emergent_in_context_learning icon indicating copy to clipboard operation
emergent_in_context_learning copied to clipboard

Published 20 hours ago • verified publisher icon google-deepmind

Have you experiment with smaller size model

Open La-SilverLand opened this issue 2 years ago • 0 comments

Hi, in the paper, a transformer model with 12 layers and embdding size 64 is used to validate your hypothesis. Did you do any trial experiment on smaller sized model ? and if you did, what's the result ?

La-SilverLand avatar Feb 10 '23 07:02 La-SilverLand