Add with torch.no_grad() to DistilBERT integration test forward pass (

huggingface#14979)

* refactor: wrap forward pass around no_grad context

* Update tests/test_modeling_distilbert.py

* fix: rm `no_grad` from non-integration tests

* chore: rm whitespace change

tests/test_modeling_distilbert.py

@@ -284,7 +284,8 @@ def test_inference_no_head_absolute_embedding(self):
         model = DistilBertModel.from_pretrained("distilbert-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(