AI4Finance-Foundation / FinGPT-Research

FinGPT-AI-Research

Replicating FinGPT - Supervised Fine-tuning (SFT)

Inference

Utilized the FinGPT API from Hugging Face for direct inference and reported test results. Benchmarks

Benchmark Results for Financial Sentiment Analysis Tasks:

Employed a base model(ChatGLM2-6B) to follow the LoRA training process. Compared the performance of the custom-trained model with the direct inference results. LoRA with ChatGLM2-6B

Base Model Comparison

Conducted experiments with multiple base models for LoRA training, including InternLM, Falcon, llama2, and ChatGLM2. Compared their respective performances.

Articles of My Work

• SFT Research Report

• LoRA, Financial Sentimental Analysis with Baichuan2–7B

• Beginner’s Guide to FinGPT: Training with LoRA and ChatGLM2–6B

• Replicate FinGPT Sentiment Analysis Task with Google Colab

FinGPT

• GitHub

• Hugging Face