Model training does not stop when accuracy reaches a particular threshold

[ai_curious]

These two issues are related…

Week 2: Implementing Callbacks in TensorFlow using the MNIST Dataset

Ok, but maybe that always keeps the rule from firing, because logs.get(‘acc’) is always None. Are you ever seeing the early stopping message? I think I have my answer… https://community.deeplearning.ai/t/model-training-does-not-stop-when-accuracy-reaches-a-particular-threshold/271554

You didn’t actually fix that problem, merely masked it, which creates this problem.