Leveraging Sequential Nature: Large Language Models for Sequential Recommendation

Environment Setup

To replicate our environment, please use the provided requirements.txt file.

To set up the environment, run the following command:

pip install -r requirements.txt

Training & Evaluation

1. First, update the paths in the training scripts (llm_path, data_dir, ckpt_dir, and log_dir) with your own folder paths.
2. To train the model using a single A100 GPU, run the following command:

sh train_lastfm.sh

3. Once the model is trained, you can test it by updating the necessary paths in the test scripts and running:

sh test_lastfm.sh

For other shell files, you can train other datasets.

Dataset

The data folder contains three datasets that can be easily utilized for training and evaluation.