A mixture of datasets focused on Source/Intermediate Representation training for multilingual code generation. For SCU CSEN346.
Usage: Pass to continued_pretrain.py with
--dataset_name './tokenized_dataset'
Sources: UKPLab/SLTrans bigcode/the-stack allenai/peS2o open-web-math/open-web-math