From 75518f7042d925cba02f939a66ab49a62193b998 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sat, 23 Oct 2021 00:19:01 +0800
Subject: [PATCH 1/7] [DLMED] add utitlity

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/__init__.py        |  1 +
 monai/networks/utils.py           | 44 +++++++++++++++++++++++++++++++
 tests/test_save_to_torchscript.py | 40 ++++++++++++++++++++++++++++
 3 files changed, 85 insertions(+)
 create mode 100644 tests/test_save_to_torchscript.py

diff --git a/monai/networks/__init__.py b/monai/networks/__init__.py
index 3c347dad22..65e7b26331 100644
--- a/monai/networks/__init__.py
+++ b/monai/networks/__init__.py
@@ -18,6 +18,7 @@
     one_hot,
     pixelshuffle,
     predict_segmentation,
+    save_to_torchscript,
     slice_channels,
     to_norm_affine,
     train_mode,
diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index 529dfbf977..94828a15e6 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -34,6 +34,7 @@
     "eval_mode",
     "train_mode",
     "copy_model_state",
+    "save_to_torchscript",
 ]
 
 
@@ -424,3 +425,46 @@ def copy_model_state(
     if inplace and isinstance(dst, torch.nn.Module):
         dst.load_state_dict(dst_dict)
     return dst_dict, updated_keys, unchanged_keys
+
+
+def save_to_torchscript(
+    model: nn.Module,
+    output_path: str,
+    verify: bool = False,
+    input_shape: Optional[Sequence[int]] = None,
+    device: Optional[torch.device] = None,
+    rtol: float = 1e-4,
+    atol: float = 0.0,
+):
+    """
+    Utility to save a model into TorchScript model with optional input / output data verification.
+
+    Args:
+        model: source PyTorch model to save.
+        output_path: the path to save converted TorchScript model.
+        verify: whether to verify the input and output of TorchScript model.
+        input_shape: shape of the input data to verify model.
+        device: target device to verify the model.
+        rtol: the relative tolerance when comparing the outputs of PyTorch model and TorchScript model.
+        atol: the absolute tolerance when comparing the outputs of PyTorch model and TorchScript model.
+
+    """
+    model.eval()
+    with torch.no_grad():
+        script_module = torch.jit.script(model)
+        script_module.save(output_path)
+
+    if verify:
+        if input_shape is None:
+            raise ValueError("missing input_shape argument for verification.")
+        dummy_input = torch.randn(tuple(input_shape), requires_grad=False).to(device)
+        torchscript_model = torch.jit.load(output_path).eval().to(device)
+        model = model.to(device)
+
+        with torch.no_grad():
+            torch_out = model(dummy_input)
+            torchscript_out = torchscript_model(dummy_input)
+        # compare TorchScript and PyTorch results
+        torch.testing.assert_allclose(torch_out, torchscript_out, rtol=rtol, atol=atol)
+
+    return script_module
diff --git a/tests/test_save_to_torchscript.py b/tests/test_save_to_torchscript.py
new file mode 100644
index 0000000000..fe7be5cc8c
--- /dev/null
+++ b/tests/test_save_to_torchscript.py
@@ -0,0 +1,40 @@
+# Copyright 2020 - 2021 MONAI Consortium
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#     http://www.apache.org/licenses/LICENSE-2.0
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import tempfile
+import unittest
+
+import torch
+
+from monai.networks import save_to_torchscript
+from monai.networks.nets import UNet
+
+
+class TestArrayDataset(unittest.TestCase):
+    def test_shape(self):
+        model = UNet(
+            spatial_dims=2, in_channels=1, out_channels=3, channels=(16, 32, 64), strides=(2, 2), num_res_units=0
+        )
+        with tempfile.TemporaryDirectory() as tempdir:
+            torchscript_model = save_to_torchscript(
+                model=model,
+                output_path=os.path.join(tempdir, "model.ts"),
+                verify=True,
+                input_shape=(16, 1, 32, 32),
+                device="cuda" if torch.cuda.is_available() else "cpu",
+                rtol=1e-3,
+            )
+            self.assertTrue(isinstance(torchscript_model, torch.jit._script.RecursiveScriptModule))
+
+
+if __name__ == "__main__":
+    unittest.main()

From 2d69164e4febf04f54b072c4a9ec16956ebb57d1 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sat, 23 Oct 2021 01:13:15 +0800
Subject: [PATCH 2/7] [DLMED] update according to comments

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/__init__.py                    |  2 +-
 monai/networks/utils.py                       | 23 +++++++++++--------
 ...ript.py => test_convert_to_torchscript.py} |  6 ++---
 tests/utils.py                                |  3 ++-
 4 files changed, 20 insertions(+), 14 deletions(-)
 rename tests/{test_save_to_torchscript.py => test_convert_to_torchscript.py} (90%)

diff --git a/monai/networks/__init__.py b/monai/networks/__init__.py
index 65e7b26331..4dec09c889 100644
--- a/monai/networks/__init__.py
+++ b/monai/networks/__init__.py
@@ -10,6 +10,7 @@
 # limitations under the License.
 
 from .utils import (
+    convert_to_torchscript,
     copy_model_state,
     eval_mode,
     icnr_init,
@@ -18,7 +19,6 @@
     one_hot,
     pixelshuffle,
     predict_segmentation,
-    save_to_torchscript,
     slice_channels,
     to_norm_affine,
     train_mode,
diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index 94828a15e6..77ce2a2277 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -34,7 +34,7 @@
     "eval_mode",
     "train_mode",
     "copy_model_state",
-    "save_to_torchscript",
+    "convert_to_torchscript",
 ]
 
 
@@ -427,43 +427,48 @@ def copy_model_state(
     return dst_dict, updated_keys, unchanged_keys
 
 
-def save_to_torchscript(
+def convert_to_torchscript(
     model: nn.Module,
-    output_path: str,
+    output_path: Optional[str] = None,
     verify: bool = False,
     input_shape: Optional[Sequence[int]] = None,
     device: Optional[torch.device] = None,
     rtol: float = 1e-4,
     atol: float = 0.0,
+    **kwargs,
 ):
     """
-    Utility to save a model into TorchScript model with optional input / output data verification.
+    Utility to convert a model into TorchScript model and save to file,
+    with optional input / output data verification.
 
     Args:
         model: source PyTorch model to save.
-        output_path: the path to save converted TorchScript model.
+        output_path: if not None, specify the path to save converted TorchScript model.
         verify: whether to verify the input and output of TorchScript model.
+            if `output_path` is not None, load the saved TorchScript model and verify.
         input_shape: shape of the input data to verify model.
         device: target device to verify the model.
         rtol: the relative tolerance when comparing the outputs of PyTorch model and TorchScript model.
         atol: the absolute tolerance when comparing the outputs of PyTorch model and TorchScript model.
+        kwargs: besides input data, other arguments for the call function of model.
 
     """
     model.eval()
     with torch.no_grad():
         script_module = torch.jit.script(model)
-        script_module.save(output_path)
+        if output_path is not None:
+            script_module.save(output_path)
 
     if verify:
         if input_shape is None:
             raise ValueError("missing input_shape argument for verification.")
         dummy_input = torch.randn(tuple(input_shape), requires_grad=False).to(device)
-        torchscript_model = torch.jit.load(output_path).eval().to(device)
+        ts_model = (torch.jit.load(output_path) if output_path is not None else script_module).eval().to(device)
         model = model.to(device)
 
         with torch.no_grad():
-            torch_out = model(dummy_input)
-            torchscript_out = torchscript_model(dummy_input)
+            torch_out = model(dummy_input, **kwargs)
+            torchscript_out = ts_model(dummy_input, **kwargs)
         # compare TorchScript and PyTorch results
         torch.testing.assert_allclose(torch_out, torchscript_out, rtol=rtol, atol=atol)
 
diff --git a/tests/test_save_to_torchscript.py b/tests/test_convert_to_torchscript.py
similarity index 90%
rename from tests/test_save_to_torchscript.py
rename to tests/test_convert_to_torchscript.py
index fe7be5cc8c..31648511ab 100644
--- a/tests/test_save_to_torchscript.py
+++ b/tests/test_convert_to_torchscript.py
@@ -15,7 +15,7 @@
 
 import torch
 
-from monai.networks import save_to_torchscript
+from monai.networks import convert_to_torchscript
 from monai.networks.nets import UNet
 
 
@@ -25,7 +25,7 @@ def test_shape(self):
             spatial_dims=2, in_channels=1, out_channels=3, channels=(16, 32, 64), strides=(2, 2), num_res_units=0
         )
         with tempfile.TemporaryDirectory() as tempdir:
-            torchscript_model = save_to_torchscript(
+            torchscript_model = convert_to_torchscript(
                 model=model,
                 output_path=os.path.join(tempdir, "model.ts"),
                 verify=True,
@@ -33,7 +33,7 @@ def test_shape(self):
                 device="cuda" if torch.cuda.is_available() else "cpu",
                 rtol=1e-3,
             )
-            self.assertTrue(isinstance(torchscript_model, torch.jit._script.RecursiveScriptModule))
+            self.assertTrue(isinstance(torchscript_model, torch.nn.Module))
 
 
 if __name__ == "__main__":
diff --git a/tests/utils.py b/tests/utils.py
index c73a87d141..a12d9f8ea6 100644
--- a/tests/utils.py
+++ b/tests/utils.py
@@ -35,6 +35,7 @@
 from monai.config.deviceconfig import USE_COMPILED
 from monai.config.type_definitions import NdarrayOrTensor
 from monai.data import create_test_image_2d, create_test_image_3d
+from monai.networks import convert_to_torchscript
 from monai.utils import ensure_tuple, optional_import, set_determinism
 from monai.utils.misc import is_module_ver_at_least
 from monai.utils.module import version_leq
@@ -598,7 +599,7 @@ def test_script_save(net, *inputs, eval_nets=True, device=None, rtol=1e-4):
     # Convert to device
     inputs = [i.to(device) for i in inputs]
 
-    scripted = torch.jit.script(net.cpu())
+    scripted = convert_to_torchscript(net.cpu(), output_path=None)
     buffer = scripted.save_to_buffer()
     reloaded_net = torch.jit.load(BytesIO(buffer)).to(device)
     net.to(device)

From c6e2d89b9b5040813b805a70fad3215aac499e92 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sat, 23 Oct 2021 08:29:12 +0800
Subject: [PATCH 3/7] [DLMED] update according to comments

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/utils.py              | 17 +++++++++++------
 tests/test_convert_to_torchscript.py |  8 +++++---
 tests/utils.py                       |  7 ++++---
 3 files changed, 20 insertions(+), 12 deletions(-)

diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index 77ce2a2277..f054b27992 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -15,7 +15,7 @@
 import warnings
 from collections import OrderedDict
 from contextlib import contextmanager
-from typing import Any, Callable, Mapping, Optional, Sequence, Union
+from typing import Any, Callable, Dict, Mapping, Optional, Sequence, Union
 
 import torch
 import torch.nn as nn
@@ -429,7 +429,8 @@ def copy_model_state(
 
 def convert_to_torchscript(
     model: nn.Module,
-    output_path: Optional[str] = None,
+    filename_or_obj: Optional[Any] = None,
+    extra_files: Optional[Dict] = None,
     verify: bool = False,
     input_shape: Optional[Sequence[int]] = None,
     device: Optional[torch.device] = None,
@@ -443,7 +444,10 @@ def convert_to_torchscript(
 
     Args:
         model: source PyTorch model to save.
-        output_path: if not None, specify the path to save converted TorchScript model.
+        filename_or_obj: if not None, specify a file-like object (has to implement write and flush)
+            or a string containing a file path name to save the TorchScript model.
+        extra_files: map from filename to contents which will be stored as part of the save model file.
+            for more details: https://pytorch.org/docs/stable/generated/torch.jit.save.html.
         verify: whether to verify the input and output of TorchScript model.
             if `output_path` is not None, load the saved TorchScript model and verify.
         input_shape: shape of the input data to verify model.
@@ -456,14 +460,15 @@ def convert_to_torchscript(
     model.eval()
     with torch.no_grad():
         script_module = torch.jit.script(model)
-        if output_path is not None:
-            script_module.save(output_path)
+        if filename_or_obj is not None:
+            torch.jit.save(m=script_module, f=filename_or_obj, _extra_files=extra_files)
 
     if verify:
         if input_shape is None:
             raise ValueError("missing input_shape argument for verification.")
         dummy_input = torch.randn(tuple(input_shape), requires_grad=False).to(device)
-        ts_model = (torch.jit.load(output_path) if output_path is not None else script_module).eval().to(device)
+        ts_model = torch.jit.load(filename_or_obj) if filename_or_obj is not None else script_module
+        ts_model.eval().to(device)
         model = model.to(device)
 
         with torch.no_grad():
diff --git a/tests/test_convert_to_torchscript.py b/tests/test_convert_to_torchscript.py
index 31648511ab..b313038f61 100644
--- a/tests/test_convert_to_torchscript.py
+++ b/tests/test_convert_to_torchscript.py
@@ -19,19 +19,21 @@
 from monai.networks.nets import UNet
 
 
-class TestArrayDataset(unittest.TestCase):
-    def test_shape(self):
+class TestConvertToTorchScript(unittest.TestCase):
+    def test_value(self):
         model = UNet(
             spatial_dims=2, in_channels=1, out_channels=3, channels=(16, 32, 64), strides=(2, 2), num_res_units=0
         )
         with tempfile.TemporaryDirectory() as tempdir:
             torchscript_model = convert_to_torchscript(
                 model=model,
-                output_path=os.path.join(tempdir, "model.ts"),
+                filename_or_obj=os.path.join(tempdir, "model.ts"),
+                extra_files={"foo.txt": b"bar"},
                 verify=True,
                 input_shape=(16, 1, 32, 32),
                 device="cuda" if torch.cuda.is_available() else "cpu",
                 rtol=1e-3,
+                atol=1e-4,
             )
             self.assertTrue(isinstance(torchscript_model, torch.nn.Module))
 
diff --git a/tests/utils.py b/tests/utils.py
index a12d9f8ea6..38c7dc3f33 100644
--- a/tests/utils.py
+++ b/tests/utils.py
@@ -599,9 +599,10 @@ def test_script_save(net, *inputs, eval_nets=True, device=None, rtol=1e-4):
     # Convert to device
     inputs = [i.to(device) for i in inputs]
 
-    scripted = convert_to_torchscript(net.cpu(), output_path=None)
-    buffer = scripted.save_to_buffer()
-    reloaded_net = torch.jit.load(BytesIO(buffer)).to(device)
+    buffer = BytesIO()
+    convert_to_torchscript(net.cpu(), filename_or_obj=buffer, verify=False)
+    buffer.seek(0)
+    reloaded_net = torch.jit.load(buffer).to(device)
     net.to(device)
 
     if eval_nets:

From c1f00d6663bbb080f123bd78bde5815232bbb8c1 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sat, 23 Oct 2021 08:47:04 +0800
Subject: [PATCH 4/7] [DLMED] fix CI for old PyTorch

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/utils.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index f054b27992..0e9ad0708c 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -21,6 +21,7 @@
 import torch.nn as nn
 
 from monai.utils.deprecate_utils import deprecated_arg
+from monai.utils.module import PT_BEFORE_1_7
 
 __all__ = [
     "one_hot",
@@ -447,6 +448,7 @@ def convert_to_torchscript(
         filename_or_obj: if not None, specify a file-like object (has to implement write and flush)
             or a string containing a file path name to save the TorchScript model.
         extra_files: map from filename to contents which will be stored as part of the save model file.
+            works for PyTorch 1.7 or later.
             for more details: https://pytorch.org/docs/stable/generated/torch.jit.save.html.
         verify: whether to verify the input and output of TorchScript model.
             if `output_path` is not None, load the saved TorchScript model and verify.
@@ -461,7 +463,10 @@ def convert_to_torchscript(
     with torch.no_grad():
         script_module = torch.jit.script(model)
         if filename_or_obj is not None:
-            torch.jit.save(m=script_module, f=filename_or_obj, _extra_files=extra_files)
+            if PT_BEFORE_1_7:
+                torch.jit.save(m=script_module, f=filename_or_obj)
+            else:
+                torch.jit.save(m=script_module, f=filename_or_obj, _extra_files=extra_files)
 
     if verify:
         if input_shape is None:

From 54300691f06723e19606ea8ecf782ad33a3fd198 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sun, 24 Oct 2021 09:01:49 +0800
Subject: [PATCH 5/7] [DLMED] update according to comments

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/utils.py              | 29 +++++++++-----
 tests/test_convert_to_torchscript.py |  2 +-
 tests/utils.py                       | 59 ++++++----------------------
 3 files changed, 33 insertions(+), 57 deletions(-)

diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index 0e9ad0708c..0e8d2b8f79 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -21,6 +21,7 @@
 import torch.nn as nn
 
 from monai.utils.deprecate_utils import deprecated_arg
+from monai.utils.misc import ensure_tuple, set_determinism
 from monai.utils.module import PT_BEFORE_1_7
 
 __all__ = [
@@ -433,7 +434,7 @@ def convert_to_torchscript(
     filename_or_obj: Optional[Any] = None,
     extra_files: Optional[Dict] = None,
     verify: bool = False,
-    input_shape: Optional[Sequence[int]] = None,
+    inputs: Optional[Sequence[Any]] = None,
     device: Optional[torch.device] = None,
     rtol: float = 1e-4,
     atol: float = 0.0,
@@ -452,11 +453,11 @@ def convert_to_torchscript(
             for more details: https://pytorch.org/docs/stable/generated/torch.jit.save.html.
         verify: whether to verify the input and output of TorchScript model.
             if `output_path` is not None, load the saved TorchScript model and verify.
-        input_shape: shape of the input data to verify model.
-        device: target device to verify the model.
+        inputs: input test data to verify model, should be a sequence of data, every item maps to a argument
+            of `model()` function.
+        device: target device to verify the model, if None, use CUDA if available.
         rtol: the relative tolerance when comparing the outputs of PyTorch model and TorchScript model.
         atol: the absolute tolerance when comparing the outputs of PyTorch model and TorchScript model.
-        kwargs: besides input data, other arguments for the call function of model.
 
     """
     model.eval()
@@ -469,17 +470,25 @@ def convert_to_torchscript(
                 torch.jit.save(m=script_module, f=filename_or_obj, _extra_files=extra_files)
 
     if verify:
-        if input_shape is None:
-            raise ValueError("missing input_shape argument for verification.")
-        dummy_input = torch.randn(tuple(input_shape), requires_grad=False).to(device)
+        if device is None:
+            device = "cuda" if torch.cuda.is_available() else "cpu"
+        if inputs is None:
+            raise ValueError("missing input data for verification.")
+
+        inputs = [i.to(device) if isinstance(i, torch.Tensor) else i for i in inputs]
         ts_model = torch.jit.load(filename_or_obj) if filename_or_obj is not None else script_module
         ts_model.eval().to(device)
         model = model.to(device)
 
         with torch.no_grad():
-            torch_out = model(dummy_input, **kwargs)
-            torchscript_out = ts_model(dummy_input, **kwargs)
+            set_determinism(seed=0)
+            torch_out = ensure_tuple(model(*inputs))
+            set_determinism(seed=0)
+            torchscript_out = ensure_tuple(ts_model(*inputs))
+            set_determinism(seed=None)
         # compare TorchScript and PyTorch results
-        torch.testing.assert_allclose(torch_out, torchscript_out, rtol=rtol, atol=atol)
+        for r1, r2 in zip(torch_out, torchscript_out):
+            if isinstance(r1, torch.Tensor) or isinstance(r2, torch.Tensor):
+                torch.testing.assert_allclose(r1, r2, rtol=rtol, atol=atol)
 
     return script_module
diff --git a/tests/test_convert_to_torchscript.py b/tests/test_convert_to_torchscript.py
index b313038f61..a772610a04 100644
--- a/tests/test_convert_to_torchscript.py
+++ b/tests/test_convert_to_torchscript.py
@@ -30,7 +30,7 @@ def test_value(self):
                 filename_or_obj=os.path.join(tempdir, "model.ts"),
                 extra_files={"foo.txt": b"bar"},
                 verify=True,
-                input_shape=(16, 1, 32, 32),
+                inputs=[torch.randn((16, 1, 32, 32), requires_grad=False)],
                 device="cuda" if torch.cuda.is_available() else "cpu",
                 rtol=1e-3,
                 atol=1e-4,
diff --git a/tests/utils.py b/tests/utils.py
index 38c7dc3f33..19bf9977e6 100644
--- a/tests/utils.py
+++ b/tests/utils.py
@@ -22,7 +22,6 @@
 import unittest
 import warnings
 from functools import partial
-from io import BytesIO
 from subprocess import PIPE, Popen
 from typing import Callable, Optional, Tuple
 from urllib.error import HTTPError, URLError
@@ -36,7 +35,7 @@
 from monai.config.type_definitions import NdarrayOrTensor
 from monai.data import create_test_image_2d, create_test_image_3d
 from monai.networks import convert_to_torchscript
-from monai.utils import ensure_tuple, optional_import, set_determinism
+from monai.utils import optional_import
 from monai.utils.misc import is_module_ver_at_least
 from monai.utils.module import version_leq
 from monai.utils.type_conversion import convert_data_type
@@ -579,55 +578,23 @@ def setUp(self):
         self.segn = torch.tensor(self.segn)
 
 
-def test_script_save(net, *inputs, eval_nets=True, device=None, rtol=1e-4):
+def test_script_save(net, *inputs, device=None, rtol=1e-4):
     """
     Test the ability to save `net` as a Torchscript object, reload it, and apply inference. The value `inputs` is
-    forward-passed through the original and loaded copy of the network and their results returned. Both `net` and its
-    reloaded copy are set to evaluation mode if `eval_nets` is True. The forward pass for both is done without
-    gradient accumulation.
+    forward-passed through the original and loaded copy of the network and their results returned.
+    The forward pass for both is done without gradient accumulation.
 
     The test will be performed with CUDA if available, else CPU.
     """
-    if True:
-        device = "cpu"
-    else:
-        # TODO: It would be nice to be able to use GPU if
-        # available, but this currently causes CI failures.
-        if not device:
-            device = "cuda" if torch.cuda.is_available() else "cpu"
-
-    # Convert to device
-    inputs = [i.to(device) for i in inputs]
-
-    buffer = BytesIO()
-    convert_to_torchscript(net.cpu(), filename_or_obj=buffer, verify=False)
-    buffer.seek(0)
-    reloaded_net = torch.jit.load(buffer).to(device)
-    net.to(device)
-
-    if eval_nets:
-        net.eval()
-        reloaded_net.eval()
-
-    with torch.no_grad():
-        set_determinism(seed=0)
-        result1 = net(*inputs)
-        result2 = reloaded_net(*inputs)
-        set_determinism(seed=None)
-
-    # convert results to tuples if needed to allow iterating over pairs of outputs
-    result1 = ensure_tuple(result1)
-    result2 = ensure_tuple(result2)
-
-    for i, (r1, r2) in enumerate(zip(result1, result2)):
-        if None not in (r1, r2):  # might be None
-            np.testing.assert_allclose(
-                r1.detach().cpu().numpy(),
-                r2.detach().cpu().numpy(),
-                rtol=rtol,
-                atol=0,
-                err_msg=f"failed on comparison number: {i}",
-            )
+    with tempfile.TemporaryDirectory() as tempdir:
+        convert_to_torchscript(
+            model=net,
+            filename_or_obj=os.path.join(tempdir, "model.ts"),
+            verify=True,
+            inputs=inputs,
+            device=device,
+            rtol=rtol,
+        )
 
 
 def query_memory(n=2):

From 7fd52581512b19360cb5c1059c461409b49ec964 Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sun, 24 Oct 2021 09:10:18 +0800
Subject: [PATCH 6/7] [DLMED] fix flake8

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 monai/networks/utils.py                       | 4 ++--
 tests/test_utils_pytorch_numpy_unification.py | 3 ++-
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/monai/networks/utils.py b/monai/networks/utils.py
index 0e8d2b8f79..c60867d8e6 100644
--- a/monai/networks/utils.py
+++ b/monai/networks/utils.py
@@ -452,7 +452,7 @@ def convert_to_torchscript(
             works for PyTorch 1.7 or later.
             for more details: https://pytorch.org/docs/stable/generated/torch.jit.save.html.
         verify: whether to verify the input and output of TorchScript model.
-            if `output_path` is not None, load the saved TorchScript model and verify.
+            if `filename_or_obj` is not None, load the saved TorchScript model and verify.
         inputs: input test data to verify model, should be a sequence of data, every item maps to a argument
             of `model()` function.
         device: target device to verify the model, if None, use CUDA if available.
@@ -471,7 +471,7 @@ def convert_to_torchscript(
 
     if verify:
         if device is None:
-            device = "cuda" if torch.cuda.is_available() else "cpu"
+            device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
         if inputs is None:
             raise ValueError("missing input data for verification.")
 
diff --git a/tests/test_utils_pytorch_numpy_unification.py b/tests/test_utils_pytorch_numpy_unification.py
index c8e0a35c92..c3b1bc259b 100644
--- a/tests/test_utils_pytorch_numpy_unification.py
+++ b/tests/test_utils_pytorch_numpy_unification.py
@@ -15,7 +15,8 @@
 import torch
 
 from monai.transforms.utils_pytorch_numpy_unification import percentile
-from tests.utils import TEST_NDARRAYS, assert_allclose, set_determinism
+from monai.utils import set_determinism
+from tests.utils import TEST_NDARRAYS, assert_allclose
 
 
 class TestPytorchNumpyUnification(unittest.TestCase):

From 727b3431798ff985cd03924c4da7583ec1a804cf Mon Sep 17 00:00:00 2001
From: Nic Ma <nma@nvidia.com>
Date: Sun, 24 Oct 2021 09:56:05 +0800
Subject: [PATCH 7/7] [DLMED] fix CI issue

Signed-off-by: Nic Ma <nma@nvidia.com>
---
 tests/utils.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/tests/utils.py b/tests/utils.py
index 19bf9977e6..97e57c6be6 100644
--- a/tests/utils.py
+++ b/tests/utils.py
@@ -578,7 +578,7 @@ def setUp(self):
         self.segn = torch.tensor(self.segn)
 
 
-def test_script_save(net, *inputs, device=None, rtol=1e-4):
+def test_script_save(net, *inputs, device=None, rtol=1e-4, atol=0.0):
     """
     Test the ability to save `net` as a Torchscript object, reload it, and apply inference. The value `inputs` is
     forward-passed through the original and loaded copy of the network and their results returned.
@@ -586,6 +586,8 @@ def test_script_save(net, *inputs, device=None, rtol=1e-4):
 
     The test will be performed with CUDA if available, else CPU.
     """
+    # TODO: would be nice to use GPU if available, but it currently causes CI failures.
+    device = "cpu"
     with tempfile.TemporaryDirectory() as tempdir:
         convert_to_torchscript(
             model=net,
@@ -594,6 +596,7 @@ def test_script_save(net, *inputs, device=None, rtol=1e-4):
             inputs=inputs,
             device=device,
             rtol=rtol,
+            atol=atol,
         )