FinGPT

# Install dependencies
pip install -r requirements.txt
pip install -e .
From the README · python
# Using cloud API (no GPU required)
import os
os.environ['FINGPT_LLM_PROVIDER'] = 'openai'

# Using local models (requires GPU)
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# See SETUP.md for detailed examples

Who it is for

FinGPT-Forecaster is currently trained on US market data (DOW 30). The README says to consider using FinGPT v1.x models, which were designed for Chinese markets. The README says RLHF enables an LLM to learn individual preferences, such as risk-aversion level, investing habits and personalized robo-advisor.

The README says retraining an LLM like BloombergGPT every month or every week is costly, so lightweight adaptation is highly favorable. The README says democratizing Internet-scale financial data is critical, for example allowing timely updates of the model (monthly or weekly updates) using an automatic data curation pipeline. The README says finance is highly dynamic.

FinGPT is under the MIT License. The project shares codes for academic purposes under the MIT education license. The README content is not financial advice and not a recommendation to trade real money.

Sources

This page is written from the official sources below. Open them to check the original text.

Verified ·

This page is written by AI from the official sources above and checked automatically. This site has not tested it. Found a mistake? Tell us.