Refactor callbacks (#776)

kuynzereb · web-flow · commit edd4a87fb0e9 · 2020-02-16T00:03:05.000-05:00
* Refactor callbacks

* flake8

* Update docstrings

* Simplified callback, protected trainer

* .set_trainer() check

* update docs

* missed super().__ini__()

* Updated tests

* Use uppercase

* refine checkpoint callback tests

* Added test_begin() and test_end()
diff --git a/pytorch_lightning/callbacks/pt_callbacks.py b/pytorch_lightning/callbacks/pt_callbacks.py
@@ -12,60 +12,62 @@
 
 import numpy as np
 
-from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-
 
 class Callback(object):
-    r"""Abstract base class used to build new callbacks.
-    """
+    """Abstract base class used to build new callbacks."""
 
     def __init__(self):
-        self.validation_data = None
-        self.model = None
-
-    def set_params(self, params):
-        self.params = params
+        self._trainer = None
 
-    def set_model(self, model):
-        if isinstance(model, LightningDistributedDataParallel):
-            model = model.module
-        self.model = model
+    def set_trainer(self, trainer):
+        """Make a link to the trainer, so different things like `trainer.current_epoch`,
+        `trainer.batch_idx`, `trainer.global_step` can be used."""
+        self._trainer = trainer
 
-    def on_epoch_begin(self, epoch, logs=None):
-        """
-        called when the epoch begins
+    def on_epoch_begin(self):
+        """Called when the epoch begins."""
+        pass
 
-        Args:
-            epoch (int): current epoch
-            logs (dict): key-value pairs of quantities to monitor
+    def on_epoch_end(self):
+        """Called when the epoch ends."""
+        pass
 
-            Example:
+    def on_batch_begin(self):
+        """Called when the training batch begins."""
+        pass
 
-                on_epoch_begin(epoch=2, logs={'val_loss': 0.2})
-        """
+    def on_batch_end(self):
+        """Called when the training batch ends."""
+        pass
 
-    def on_epoch_end(self, epoch, logs=None):
+    def on_train_begin(self):
+        """Called when the train begins."""
         pass
 
-    def on_batch_begin(self, batch, logs=None):
-        """
-        called when the batch starts.
+    def on_train_end(self):
+        """Called when the train ends."""
+        pass
 
-        Args:
-            batch (Tensor): current batch tensor
-            logs (dict): key-value pairs of quantities to monitor
-        """
+    def on_validation_begin(self):
+        """Called when the validation loop begins."""
+        pass
 
-    def on_batch_end(self, batch, logs=None):
+    def on_validation_end(self):
+        """Called when the validation loop ends."""
         pass
 
-    def on_train_begin(self, logs=None):
+    def on_test_begin(self):
+        """Called when the test begins."""
         pass
 
-    def on_train_end(self, logs=None):
+    def on_test_end(self):
+        """Called when the test ends."""
         pass
 
 
+_NO_TRAINER_ERROR_MSG = ".set_trainer() should be called after the callback initialization"
+
+
 class EarlyStopping(Callback):
     r"""
     Stop training when a monitored quantity has stopped improving.
@@ -148,13 +150,16 @@ def check_metrics(self, logs):
 
         return True
 
-    def on_train_begin(self, logs=None):
+    def on_train_begin(self):
         # Allow instances to be re-used
         self.wait = 0
         self.stopped_epoch = 0
         self.best = np.Inf if self.monitor_op == np.less else -np.Inf
 
-    def on_epoch_end(self, epoch, logs=None):
+    def on_epoch_end(self):
+        assert self._trainer is not None, _NO_TRAINER_ERROR_MSG
+
+        logs = self._trainer.callback_metrics
         stop_training = False
         if not self.check_metrics(logs):
             return stop_training
@@ -166,13 +171,13 @@ def on_epoch_end(self, epoch, logs=None):
         else:
             self.wait += 1
             if self.wait >= self.patience:
-                self.stopped_epoch = epoch
+                self.stopped_epoch = self._trainer.current_epoch
                 stop_training = True
                 self.on_train_end()
 
         return stop_training
 
-    def on_train_end(self, logs=None):
+    def on_train_end(self):
         if self.stopped_epoch > 0 and self.verbose > 0:
             warnings.warn('Displayed epoch numbers by `EarlyStopping` start from "1" until v0.6.x,'
                           ' but will start from "0" in v0.8.0.', DeprecationWarning)
@@ -306,8 +311,11 @@ def check_monitor_top_k(self, current):
             return True
         return self.monitor_op(current, self.best_k_models[self.kth_best_model])
 
-    def on_epoch_end(self, epoch, logs=None):
-        logs = logs or {}
+    def on_validation_end(self):
+        assert self._trainer is not None, _NO_TRAINER_ERROR_MSG
+
+        logs = self._trainer.callback_metrics
+        epoch = self._trainer.current_epoch
         self.epochs_since_last_check += 1
 
         if self.save_top_k == 0:
@@ -389,6 +397,8 @@ class GradientAccumulationScheduler(Callback):
     """
 
     def __init__(self, scheduling: dict):
+        super().__init__()
+
         if scheduling == {}:  # empty dict error
             raise TypeError("Empty dict cannot be interpreted correct")
 
@@ -408,21 +418,14 @@ def __init__(self, scheduling: dict):
         self.scheduling = scheduling
         self.epochs = sorted(scheduling.keys())
 
-    def on_epoch_begin(self, epoch, trainer):
+    def on_epoch_begin(self):
+        assert self._trainer is not None, _NO_TRAINER_ERROR_MSG
+
+        trainer = self._trainer
         # indexing epochs from 1 (until v0.6.x)
-        # In v0.8.0, `epoch += 1` should be removed.
-        epoch += 1
+        # In v0.8.0, ` + 1` should be removed.
+        epoch = trainer.current_epoch + 1
         for i in reversed(range(len(self.epochs))):
             if epoch >= self.epochs[i]:
                 trainer.accumulate_grad_batches = self.scheduling.get(self.epochs[i])
                 break
-
-
-# if __name__ == '__main__':
-#     c = EarlyStopping(min_delta=0.9, patience=2, verbose=True)
-#     losses = [10, 9, 8, 8, 6, 4.3, 5, 4.4, 2.8, 2.5]
-#     for i, loss in enumerate(losses):
-#         should_stop = c.on_epoch_end(i, logs={'val_loss': loss})
-#         log.info(loss)
-#         if should_stop:
-#             break
diff --git a/pytorch_lightning/trainer/callback_config.py b/pytorch_lightning/trainer/callback_config.py
@@ -48,6 +48,9 @@ def configure_checkpoint_callback(self):
             # if checkpoint callback used, then override the weights path
             self.weights_save_path = self.checkpoint_callback.filepath
 
+            # link to the trainer
+            self.checkpoint_callback.set_trainer(self)
+
         # if weights_save_path is still none here, set to current working dir
         if self.weights_save_path is None:
             self.weights_save_path = self.default_save_path
@@ -77,3 +80,6 @@ def configure_early_stopping(self, early_stop_callback):
         else:
             self.early_stop_callback = early_stop_callback
             self.enable_early_stop = True
+
+        if self.early_stop_callback is not None:
+            self.early_stop_callback.set_trainer(self)
diff --git a/pytorch_lightning/trainer/evaluation_loop.py b/pytorch_lightning/trainer/evaluation_loop.py
@@ -330,8 +330,7 @@ def run_evaluation(self, test=False):
 
         # model checkpointing
         if self.proc_rank == 0 and self.checkpoint_callback is not None and not test:
-            self.checkpoint_callback.on_epoch_end(epoch=self.current_epoch,
-                                                  logs=self.callback_metrics)
+            self.checkpoint_callback.on_validation_end()
 
     def evaluation_forward(self, model, batch, batch_idx, dataloader_idx, test=False):
         # make dataloader_idx arg in validation_step optional
diff --git a/pytorch_lightning/trainer/training_loop.py b/pytorch_lightning/trainer/training_loop.py
@@ -328,7 +328,7 @@ def train(self):
             self.main_progress_bar.set_description(desc)
 
             # changing gradient according accumulation_scheduler
-            self.accumulation_scheduler.on_epoch_begin(epoch, self)
+            self.accumulation_scheduler.on_epoch_begin()
 
             # -----------------
             # RUN TNG EPOCH
@@ -352,8 +352,7 @@ def train(self):
             met_min_epochs = epoch >= self.min_epochs - 1
             if (self.enable_early_stop and not self.disable_validation and is_val_epoch and
                     (met_min_epochs or self.fast_dev_run)):
-                should_stop = self.early_stop_callback.on_epoch_end(epoch=epoch,
-                                                                    logs=self.callback_metrics)
+                should_stop = self.early_stop_callback.on_epoch_end()
                 # stop training
                 stop = should_stop and met_min_epochs
                 if stop:
diff --git a/pytorch_lightning/trainer/training_tricks.py b/pytorch_lightning/trainer/training_tricks.py
@@ -39,3 +39,5 @@ def configure_accumulated_gradients(self, accumulate_grad_batches):
             self.accumulation_scheduler = GradientAccumulationScheduler(schedule)
         else:
             raise TypeError("Gradient accumulation supports only int and dict types")
+
+        self.accumulation_scheduler.set_trainer(self)
diff --git a/tests/test_trainer.py b/tests/test_trainer.py
@@ -229,10 +229,16 @@ def mock_save_function(filepath):
 
     # -----------------
     # CASE K=-1  (all)
-    w = ModelCheckpoint(save_dir, save_top_k=-1, verbose=1)
-    w.save_function = mock_save_function
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=-1, verbose=1)
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for i, loss in enumerate(losses):
-        w.on_epoch_end(i, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = i
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = set(os.listdir(save_dir))
 
@@ -247,10 +253,16 @@ def mock_save_function(filepath):
 
     # -----------------
     # CASE K=0 (none)
-    w = ModelCheckpoint(save_dir, save_top_k=0, verbose=1)
-    w.save_function = mock_save_function
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=0, verbose=1)
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for i, loss in enumerate(losses):
-        w.on_epoch_end(i, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = i
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = os.listdir(save_dir)
 
@@ -261,10 +273,16 @@ def mock_save_function(filepath):
 
     # -----------------
     # CASE K=1 (2.5, epoch 4)
-    w = ModelCheckpoint(save_dir, save_top_k=1, verbose=1, prefix='test_prefix')
-    w.save_function = mock_save_function
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=1, verbose=1, prefix='test_prefix')
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for i, loss in enumerate(losses):
-        w.on_epoch_end(i, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = i
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = set(os.listdir(save_dir))
 
@@ -278,11 +296,17 @@ def mock_save_function(filepath):
     # CASE K=2 (2.5 epoch 4, 2.8 epoch 2)
     # make sure other files don't get deleted
 
-    w = ModelCheckpoint(save_dir, save_top_k=2, verbose=1)
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=2, verbose=1)
     open(f'{save_dir}/other_file.ckpt', 'a').close()
-    w.save_function = mock_save_function
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for i, loss in enumerate(losses):
-        w.on_epoch_end(i, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = i
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = set(os.listdir(save_dir))
 
@@ -298,10 +322,16 @@ def mock_save_function(filepath):
     # CASE K=4 (save all 4 models)
     # multiple checkpoints within same epoch
 
-    w = ModelCheckpoint(save_dir, save_top_k=4, verbose=1)
-    w.save_function = mock_save_function
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=4, verbose=1)
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for loss in losses:
-        w.on_epoch_end(0, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = 0
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = set(os.listdir(save_dir))
 
@@ -314,10 +344,16 @@ def mock_save_function(filepath):
     # CASE K=3 (save the 2nd, 3rd, 4th model)
     # multiple checkpoints within same epoch
 
-    w = ModelCheckpoint(save_dir, save_top_k=3, verbose=1)
-    w.save_function = mock_save_function
+    checkpoint_callback = ModelCheckpoint(save_dir, save_top_k=3, verbose=1)
+    checkpoint_callback.save_function = mock_save_function
+    trainer = Trainer()
+    checkpoint_callback.set_trainer(trainer)
+
+    # emulate callback's calls during the training
     for loss in losses:
-        w.on_epoch_end(0, logs={'val_loss': loss})
+        checkpoint_callback._trainer.current_epoch = 0
+        checkpoint_callback._trainer.callback_metrics = {'val_loss': loss}
+        checkpoint_callback.on_validation_end()
 
     file_lists = set(os.listdir(save_dir))
 
