release v0.3.6.8

pytorch_lightning/models/trainer.py

@@ -612,7 +612,7 @@ class Trainer(TrainerIO):
 
         # enable cluster checkpointing
         # also restores training state
-        if self.cluster is not None and self.proc_rank == 0:  # pragma: no cover
+        if self.cluster is not None:  # pragma: no cover
             self.enable_auto_hpc_walltime_manager()
 
         # ---------------------------

pytorch_lightning/root_module/model_saving.py

@@ -102,13 +102,16 @@ class TrainerIO(object):
             return
 
         # allow test tube to handle model check pointing automatically
-        self.cluster.set_checkpoint_save_function(
-            self.hpc_save,
-            kwargs={
-                'folderpath': self.checkpoint_callback.filepath,
-                'experiment': self.experiment
-            }
-        )
+        # only if proc 0 so we don't trigger world_size resubmits
+        if self.proc_rank == 0:
+            self.cluster.set_checkpoint_save_function(
+                self.hpc_save,
+                kwargs={
+                    'folderpath': self.checkpoint_callback.filepath,
+                    'experiment': self.experiment
+                }
+            )
+
         self.cluster.set_checkpoint_load_function(
             self.hpc_load,
             kwargs={

setup.py

@@ -7,7 +7,7 @@ from setuptools import setup, find_packages
 # http://blog.ionelmc.ro/2014/05/25/python-packaging/
 setup(
     name="pytorch-lightning",
-    version='0.3.6.7',
+    version='0.3.6.8',
     description="The Keras for ML researchers using PyTorch",
     author="William Falcon",
     author_email="waf2107@columbia.edu",