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
No, it doesn't seem to be official, I'm not sure why that account on huggingface has it.
However, it is the expected size and even includes special layers (SwiGLUFFNFused) that are only meant to be present in the vitg model. As far as I can tell, it also produces predictions that are as good or better than the vit-large model.
So it seems like either someone independently trained their own vitg model that's competitive with the official releases, or it's some (bootleg?) copy of the original model.
Hi, I wonder when will the Depth-Anything-V2-Giant model release? Thanks a lot!