From 0940e9b242a8bd3efa5b4214febe577e611b9a24 Mon Sep 17 00:00:00 2001
From: Henrik Holm <33653878+henholm@users.noreply.github.com>
Date: Mon, 20 Dec 2021 15:28:17 +0100
Subject: [PATCH] Add 'with torch.no_grad()' to integration test forward pass
 (#14820)

---
 tests/test_modeling_bert.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/tests/test_modeling_bert.py b/tests/test_modeling_bert.py
index 6c4613e6a3..7a66285097 100755
--- a/tests/test_modeling_bert.py
+++ b/tests/test_modeling_bert.py
@@ -587,7 +587,8 @@ class BertModelIntegrationTest(unittest.TestCase):
         model = BertModel.from_pretrained("bert-base-uncased")
         input_ids = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]])
         attention_mask = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])
-        output = model(input_ids, attention_mask=attention_mask)[0]
+        with torch.no_grad():
+            output = model(input_ids, attention_mask=attention_mask)[0]
         expected_shape = torch.Size((1, 11, 768))
         self.assertEqual(output.shape, expected_shape)
         expected_slice = torch.tensor([[[0.4249, 0.1008, 0.7531], [0.3771, 0.1188, 0.7467], [0.4152, 0.1098, 0.7108]]])
@@ -599,7 +600,8 @@ class BertModelIntegrationTest(unittest.TestCase):
         model = BertModel.from_pretrained("zhiheng-huang/bert-base-uncased-embedding-relative-key")
         input_ids = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]])
         attention_mask = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])
-        output = model(input_ids, attention_mask=attention_mask)[0]
+        with torch.no_grad():
+            output = model(input_ids, attention_mask=attention_mask)[0]
         expected_shape = torch.Size((1, 11, 768))
         self.assertEqual(output.shape, expected_shape)
         expected_slice = torch.tensor(
@@ -613,7 +615,8 @@ class BertModelIntegrationTest(unittest.TestCase):
         model = BertModel.from_pretrained("zhiheng-huang/bert-base-uncased-embedding-relative-key-query")
         input_ids = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]])
         attention_mask = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]])
-        output = model(input_ids, attention_mask=attention_mask)[0]
+        with torch.no_grad():
+            output = model(input_ids, attention_mask=attention_mask)[0]
         expected_shape = torch.Size((1, 11, 768))
         self.assertEqual(output.shape, expected_shape)
         expected_slice = torch.tensor(