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
