nb steps in early stop (#3909)

* nb steps

* if

* skip

* rev

* seed

* seed

.github/workflows/ci_test-conda.yml

@@ -31,6 +31,7 @@ jobs:
         pip list
 
     - name: Cache datasets
+      # todo this probably does not work with docker images, rather cache dockers
       uses: actions/cache@v2
       with:
         path: Datasets # This path is specific to Ubuntu

.pyrightconfig.json

@@ -35,6 +35,7 @@
     "pytorch_lightning/trainer/connectors/checkpoint_connector.py",
     "pytorch_lightning/trainer/connectors/data_connector.py",
     "pytorch_lightning/trainer/connectors/logger_connector.py",
+    "pytorch_lightning/trainer/connectors/slurm_connector.py",
     "pytorch_lightning/distributed/dist.py",
     "pytorch_lightning/tuner",
     "pytorch_lightning/plugins"

pytorch_lightning/callbacks/early_stopping.py

@@ -205,6 +205,7 @@ def _run_early_stopping_check(self, trainer, pl_module):
 
     def on_train_end(self, trainer, pl_module):
         if self.stopped_epoch > 0 and self.verbose > 0:
+            # todo: remove this old warning
             rank_zero_warn('Displayed epoch numbers by `EarlyStopping` start from "1" until v0.6.x,'
                            ' but will start from "0" in v0.8.0.', DeprecationWarning)
             log.info(f'Epoch {self.stopped_epoch + 1:05d}: early stopping triggered.')

pytorch_lightning/trainer/connectors/checkpoint_connector.py

@@ -149,7 +149,7 @@ def restore_training_state(self, checkpoint):
         self.trainer.global_step = checkpoint['global_step']
         self.trainer.current_epoch = checkpoint['epoch']
 
-        # crash if max_epochs is lower than the current epoch from the checkpoint
+        # crash if max_epochs is lower then the current epoch from the checkpoint
         if self.trainer.current_epoch > self.trainer.max_epochs:
             m = f"""
             you restored a checkpoint with current_epoch={self.trainer.current_epoch}

tests/callbacks/test_early_stopping.py

@@ -6,7 +6,7 @@
 import pytest
 import torch
 
-from pytorch_lightning import Trainer
+from pytorch_lightning import Trainer, seed_everything
 from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint
 from tests.base import EvalModelTemplate
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
@@ -35,7 +35,7 @@ def test_resume_early_stopping_from_checkpoint(tmpdir):
     https://github.com/PyTorchLightning/pytorch-lightning/issues/1464
     https://github.com/PyTorchLightning/pytorch-lightning/issues/1463
     """
-
+    seed_everything(42)
     model = EvalModelTemplate()
     checkpoint_callback = ModelCheckpoint(monitor="early_stop_on", save_top_k=1)
     early_stop_callback = EarlyStoppingTestRestore()
@@ -60,7 +60,7 @@ def test_resume_early_stopping_from_checkpoint(tmpdir):
     early_stop_callback = EarlyStoppingTestRestore(early_stop_callback_state)
     new_trainer = Trainer(
         default_root_dir=tmpdir,
-        max_epochs=2,
+        max_epochs=1,
         resume_from_checkpoint=checkpoint_filepath,
         early_stop_callback=early_stop_callback,
     )