Pass datasets trust_remote_code (#31406)

* Pass datasets trust_remote_code

* Pass trust_remote_code in more tests

* Add trust_remote_dataset_code arg to some tests

* Revert "Temporarily pin datasets upper version to fix CI"

This reverts commit b7672826ca.

* Pass trust_remote_code in librispeech_asr_dummy docstrings

* Revert "Pin datasets<2.20.0 for examples"

This reverts commit 833fc17a3e.

* Pass trust_remote_code to all examples

* Revert "Add trust_remote_dataset_code arg to some tests" to research_projects

* Pass trust_remote_code to tests

* Pass trust_remote_code to docstrings

* Fix flax examples tests requirements

* Pass trust_remote_dataset_code arg to tests

* Replace trust_remote_dataset_code with trust_remote_code in one example

* Fix duplicate trust_remote_code

* Replace args.trust_remote_dataset_code with args.trust_remote_code

* Replace trust_remote_dataset_code with trust_remote_code in parser

* Replace trust_remote_dataset_code with trust_remote_code in dataclasses

* Replace trust_remote_dataset_code with trust_remote_code arg

tests/models/whisper/test_feature_extraction_whisper.py

@@ -215,7 +215,9 @@ class WhisperFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.
             self.assertTrue(pt_processed.input_features.dtype == torch.float32)
 
     def _load_datasamples(self, num_samples):
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        ds = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
+        )
         # automatic decoding with librispeech
         speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
 

tests/models/whisper/test_modeling_flax_whisper.py

@@ -410,7 +410,9 @@ class FlaxWhisperModelIntegrationTest(unittest.TestCase):
         return WhisperProcessor.from_pretrained("openai/whisper-base")
 
     def _load_datasamples(self, num_samples):
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        ds = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
+        )
         # automatic decoding with librispeech
         speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
 
@@ -561,7 +563,7 @@ class FlaxWhisperModelIntegrationTest(unittest.TestCase):
         processor = WhisperProcessor.from_pretrained("openai/whisper-large")
         model = FlaxWhisperForConditionalGeneration.from_pretrained("openai/whisper-large", from_pt=True)
 
-        ds = load_dataset("common_voice", "ja", split="test", streaming=True)
+        ds = load_dataset("legacy-datasets/common_voice", "ja", split="test", streaming=True, trust_remote_code=True)
         ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16_000))
         input_speech = next(iter(ds))["audio"]["array"]
         input_features = processor.feature_extractor(raw_speech=input_speech, return_tensors="np")

tests/models/whisper/test_modeling_tf_whisper.py

@@ -704,7 +704,7 @@ class TFWhisperModelTest(TFModelTesterMixin, PipelineTesterMixin, unittest.TestC
 
 
 def _load_datasamples(num_samples):
-    ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+    ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True)
     # automatic decoding with librispeech
     speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
 
@@ -795,7 +795,7 @@ def _test_large_generation_multilingual(in_queue, out_queue, timeout):
         processor = WhisperProcessor.from_pretrained("openai/whisper-large")
         model = TFWhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
 
-        ds = load_dataset("common_voice", "ja", split="test", streaming=True)
+        ds = load_dataset("legacy-datasets/common_voice", "ja", split="test", streaming=True, trust_remote_code=True)
         ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16_000))
         input_speech = next(iter(ds))["audio"]["array"]
         input_features = processor.feature_extractor(raw_speech=input_speech, return_tensors="tf").input_features

tests/models/whisper/test_modeling_whisper.py

@@ -1552,7 +1552,9 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         return WhisperProcessor.from_pretrained("openai/whisper-base")
 
     def _load_datasamples(self, num_samples):
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        ds = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
+        )
         # automatic decoding with librispeech
         speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
 
@@ -1763,7 +1765,9 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
         model.to(torch_device)
 
-        ds = load_dataset("facebook/multilingual_librispeech", "german", split="test", streaming=True)
+        ds = load_dataset(
+            "facebook/multilingual_librispeech", "german", split="test", streaming=True, trust_remote_code=True
+        )
         ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16_000))
 
         input_speech = next(iter(ds))["audio"]["array"]
@@ -1830,7 +1834,14 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model.to(torch_device)
 
         token = os.getenv("HF_HUB_READ_TOKEN", True)
-        ds = load_dataset("mozilla-foundation/common_voice_6_1", "ja", split="test", streaming=True, token=token)
+        ds = load_dataset(
+            "mozilla-foundation/common_voice_6_1",
+            "ja",
+            split="test",
+            streaming=True,
+            token=token,
+            trust_remote_code=True,
+        )
         ds = ds.cast_column("audio", datasets.Audio(sampling_rate=16_000))
 
         input_speech = next(iter(ds))["audio"]["array"]
@@ -2358,7 +2369,9 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         )
         assistant_model.to(torch_device)
 
-        dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        dataset = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
+        )
         sample = dataset[0]["audio"]
 
         input_features = processor(sample["array"], return_tensors="pt", sampling_rate=16_000).input_features
@@ -2407,7 +2420,9 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         )
         assistant_model.to(torch_device)
 
-        dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+        dataset = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
+        )
         sample = dataset[0]["audio"]
 
         input_features = processor(sample["array"], return_tensors="pt", sampling_rate=16_000).input_features
@@ -2448,7 +2463,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny.en")
         model = model.to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", trust_remote_code=True)
         one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
 
         input_features = processor(
@@ -2484,7 +2499,9 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         prompt = "Mr. Kilter, Brionno."  # let's force Quilter -> Kilter, Brion -> Brionno
         prompt_ids = processor.get_prompt_ids(prompt, return_tensors="pt").to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation[:-1]")
+        ds = load_dataset(
+            "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation[:-1]", trust_remote_code=True
+        )
         one_audio = np.concatenate([x["array"] for x in ds["audio"]], dtype=np.float32)
 
         first_text = ds[0]["text"].lower()
@@ -2535,7 +2552,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny.en")
         model = model.to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", trust_remote_code=True)
         one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
 
         input_features = processor(
@@ -2568,7 +2585,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny.en")
         model = model.to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", trust_remote_code=True)
         one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
 
         input_features = processor(
@@ -2610,7 +2627,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny.en")
         model = model.to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", trust_remote_code=True)
         one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
         audios = []
         audios.append(one_audio[110000:])
@@ -2664,7 +2681,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
         model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny")
         model = model.to(torch_device)
 
-        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
+        ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", trust_remote_code=True)
         one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
         audios = []
         audios.append(one_audio[110000:])