Please describe the use case that requires this feature.
For transfer learning, one might want to resume from trained weights and store after every epoch rather than only when the validation loss improved.
Describe the solution you'd like
Introduce config parameter that is passed to the checkpoint callback,
TjarkMiener
commented
1 month ago
Please describe the use case that requires this feature. For transfer learning, one might want to resume from trained weights and store after every epoch rather than only when the validation loss improved.
Describe the solution you'd like Introduce config parameter that is passed to the checkpoint callback,