chai _layer for all tasks

facebookresearch / chai

CHAI is a library for dynamic pruning of attention heads for efficient LLM inference.

GNU General Public License v3.0

9 stars 0 forks source link

Open rippleD030 opened 1 week ago

rippleD030 commented 1 week ago

Could you please clarify whether the number of clusters you chose for each layer is the same across all five tasks?

iidsample commented 4 days ago

Yes we keep the number of clusters across all five tasks same.