Please follow the command to prepare dependency.

conda create -n canbert python=3.12
conda activate canbert
pip install -r requirements.txt

It is not necessary to use conda environment.

It needs to prepare the data.

Go to model and execute the following command.

python gen_vocab.py -i /path/to/data/

The next step is to pretrain the model. It will take few hours.

python can_ids.py -i /path/to/data/orig.csv

With the pretrained model, it can fine-tune it based on new datasets.

python can_finetune.py -i /path/to/data/mcan/g80_mcan_fuzzing_data.csv
python can_finetune.py -i /path/to/data/mcan/g80_mcan_ddos_data.csv

When the size of data set is too small, it may be necessary to increase patience to obtain reliable result.

The evaluated datasets are not collected by authors. Hence, please cite each dataset when you are planning to use it.