LiquiLM

To ensure that the experimental results of LiquiLM can be fully reproduced, we have compiled and uploaded all relevant resources to this project. This includes the dataset used in the experiment, the complete experimental code, the hyperparameter settings used in the model training process, and detailed environment configuration information. By providing these key elements, we aim to promote the transparency and reproducibility of research, so that other researchers can understand, verify and further explore the performance and potential of LiquiLM. For details on experimental reproduction, please refer to README.md.