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
I just find there is a mistake in your released DARTS code.
In architect.py, the hessian is computed as follows: hessian = [(p-n) / 2.*eps for p, n in zip(dalpha_pos, dalpha_neg)]
However, referring to paper and the official release code, it should be computed as follows: hessian = [(p-n) / (2.*eps) for p, n in zip(dalpha_pos, dalpha_neg)]
But it seems this mistake does not hurt the performance. LOL
The text was updated successfully, but these errors were encountered:
Hi, thanks for noticing. I corrected it. You may as well warn the original repo of DARTS that I refer to in the Readme though. I am not too surprised that the performance is not much affected :)
Hi,
I just find there is a mistake in your released DARTS code.
In architect.py, the hessian is computed as follows:
hessian = [(p-n) / 2.*eps for p, n in zip(dalpha_pos, dalpha_neg)]
However, referring to paper and the official release code, it should be computed as follows:
hessian = [(p-n) / (2.*eps) for p, n in zip(dalpha_pos, dalpha_neg)]
But it seems this mistake does not hurt the performance. LOL
The text was updated successfully, but these errors were encountered: