For the train.py
and predict.py
scripts it is possible setup readers:
JSONL
utilized by default:
from arenets.arekit.contrib.utils.data.readers.jsonl import JsonlReader
reader = JsonlReader()
CSV
reader: requires installation of thepandas
:
from arenets.external.readers.pandas_csv_reader import PandasCsvReader
reader = PandasCsvReader(sep=',', # Column separator
target_extension=".csv") # Input file type