lob/                    Source code for LOB models, datasets, etc.
    ar_pred.ipynb           Test set evaluations and plotting for trained model.
    dataloading.py          Dataloading functions.
    encoding.py             Message tokenization: encoding and decoding
    evaluation.py           Model evaluation logic
    inference.py            Logic for model inference loop
    init_train.py           Train state initialisation
    lob_seq_model.py        Defines LOB deep sequence model that consist of stacks of S5
    lobster_dataloader.py   Defines dataset and dataloading
    preproc.py              Pre-processes LOBSTER data for the model
    run_eval.py             Script to run model inference with trained model
    sweep.py                Hyperparamter sweep (WanDB)
    train_helpers.py        Functions for optimization, training and evaluation steps.
    train.py                Training loop code.
    validation_helpers.py   Helper functions for model validation
s5/                     Original S5 code
bin/                    Shell scripts for downloading data and running experiments.
requirements.txt            Package requirements
run_train.py            Training loop entrypoint.

@article{nagy2023generative,
  doi = {},
  url = {https://arxiv.org/abs/xxxx.xxxxx},
  author = {Nagy, Peer and Frey, Sascha and Sapora, Silvia and Li, Kang and Calinescu, Anisoara and Zohren, Stefan and Foerster, Jakob},
  keywords = {generative AI, structured state space models, limit order books, ML},
  title = {Generative AI for End-to-End Limit Order Book Modelling: A Token-Level Autoregressive Generative Model of Message Flow Using a Deep State Space Network},
  publisher = {arXiv},
  year = {2023},
  copyright = {Creative Commons Attribution 4.0 International}
}