Utilized the FinGPT API from Hugging Face for direct inference and reported test results. Benchmarks
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
Conducted experiments with multiple base models for LoRA training, including InternLM, Falcon, llama2, and ChatGLM2. Compared their respective performances.