ddp pickle

pytorch_lightning/callbacks/early_stopping.py

@@ -57,6 +57,7 @@ class EarlyStopping(Callback):
         self.min_delta = min_delta
         self.wait = 0
         self.stopped_epoch = 0
+        self.mode = mode
 
         mode_dict = {
             'min': torch.lt,
@@ -67,9 +68,8 @@ class EarlyStopping(Callback):
         if mode not in mode_dict:
             if self.verbose > 0:
                 log.info(f'EarlyStopping mode {mode} is unknown, fallback to auto mode.')
-            mode = 'auto'
+            self.mode = 'auto'
 
-        self.monitor_op = mode_dict[mode]
         self.min_delta *= 1 if self.monitor_op == torch.gt else -1
 
     def _validate_condition_metric(self, logs):
@@ -94,6 +94,15 @@ class EarlyStopping(Callback):
 
         return True
 
+    @property
+    def monitor_op(self):
+        mode_dict = {
+            'min': torch.lt,
+            'max': torch.gt,
+            'auto': torch.gt if 'acc' in self.monitor else torch.lt
+        }
+        return mode_dict[self.mode]
+
     def on_train_start(self, trainer, pl_module):
         # Allow instances to be re-used
         self.wait = 0

pytorch_lightning/trainer/distrib_data_parallel.py

@@ -378,6 +378,7 @@ class TrainerDDPMixin(ABC):
         :param model:
         :return:
         """
+        import pdb; pdb.set_trace()
         if self.proc_rank == 0:
             path = os.path.join(self.default_root_dir, '__temp_weight_ddp_end.ckpt')
             self.save_checkpoint(path)