Fix tests (#14703)

tests/test_modeling_perceiver.py

@@ -860,7 +860,8 @@ class PerceiverModelIntegrationTest(unittest.TestCase):
         self.assertEqual(logits.shape, expected_shape)
 
         expected_slice = torch.tensor(
-            [[-10.8609, -10.7651, -10.9187], [-12.1689, -11.9389, -12.1479], [-12.1518, -11.9707, -12.2073]]
+            [[-10.8609, -10.7651, -10.9187], [-12.1689, -11.9389, -12.1479], [-12.1518, -11.9707, -12.2073]],
+            device=torch_device,
         )
 
         self.assertTrue(torch.allclose(logits[0, :3, :3], expected_slice, atol=1e-4))
@@ -970,7 +971,7 @@ class PerceiverModelIntegrationTest(unittest.TestCase):
 
         # forward pass
         with torch.no_grad():
-            outputs = model(inputs=patches)
+            outputs = model(inputs=patches.to(torch_device))
         logits = outputs.logits
 
         # verify logits

tests/test_tokenization_perceiver.py

@@ -99,17 +99,17 @@ class PerceiverTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
 
         # decoding
         decoded = tokenizer.decode(encoded_ids)
-        self.assertEqual(decoded, "<cls>Unicode €.<sep>")
+        self.assertEqual(decoded, "[CLS]Unicode €.[SEP]")
 
         encoded = tokenizer("e è é ê ë")
         encoded_ids = [4, 107, 38, 201, 174, 38, 201, 175, 38, 201, 176, 38, 201, 177, 5]
         self.assertEqual(encoded["input_ids"], encoded_ids)
         # decoding
         decoded = tokenizer.decode(encoded_ids)
-        self.assertEqual(decoded, "<cls>e è é ê ë<sep>")
+        self.assertEqual(decoded, "[CLS]e è é ê ë[SEP]")
 
         # encode/decode, but with `encode` instead of `__call__`
-        self.assertEqual(tokenizer.decode(tokenizer.encode("e è é ê ë")), "<cls>e è é ê ë<sep>")
+        self.assertEqual(tokenizer.decode(tokenizer.encode("e è é ê ë")), "[CLS]e è é ê ë[SEP]")
 
     def test_prepare_batch_integration(self):
         tokenizer = self.perceiver_tokenizer