Code for the paper "On LLM-Enhanced Mixed-Type Data Imputation with High-Order Message Passing" which is submitted for reviewing in VLDB 2025.

Pre-training

bash ./scripts/run_linear_all.sh
bash ./scripts/run_LLM_all.sh

Finetune

bash ./scripts/run_finetune.sh

Re-produce Experiments

bash ./scripts/xxx.sh

Folder Structure

.
├── data                        # the folder containing all the datasets
├── models                      # the implementation of the model
├── logs                        # the running logs
├── scripts                     # the scripts to run the model and to reproduce the experiments
├── data_loader.py              # data loader
├── finetune.py                 # the entrance for the finetune mode
├── main.py                     # the overall entrance for the framework
├── training.py                 # the entrance for the training mode
├── testing.py                  # the entrance for the testing mode
└── README.md