UTCM

We conducted experiments using the LLaMA-Factory framework on several NVIDIA A800 80G GPUs. The environment configuration is consistent with LLaMA-Factory. The code is primarily divided into two parts: an entry script (.sh files) and model configuration files (.yaml files). The entry script is located in the main directory, while the configuration files are stored under the path ./Configs, including ./Configs/pt, ./Configs/sft and ./Configs/predict, which are used for incremental pretraining ,supervised fine-tuning and testing(predicting), respectively. Additionally, some of the test(predict) configuration files are also placed under ./Configs/sft.