In this research, we tackle the fundamental challenge of accurately measuring dataset diversity for instruction tuning and introduce NovelSum, a reliable diversity metric that jointly accounts for ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results