cerebras.modelzoo.data_preparation.nlp.bert.create_hdf5_files#

Script to write HDF5 files for MLM_only and MLM + NSP datasets.

Usage:

# For help related to MLM_only dataset creation python mlm_only -h

# For help related to MLM + NSP dataset creation python mlm_nsp -h

Functions

create_h5

create_h5_mp

main