Would training LLMs on either smaller batches of text first instead of the (current) large batches of text or on easier subjects/topics first then later train them on subjects/topics with increasing order of difficulty...enable the LLMs to learn better.…