how can i reduce massive tsv file using the Dask library and transforme it to pandas dataframe

Question

i want to reduce a corpus of documents which is in a tsv file, because it connot fit in to a pandas.dataframe without loosing RAM. And if i used the dask dataframe it still a lo