In running 03_continued_pretraining_titan_text.ipynb in 03_Model_customization module, I couldn't properly finish training, and saw an error message of "Maximum input token count 4732 exceeds limit of 4096 for train data" on the console. please kindly advise ways to mitigate or help fix the dataset used for the training job as part of the workshop artifact. thanks.
In running 03_continued_pretraining_titan_text.ipynb in 03_Model_customization module, I couldn't properly finish training, and saw an error message of "Maximum input token count 4732 exceeds limit of 4096 for train data" on the console. please kindly advise ways to mitigate or help fix the dataset used for the training job as part of the workshop artifact. thanks.