The behavior of EarlyStopping is not the same between 0.7.5 with 0.7.3

[nickyi1990]

🐛 Bug

The behavior of EarlyStopping is not the same between 0.7.5 with 0.7.3, If i set patience=1, 0.7.3 will print two epoch when the second is worse than the first, but 0.7.5 will just print the first which is not consistent with the description of patience parameter

To Reproduce

Code sample

early_stop_callback = EarlyStopping(monitor='valid_roc_auc', patience=0, mode='max', verbose=True)
# Instance Model, Trainer and train model
model = PlantModel(hparams)
trainer = pl.Trainer(gpus=[hparams.gpus],
                     max_epochs=hparams.max_epochs,
                     early_stop_callback=early_stop_callback,
                     progress_bar_refresh_rate=0,
                     num_sanity_val_steps=0,
                     profiler=True,
                     gradient_clip_val=hparams.gradient_clip_val)
trainer.fit(model)
print(trainer.early_stop_callback.wait)

Expected behavior

Environment

• CUDA:
  - GPU:
  - GeForce RTX 2080 Ti
  - GeForce RTX 2080 Ti
  - GeForce RTX 2080 Ti
  - GeForce RTX 2080 Ti
  - available: True
  - version: 10.1
• Packages:
  - numpy: 1.18.1
  - pyTorch_debug: False
  - pyTorch_version: 1.4.0
  - pytorch-lightning: 0.7.5
  - tensorboard: 2.2.1
  - tqdm: 4.42.1
• System:
  - OS: Linux
  - architecture:
  - 64bit
  -
  - processor: x86_64
  - python: 3.7.6
  - version: Fixed typo in single_cpu_template #38~18.04.1-Ubuntu SMP Tue Mar 31 04:17:56 UTC 2020

Additional context

[williamFalcon]

@jeremyjordan

[jeremyjordan]

yes i believe this is the same issue reported in this comment and will be addressed in #1504

[nickyi1990]

reported

Yes, exactly the same problem, It seems the only way to use 0.7.5 is to change the source code my self😢