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
Originally posted by Franklalalala October 8, 2021
I saw this feature in the code, loss function, which is exactly what i'm looking for. Since my data is not from md trajectory, most of the atoms have a extremely low force. Apply this "atom prefernce" feature may greatly improve the force prediction.
I've seen this. #358
And checked my data with this #709
I've tried this feature on my cpu version, the results looks acceptable. However, I can't run it with GPU version. BUG is: ArgumentKeyError: [at location loss] undefined key start_pref_pf is not allowed in strict mode.
What can i do to implement this feature with GPU? Lower my version? If so, which version? Thanks!
The text was updated successfully, but these errors were encountered:
Discussed in #1197
Originally posted by Franklalalala October 8, 2021
I saw this feature in the code, loss function, which is exactly what i'm looking for. Since my data is not from md trajectory, most of the atoms have a extremely low force. Apply this "atom prefernce" feature may greatly improve the force prediction.
I've seen this.
#358
And checked my data with this
#709
I've tried this feature on my cpu version, the results looks acceptable. However, I can't run it with GPU version. BUG is: ArgumentKeyError: [at location
loss
] undefined keystart_pref_pf
is not allowed in strict mode.What can i do to implement this feature with GPU? Lower my version? If so, which version? Thanks!
The text was updated successfully, but these errors were encountered: