Prepare new data for fine tuning? about siatl HOT 1 CLOSED

Check this line here https://github.com/alexandra-chron/siatl/blob/master/models/run_clf.py#L26

You need to provide the dataset in the following form:

dataset (list of 4 lists) = [X_train (list of str), y_train (list of int), X_val (list of str), y_val (list of int)] .

y_train, y_val can also be lists of str in the form ["positive", "positive", "neutral", "negative", ...] etc.

So X_train are your training sentences in the form of strings, y_train are your labels, etc. The way that you parse it (if it originally was a .csv, a .pkl, .json or a different form of input file is not important - meaning that you can parse any file, you will just need to write a different load_dataset() for your type of input file.

from siatl.