Removed old eval logic, added eval tests

pytorch_lightning/plugins/sharded_plugin.py

@@ -33,11 +33,6 @@ class DDPShardedPlugin(DDPPlugin):
             self, model: LightningModule, device_ids: List[int]
     ):
         self._wrap_optimizers(model)
-        if model.trainer.testing:  # Revert to standard DDP if testing
-            return super().configure_ddp(
-                model=model,
-                device_ids=device_ids
-            )
         return LightningShardedDataParallel(model, sharded_optimizer=model.trainer.optimizers)
 
     def optimizer_state(self, optimizer: 'OSS') -> Optional[dict]:

tests/plugins/test_sharded_plugin.py

@@ -278,3 +278,41 @@ def test_ddp_sharded_plugin_resume_from_checkpoint_gpu_to_cpu(tmpdir):
 
     trainer.fit(model)
     return 1
+
+
+@pytest.mark.skipif(platform.system() == "Windows",
+                    reason="Distributed training is not supported on Windows")
+@pytest.mark.skipif(not FAIRSCALE_AVAILABLE, reason="Fairscale is not available")
+def test_ddp_sharded_plugin_test(tmpdir):
+    """
+        Test to ensure we can use test without fit
+    """
+    model = BoringModel()
+    trainer = Trainer(
+        accelerator='ddp_cpu',
+        plugins=[DDPShardedPlugin()],
+        fast_dev_run=True,
+    )
+
+    trainer.test(model)
+    return 1
+
+
+@pytest.mark.skipif(torch.cuda.device_count() < 2, reason="test requires multi-GPU machine")
+@pytest.mark.skipif(platform.system() == "Windows",
+                    reason="Distributed training is not supported on Windows")
+@pytest.mark.skipif(not FAIRSCALE_AVAILABLE, reason="Fairscale is not available")
+def test_ddp_sharded_plugin_test_multigpu(tmpdir):
+    """
+        Test to ensure we can use test without fit
+    """
+    model = BoringModel()
+    trainer = Trainer(
+        accelerator='ddp_spawn',
+        gpus=2,
+        plugins=[DDPShardedPlugin()],
+        fast_dev_run=True,
+    )
+
+    trainer.test(model)
+    return 1