Script for data generation #2
Description
Hi,
Thank you for your work! The paper and the implementation are incredibly interesting. I noticed that you provided a constructed dataset of 12,000 samples in the project, with a maximum length of 3,000 tokens for each sample.
However, for my experimental needs, I need to generate data with longer token lengths. Could you kindly provide the script used for data generation or explain how to modify the existing code to generate longer samples?