Skip tests properly (#31308)

* Skip tests properly

* [test_all]

* Add 'reason' as kwarg for skipTest

* [test_all] Fix up

* [test_all]

tests/models/mgp_str/test_tokenization_mgp_str.py

@@ -51,7 +51,7 @@ class MgpstrTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
         output_text = "tester"
         return input_text, output_text
 
-    @unittest.skip("MGP-STR always lower cases letters.")
+    @unittest.skip(reason="MGP-STR always lower cases letters.")
     def test_added_tokens_do_lower_case(self):
         pass
 
@@ -86,10 +86,10 @@ class MgpstrTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
 
                 self.assertEqual(text_2.replace(" ", ""), output_text)
 
-    @unittest.skip("MGP-STR tokenizer only handles one sequence.")
+    @unittest.skip(reason="MGP-STR tokenizer only handles one sequence.")
     def test_maximum_encoding_length_pair_input(self):
         pass
 
-    @unittest.skip("inputs cannot be pretokenized in MgpstrTokenizer")
+    @unittest.skip(reason="inputs cannot be pretokenized in MgpstrTokenizer")
     def test_pretokenized_inputs(self):
         pass

tests/models/mistral/test_modeling_mistral.py

@@ -397,11 +397,11 @@ class MistralModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
             (self.model_tester.batch_size, self.model_tester.seq_length, self.model_tester.num_labels),
         )
 
-    @unittest.skip("Mistral buffers include complex numbers, which breaks this test")
+    @unittest.skip(reason="Mistral buffers include complex numbers, which breaks this test")
     def test_save_load_fast_init_from_base(self):
         pass
 
-    @unittest.skip("Mistral uses GQA on all models so the KV cache is a non standard format")
+    @unittest.skip(reason="Mistral uses GQA on all models so the KV cache is a non standard format")
     def test_past_key_values_format(self):
         pass
 
@@ -489,7 +489,7 @@ class MistralModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
     @pytest.mark.flash_attn_test
     @slow
     def test_flash_attn_2_inference_equivalence_right_padding(self):
-        self.skipTest("Mistral flash attention does not support right padding")
+        self.skipTest(reason="Mistral flash attention does not support right padding")
 
 
 @require_torch_gpu
@@ -647,10 +647,10 @@ class MistralIntegrationTest(unittest.TestCase):
         # `torch==2.2` will throw an error on this test (as in other compilation tests), but torch==2.1.2 and torch>2.2
         # work as intended. See https://github.com/pytorch/pytorch/issues/121943
         if version.parse(torch.__version__) < version.parse("2.3.0"):
-            self.skipTest("This test requires torch >= 2.3 to run.")
+            self.skipTest(reason="This test requires torch >= 2.3 to run.")
 
         if self.cuda_compute_capability_major_version == 7:
-            self.skipTest("This test is failing (`torch.compile` fails) on Nvidia T4 GPU.")
+            self.skipTest(reason="This test is failing (`torch.compile` fails) on Nvidia T4 GPU.")
 
         NUM_TOKENS_TO_GENERATE = 40
         EXPECTED_TEXT_COMPLETION = {

tests/models/mixtral/test_modeling_mixtral.py

@@ -398,11 +398,11 @@ class MixtralModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
             (self.model_tester.batch_size, self.model_tester.seq_length, self.model_tester.num_labels),
         )
 
-    @unittest.skip("Mixtral buffers include complex numbers, which breaks this test")
+    @unittest.skip(reason="Mixtral buffers include complex numbers, which breaks this test")
     def test_save_load_fast_init_from_base(self):
         pass
 
-    @unittest.skip("Mixtral uses GQA on all models so the KV cache is a non standard format")
+    @unittest.skip(reason="Mixtral uses GQA on all models so the KV cache is a non standard format")
     def test_past_key_values_format(self):
         pass
 
@@ -490,7 +490,7 @@ class MixtralModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
     @pytest.mark.flash_attn_test
     @slow
     def test_flash_attn_2_inference_equivalence_right_padding(self):
-        self.skipTest("Mixtral flash attention does not support right padding")
+        self.skipTest(reason="Mixtral flash attention does not support right padding")
 
     # Ignore copy
     def test_load_balancing_loss(self):

tests/models/mluke/test_tokenization_mluke.py

@@ -93,6 +93,7 @@ class MLukeTokenizerTest(TokenizerTesterMixin, unittest.TestCase):
         ids = tokenizer.encode(txt, add_special_tokens=False)
         return txt, ids
 
+    @unittest.skip
     def test_pretokenized_inputs(self):
         pass
 

tests/models/mobilebert/test_modeling_mobilebert.py

@@ -298,7 +298,7 @@ class MobileBertModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCa
         return inputs_dict
 
     # TODO (@SunMarc): Fix me
-    @unittest.skip("It's broken.")
+    @unittest.skip(reason="It's broken.")
     def test_resize_tokens_embeddings(self):
         super().test_resize_tokens_embeddings()
 

tests/models/mobilebert/test_tokenization_mobilebert.py

@@ -87,7 +87,7 @@ class MobileBERTTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     # Copied from tests.models.bert.test_tokenization_bert.BertTokenizationTest.test_rust_and_python_full_tokenizers
     def test_rust_and_python_full_tokenizers(self):
         if not self.test_rust_tokenizer:
-            return
+            self.skipTest(reason="test_rust_tokenizer is set to False")
 
         tokenizer = self.get_tokenizer()
         rust_tokenizer = self.get_rust_tokenizer()

tests/models/mpnet/test_modeling_mpnet.py

@@ -246,7 +246,7 @@ class MPNetModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.create_and_check_mpnet_for_question_answering(*config_and_inputs)
 
-    @unittest.skip("TFMPNet adds poolers to all models, unlike the PT model class.")
+    @unittest.skip(reason="TFMPNet adds poolers to all models, unlike the PT model class.")
     def test_tf_from_pt_safetensors(self):
         return
 

tests/models/mpt/test_modeling_mpt.py

@@ -422,7 +422,7 @@ class MptModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.create_and_check_mpt_weight_initialization(*config_and_inputs)
 
-    @unittest.skip("For backward compatibility the lm_head is not in the model's state dict on the Hub.")
+    @unittest.skip(reason="For backward compatibility the lm_head is not in the model's state dict on the Hub.")
     def test_model_weights_reload_no_missing_tied_weights(self):
         pass
 

tests/models/mra/test_modeling_mra.py

@@ -376,7 +376,9 @@ class MraModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
     def test_training_gradient_checkpointing_use_reentrant_false(self):
         pass
 
-    @unittest.skip("Model has `nan` in hidden_states, see https://github.com/huggingface/transformers/issues/29373.")
+    @unittest.skip(
+        reason="Model has `nan` in hidden_states, see https://github.com/huggingface/transformers/issues/29373."
+    )
     def test_batching_equivalence(self):
         pass
 

tests/models/mt5/test_modeling_mt5.py

@@ -593,7 +593,7 @@ class MT5ModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
 
     def _create_and_check_torch_fx_tracing(self, config, inputs_dict, output_loss=False):
         if not is_torch_fx_available() or not self.fx_compatible:
-            return
+            self.skipTest(reason="torch.fx is not available or not compatible with this model")
 
         configs_no_init = _config_zero_init(config)  # To be sure we have no Nan
         configs_no_init.return_dict = False
@@ -837,7 +837,7 @@ class MT5ModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
         model = MT5Model.from_pretrained(model_name)
         self.assertIsNotNone(model)
 
-    @unittest.skip("Test has a segmentation fault on torch 1.8.0")
+    @unittest.skip(reason="Test has a segmentation fault on torch 1.8.0")
     def test_export_to_onnx(self):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         model = MT5Model(config_and_inputs[0]).to(torch_device)

tests/models/musicgen/test_modeling_musicgen.py

@@ -205,7 +205,7 @@ class MusicgenDecoderTest(ModelTesterMixin, GenerationTesterMixin, PipelineTeste
 
     def check_training_gradient_checkpointing(self, gradient_checkpointing_kwargs=None):
         if not self.model_tester.is_training:
-            return
+            self.skipTest(reason="model_tester.is_training is set to False")
 
         config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
         config.use_cache = False
@@ -270,15 +270,15 @@ class MusicgenDecoderTest(ModelTesterMixin, GenerationTesterMixin, PipelineTeste
     def test_inputs_embeds_matches_input_ids(self):
         pass
 
-    # skip as this model doesn't support all arguments tested
+    @unittest.skip(reason="MusicGen does not support all arguments tested")
     def test_model_outputs_equivalence(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied")
     def test_tie_model_weights(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied")
     def test_tied_weights_keys(self):
         pass
 
@@ -624,6 +624,9 @@ class MusicgenDecoderTest(ModelTesterMixin, GenerationTesterMixin, PipelineTeste
     @slow
     # Copied from tests.test_modeling_common.ModelTesterMixin.test_eager_matches_sdpa_inference
     def test_eager_matches_sdpa_inference(self, torch_dtype: str):
+        if not self.has_attentions:
+            self.skipTest(reason="Model architecture does not support attentions")
+
         if not self.all_model_classes[0]._supports_sdpa:
             self.skipTest(f"{self.all_model_classes[0].__name__} does not support SDPA")
 
@@ -1085,7 +1088,7 @@ class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
 
     def check_training_gradient_checkpointing(self, gradient_checkpointing_kwargs=None):
         if not self.model_tester.is_training:
-            return
+            self.skipTest(reason="model_tester.is_training is set to False")
 
         for model_class in self.all_model_classes:
             config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
@@ -1262,27 +1265,27 @@ class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
             model = model_class(config)
             self.assertTrue(model.is_gradient_checkpointing)
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tie_model_weights(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tied_model_weights_key_ignore(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tied_weights_keys(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage_checkpoints(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage_no_safetensors(self):
         pass
 
@@ -1569,7 +1572,7 @@ class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
 
         # if no bos token id => cannot generate from None
         if config.bos_token_id is None:
-            return
+            self.skipTest(reason="bos_token_id is None")
 
         for model_class in self.greedy_sample_model_classes:
             model = model_class(config).to(torch_device)
@@ -1615,7 +1618,9 @@ class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
 
             self.assertNotIn(config.pad_token_id, output_generate)
 
-    @unittest.skip("MusicgenModel is actually not the base of MusicgenForCausalLM as the latter is a composit model")
+    @unittest.skip(
+        reason="MusicgenModel is actually not the base of MusicgenForCausalLM as the latter is a composit model"
+    )
     def test_save_load_fast_init_from_base(self):
         pass
 
@@ -1934,6 +1939,9 @@ class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin,
     @slow
     # Copied from tests.test_modeling_common.ModelTesterMixin.test_eager_matches_sdpa_inference
     def test_eager_matches_sdpa_inference(self, torch_dtype: str):
+        if not self.has_attentions:
+            self.skipTest(reason="Model architecture does not support attentions")
+
         if not self.all_model_classes[0]._supports_sdpa:
             self.skipTest(f"{self.all_model_classes[0].__name__} does not support SDPA")
 

tests/models/musicgen/test_processing_musicgen.py

@@ -22,12 +22,9 @@ import numpy as np
 
 from transformers import T5Tokenizer, T5TokenizerFast
 from transformers.testing_utils import require_sentencepiece, require_torch
-from transformers.utils.import_utils import is_speech_available, is_torch_available
+from transformers.utils.import_utils import is_speech_available
 
 
-if is_torch_available():
-    pass
-
 if is_speech_available():
     from transformers import EncodecFeatureExtractor, MusicgenProcessor
 

tests/models/musicgen_melody/test_modeling_musicgen_melody.py

@@ -208,7 +208,7 @@ class MusicgenMelodyDecoderTest(ModelTesterMixin, GenerationTesterMixin, unittes
     # Copied from tests.models.musicgen.test_modeling_musicgen.MusicgenDecoderTest.check_training_gradient_checkpointing with Musicgen->MusicgenMelody
     def check_training_gradient_checkpointing(self, gradient_checkpointing_kwargs=None):
         if not self.model_tester.is_training:
-            return
+            self.skipTest(reason="model_tester.is_training is set to False")
 
         config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
         config.use_cache = False
@@ -273,15 +273,15 @@ class MusicgenMelodyDecoderTest(ModelTesterMixin, GenerationTesterMixin, unittes
     def test_inputs_embeds_matches_input_ids(self):
         pass
 
-    @unittest.skip("this model doesn't support all arguments tested")
+    @unittest.skip(reason="this model doesn't support all arguments tested")
     def test_model_outputs_equivalence(self):
         pass
 
-    @unittest.skip("this model has multiple inputs embeds and lm heads that should not be tied")
+    @unittest.skip(reason="this model has multiple inputs embeds and lm heads that should not be tied")
     def test_tie_model_weights(self):
         pass
 
-    @unittest.skip("this model has multiple inputs embeds and lm heads that should not be tied")
+    @unittest.skip(reason="this model has multiple inputs embeds and lm heads that should not be tied")
     def test_tied_weights_keys(self):
         pass
 
@@ -626,6 +626,9 @@ class MusicgenMelodyDecoderTest(ModelTesterMixin, GenerationTesterMixin, unittes
     @slow
     # Copied from tests.models.musicgen.test_modeling_musicgen.MusicgenDecoderTest.test_eager_matches_sdpa_inference
     def test_eager_matches_sdpa_inference(self, torch_dtype: str):
+        if not self.has_attentions:
+            self.skipTest(reason="Model architecture does not support attentions")
+
         if not self.all_model_classes[0]._supports_sdpa:
             self.skipTest(f"{self.all_model_classes[0].__name__} does not support SDPA")
 
@@ -1089,7 +1092,7 @@ class MusicgenMelodyTest(ModelTesterMixin, GenerationTesterMixin, PipelineTester
 
     def check_training_gradient_checkpointing(self, gradient_checkpointing_kwargs=None):
         if not self.model_tester.is_training:
-            return
+            self.skipTest(reason="model_tester.is_training is set to False")
 
         for model_class in self.all_model_classes:
             config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
@@ -1247,27 +1250,27 @@ class MusicgenMelodyTest(ModelTesterMixin, GenerationTesterMixin, PipelineTester
             model = model_class(config)
             self.assertTrue(model.is_gradient_checkpointing)
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tie_model_weights(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tied_model_weights_key_ignore(self):
         pass
 
-    # skip as this model has multiple inputs embeds and lm heads that should not be tied
+    @unittest.skip(reason="MusicGen has multiple inputs embeds and lm heads that should not be tied.")
     def test_tied_weights_keys(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage_checkpoints(self):
         pass
 
-    @unittest.skip("No support for low_cpu_mem_usage=True.")
+    @unittest.skip(reason="No support for low_cpu_mem_usage=True.")
     def test_save_load_low_cpu_mem_usage_no_safetensors(self):
         pass
 
@@ -1553,7 +1556,7 @@ class MusicgenMelodyTest(ModelTesterMixin, GenerationTesterMixin, PipelineTester
 
         # if no bos token id => cannot generate from None
         if config.bos_token_id is None:
-            return
+            self.skipTest(reason="bos_token_id is None")
 
         for model_class in self.greedy_sample_model_classes:
             model = model_class(config).to(torch_device)
@@ -1600,7 +1603,7 @@ class MusicgenMelodyTest(ModelTesterMixin, GenerationTesterMixin, PipelineTester
             self.assertNotIn(config.pad_token_id, output_generate)
 
     @unittest.skip(
-        "MusicgenMelodyModel is actually not the base of MusicgenMelodyForCausalLM as the latter is a composit model"
+        reason="MusicgenMelodyModel is actually not the base of MusicgenMelodyForCausalLM as the latter is a composit model"
     )
     def test_save_load_fast_init_from_base(self):
         pass

tests/models/mvp/test_modeling_mvp.py

@@ -818,6 +818,6 @@ class MvpStandaloneDecoderModelTest(ModelTesterMixin, GenerationTesterMixin, uni
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.create_and_check_decoder_model_attention_mask_past(*config_and_inputs)
 
+    @unittest.skip(reason="Decoder cannot keep gradients")
     def test_retain_grad_hidden_states_attentions(self):
-        # decoder cannot keep gradients
         return

tests/models/mvp/test_tokenization_mvp.py

@@ -146,6 +146,7 @@ class TestTokenizationMvp(TokenizerTesterMixin, unittest.TestCase):
             self.assertTrue((input_ids[:, -1] == tokenizer.eos_token_id).all().item())
             self.assertTrue((labels[:, -1] == tokenizer.eos_token_id).all().item())
 
+    @unittest.skip
     def test_pretokenized_inputs(self):
         pass
 

tests/models/nllb/test_tokenization_nllb.py

@@ -207,7 +207,7 @@ class NllbTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     @require_torch
     def test_prepare_seq2seq_batch(self):
         if not self.test_seq2seq:
-            return
+            self.skipTest(reason="test_seq2seq is set to False")
 
         tokenizers = self.get_tokenizers()
         for tokenizer in tokenizers:
@@ -236,7 +236,7 @@ class NllbTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
                         tgt_lang="ron_Latn",
                     )
                 except NotImplementedError:
-                    return
+                    self.skipTest(reason="Encountered NotImplementedError when calling prepare_seq2seq_batch")
                 self.assertEqual(batch.input_ids.shape[1], 3)
                 self.assertEqual(batch.labels.shape[1], 10)
                 # max_target_length will default to max_length if not specified
@@ -253,7 +253,7 @@ class NllbTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
                 self.assertEqual(batch_encoder_only.attention_mask.shape[1], 3)
                 self.assertNotIn("decoder_input_ids", batch_encoder_only)
 
-    @unittest.skip("Unfortunately way too slow to build a BPE with SentencePiece.")
+    @unittest.skip(reason="Unfortunately way too slow to build a BPE with SentencePiece.")
     def test_save_slow_from_fast_and_reload_fast(self):
         pass
 
@@ -290,7 +290,7 @@ class NllbTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
                     self.assertTrue(special_token_id in p_output)
                     self.assertTrue(special_token_id in cr_output)
 
-    @unittest.skip("Need to fix this after #26538")
+    @unittest.skip(reason="Need to fix this after #26538")
     def test_training_new_tokenizer(self):
         pass
 

tests/models/nllb_moe/test_modeling_nllb_moe.py

@@ -403,7 +403,7 @@ class NllbMoeModelIntegrationTests(unittest.TestCase):
         EXPECTED_LOGTIS = torch.Tensor([-0.3059, 0.0000, 9.3029, 0.6456, -0.9148, 1.7836, 0.6478, 0.9438, -0.5272, -0.6617, -1.2717, 0.4564, 0.1345, -0.2301, -1.0140, 1.1427, -1.5535, 0.1337, 0.2082, -0.8112, -0.3842, -0.3377, 0.1256, 0.6450, -0.0452, 0.0219, 1.4274, -0.4991, -0.2063, -0.4409,])  # fmt: skip
         torch.testing.assert_close(output.logits[1, 0, :30], EXPECTED_LOGTIS, rtol=6e-3, atol=9e-3)
 
-    @unittest.skip("This requires 300GB of RAM")
+    @unittest.skip(reason="This requires 300GB of RAM")
     def test_large_logits(self):
         model = self.big_model
         with torch.no_grad():
@@ -421,7 +421,7 @@ class NllbMoeModelIntegrationTests(unittest.TestCase):
         torch.testing.assert_close(output.last_hidden_state[1, 0, :30], EXPECTED_DECODER_STATE, rtol=6e-3, atol=9e-3)
         torch.testing.assert_close(output.logits[1, 0, :30], EXPECTED_LOGTIS, rtol=6e-3, atol=9e-3)
 
-    @unittest.skip("This requires 300GB of RAM")
+    @unittest.skip(reason="This requires 300GB of RAM")
     def test_seq_to_seq_generation(self):
         model = self.big_model
         tokenizer = NllbTokenizer.from_pretrained("facebook/nllb-moe-54b")

tests/models/nougat/test_tokenization_nougat.py

@@ -102,19 +102,19 @@ class NougatTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
                     padding="max_length",
                 )
 
-    @unittest.skip("NougatTokenizerFast does not have tokenizer_file in its signature")
+    @unittest.skip(reason="NougatTokenizerFast does not have tokenizer_file in its signature")
     def test_rust_tokenizer_signature(self):
         pass
 
-    @unittest.skip("NougatTokenizerFast does not support pretokenized inputs")
+    @unittest.skip(reason="NougatTokenizerFast does not support pretokenized inputs")
     def test_pretokenized_inputs(self):
         pass
 
-    @unittest.skip("NougatTokenizerFast directly inherits from PreTrainedTokenizerFast")
+    @unittest.skip(reason="NougatTokenizerFast directly inherits from PreTrainedTokenizerFast")
     def test_prepare_for_model(self):
         pass
 
-    @unittest.skip("This needs a slow tokenizer. Nougat does not have one!")
+    @unittest.skip(reason="This needs a slow tokenizer. Nougat does not have one!")
     def test_encode_decode_with_spaces(self):
         pass
 

tests/models/olmo/test_modeling_olmo.py

@@ -301,7 +301,7 @@ class OlmoModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.create_and_check_model(*config_and_inputs)
 
-    @unittest.skip("OLMo does not support head pruning.")
+    @unittest.skip(reason="OLMo does not support head pruning.")
     def test_headmasking(self):
         pass
 
@@ -311,7 +311,7 @@ class OlmoModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin
             config_and_inputs[0].position_embedding_type = type
             self.model_tester.create_and_check_model(*config_and_inputs)
 
-    @unittest.skip("OLMo buffers include complex numbers, which breaks this test")
+    @unittest.skip(reason="OLMo buffers include complex numbers, which breaks this test")
     def test_save_load_fast_init_from_base(self):
         pass
 

tests/models/oneformer/test_image_processing_oneformer.py

@@ -213,6 +213,7 @@ class OneFormerImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
 
         return inputs
 
+    @unittest.skip
     def test_init_without_params(self):
         pass