You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It's probably reduced from decoupled head, not relative to baseline. From green numbers you can see that each model is compared to the previous one, so I assume they are built on top of the previous one. So it's likely "+decoupled head + strong augmentation + anchor-free" which is 63.72M params.
Why in the paper state that the param and GLOPS reduce with anchor free but the table states inversely?
The text was updated successfully, but these errors were encountered: