Continue training when validation loss below threshold

IBM / mi-prometheus

Enabling reproducible Machine Learning research

http://mi-prometheus.rtfd.io/

Apache License 2.0

42 stars 18 forks source link

Open tkornuta-ibm opened 5 years ago

tkornuta-ibm commented 5 years ago

Consider actually NOT stopping when "converged", this will require dividing loss_stop into two parameters:

loss_threshold (DEFAULT: 1e-5) - indicating the threshold
loss_stop (DEFAULT: True or False?) - indicating whether we want to stop or continue training