Fine-tuned from the base model (Llama-3.1-8B), supervised fine-tuned with abundant crypto fundamental knowledge and popular projects.