Fewshot results

[halean]

Hello,
I have found the VMLU benchmark very useful and informative, thank you very much for developing the dataset and hosting the leaderboard. I am wondering about the inclusion of fewshot learning results in the public leaderboard. My experiments indicate that fewshot learning would add about 2 to 3% to the performance of these models, and it would be great to see whether and how the leaderboard would change in the fewshot learning settings.