Is the Reproducibility problem fixed for this repo, because i am facing similar problem when i tried with same training procedure that is mentioned in README
Issue 01:
Script used - train_classifier.py
I am able to achieve Batch accuracy of ~31% for Training and ~37% for validation.
Issue 02:
Script used - train_features.py
I am able to achieve Loss of ~23% for Training and ~22% for validation.
Can you please help me with solution or is this the expected issue. even i am unable to reproduce the results for CIFAR that you have achieved so far
Hi @mdiephuis,
Is the Reproducibility problem fixed for this repo, because i am facing similar problem when i tried with same training procedure that is mentioned in README
Issue 01:
Script used - train_classifier.py I am able to achieve Batch accuracy of ~31% for Training and ~37% for validation.
Issue 02:
Script used - train_features.py I am able to achieve Loss of ~23% for Training and ~22% for validation.
Can you please help me with solution or is this the expected issue. even i am unable to reproduce the results for CIFAR that you have achieved so far