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 b7672826.

* Pass trust_remote_code in librispeech_asr_dummy docstrings

* Revert "Pin datasets<2.20.0 for examples"

This reverts commit 833fc17a.

* 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/layoutlmv2/test_processor_layoutlmv2.py

@@ -167,7 +167,7 @@ class LayoutLMv2ProcessorTest(unittest.TestCase):
         from datasets import load_dataset
 
         # set up
-        datasets = load_dataset("nielsr/funsd")
+        datasets = load_dataset("nielsr/funsd", trust_remote_code=True)
         processor = LayoutLMv2Processor.from_pretrained("microsoft/layoutlmv2-base-uncased", revision="no_ocr")
 
         def preprocess_data(examples):
@@ -203,7 +203,7 @@ class LayoutLMv2ProcessorIntegrationTests(unittest.TestCase):
         # we verify our implementation on 2 document images from the DocVQA dataset
         from datasets import load_dataset
 
-        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
+        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test", trust_remote_code=True)
 
         image_1 = Image.open(ds[0]["file"]).convert("RGB")
         image_2 = Image.open(ds[1]["file"]).convert("RGB")

tests/models/layoutlmv3/test_image_processing_layoutlmv3.py

@@ -102,7 +102,7 @@ class LayoutLMv3ImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase)
 
         from datasets import load_dataset
 
-        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
+        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test", trust_remote_code=True)
 
         image = Image.open(ds[0]["file"]).convert("RGB")
 

tests/models/layoutlmv3/test_processor_layoutlmv3.py

@@ -183,7 +183,7 @@ class LayoutLMv3ProcessorIntegrationTests(unittest.TestCase):
         # we verify our implementation on 2 document images from the DocVQA dataset
         from datasets import load_dataset
 
-        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
+        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test", trust_remote_code=True)
 
         image_1 = Image.open(ds[0]["file"]).convert("RGB")
         image_2 = Image.open(ds[1]["file"]).convert("RGB")

tests/models/layoutxlm/test_processor_layoutxlm.py

@@ -160,7 +160,7 @@ class LayoutXLMProcessorTest(unittest.TestCase):
         from datasets import load_dataset
 
         # set up
-        datasets = load_dataset("nielsr/funsd")
+        datasets = load_dataset("nielsr/funsd", trust_remote_code=True)
         processor = LayoutXLMProcessor.from_pretrained("microsoft/layoutxlm-base", apply_ocr=False)
 
         def preprocess_data(examples):
@@ -198,7 +198,7 @@ class LayoutXLMProcessorIntegrationTests(unittest.TestCase):
         # we verify our implementation on 2 document images from the DocVQA dataset
         from datasets import load_dataset
 
-        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
+        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test", trust_remote_code=True)
 
         image_1 = Image.open(ds[0]["file"]).convert("RGB")
         image_2 = Image.open(ds[1]["file"]).convert("RGB")

tests/models/llama/test_tokenization_llama.py

@@ -514,7 +514,7 @@ class LlamaIntegrationTest(unittest.TestCase):
         pyth_tokenizer = self.tokenizer
         rust_tokenizer = self.rust_tokenizer
 
-        dataset = load_dataset("code_x_glue_ct_code_to_text", "go")
+        dataset = load_dataset("google/code_x_glue_ct_code_to_text", "go")
         for item in tqdm.tqdm(dataset["validation"]):
             string = item["code"]
             encoded1 = pyth_tokenizer.encode(string)
@@ -527,7 +527,7 @@ class LlamaIntegrationTest(unittest.TestCase):
 
             self.assertEqual(decoded1, decoded2)
 
-        dataset = load_dataset("xnli", "all_languages")
+        dataset = load_dataset("facebook/xnli", "all_languages")
 
         for item in tqdm.tqdm(dataset["train"]):
             for string in item["premise"].values():

tests/models/mobilevit/test_image_processing_mobilevit.py

@@ -87,7 +87,7 @@ class MobileViTImageProcessingTester(unittest.TestCase):
 
 
 def prepare_semantic_single_inputs():
-    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
+    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image = Image.open(dataset[0]["file"])
     map = Image.open(dataset[1]["file"])
@@ -96,7 +96,7 @@ def prepare_semantic_single_inputs():
 
 
 def prepare_semantic_batch_inputs():
-    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
+    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image1 = Image.open(dataset[0]["file"])
     map1 = Image.open(dataset[1]["file"])

tests/models/perceiver/test_modeling_perceiver.py

@@ -851,7 +851,7 @@ def prepare_img():
 
 # Helper functions for optical flow integration test
 def prepare_optical_flow_images():
-    dataset = load_dataset("hf-internal-testing/fixtures_sintel", split="test")
+    dataset = load_dataset("hf-internal-testing/fixtures_sintel", split="test", trust_remote_code=True)
     image1 = Image.open(dataset[0]["file"]).convert("RGB")
     image2 = Image.open(dataset[0]["file"]).convert("RGB")
 

tests/models/pop2piano/test_feature_extraction_pop2piano.py

@@ -136,7 +136,9 @@ class Pop2PianoFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittes
         self.assertTrue(input_features.extrapolated_beatstep.ndim == 2)
 
     def test_integration(self):
-        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
+        )
         speech_samples = ds.sort("id").select([0])["audio"]
         input_speech = [x["array"] for x in speech_samples][0]
         sampling_rate = [x["sampling_rate"] for x in speech_samples][0]

tests/models/pop2piano/test_processor_pop2piano.py

@@ -111,7 +111,9 @@ class Pop2PianoProcessorTest(unittest.TestCase):
 
     def get_inputs(self):
         """get inputs for both feature extractor and tokenizer"""
-        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
+        )
         speech_samples = ds.sort("id").select([0])["audio"]
         input_speech = [x["array"] for x in speech_samples][0]
         sampling_rate = [x["sampling_rate"] for x in speech_samples][0]

tests/models/seamless_m4t/test_feature_extraction_seamless_m4t.py

@@ -258,7 +258,9 @@ class SeamlessM4TFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unitt
             self.assertTrue(pt_processed.input_features.dtype == torch.float32)
 
     def _load_datasample(self, id):
-        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_sample = ds.sort("id")[id]["audio"]["array"]
 

tests/models/segformer/test_image_processing_segformer.py

@@ -87,7 +87,7 @@ class SegformerImageProcessingTester(unittest.TestCase):
 
 
 def prepare_semantic_single_inputs():
-    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
+    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image = Image.open(dataset[0]["file"])
     map = Image.open(dataset[1]["file"])
@@ -96,7 +96,7 @@ def prepare_semantic_single_inputs():
 
 
 def prepare_semantic_batch_inputs():
-    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
+    dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image1 = Image.open(dataset[0]["file"])
     map1 = Image.open(dataset[1]["file"])

tests/models/sew/test_modeling_sew.py

@@ -497,7 +497,9 @@ class SEWModelIntegrationTest(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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").filter(
             lambda x: x["id"] in [f"1272-141231-000{i}" for i in range(num_samples)]

tests/models/sew_d/test_modeling_sew_d.py

@@ -511,7 +511,9 @@ class SEWDModelIntegrationTest(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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").filter(
             lambda x: x["id"] in [f"1272-141231-000{i}" for i in range(num_samples)]

tests/models/speech_to_text/test_feature_extraction_speech_to_text.py

@@ -259,7 +259,9 @@ class Speech2TextFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unitt
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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/speech_to_text/test_modeling_speech_to_text.py

@@ -791,7 +791,9 @@ class Speech2TextModelIntegrationTests(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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/speech_to_text/test_modeling_tf_speech_to_text.py

@@ -587,7 +587,9 @@ class TFSpeech2TextModelIntegrationTests(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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/speecht5/test_feature_extraction_speecht5.py

@@ -380,7 +380,9 @@ class SpeechT5FeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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/speecht5/test_modeling_speecht5.py

@@ -741,7 +741,9 @@ class SpeechT5ForSpeechToTextIntegrationTests(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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 SpeechT5ForSpeechToSpeechIntegrationTests(unittest.TestCase):
     def _load_datasamples(self, num_samples):
         from datasets import load_dataset
 
-        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/t5/test_tokenization_t5.py

@@ -598,7 +598,7 @@ class CommonSpmIntegrationTests(unittest.TestCase):
         from datasets import load_dataset
         from seqio import SentencePieceVocabulary
 
-        ds = load_dataset("xnli", "all_languages", split="train+test+validation")
+        ds = load_dataset("facebook/xnli", "all_languages", split="train+test+validation")
 
         # TODO @ArthurZucker fix the 3 commented tests with #23909
         input_texts = [

tests/models/udop/test_processor_udop.py

@@ -185,7 +185,7 @@ class UdopProcessorTest(unittest.TestCase):
         from datasets import load_dataset
 
         # set up
-        datasets = load_dataset("nielsr/funsd")
+        datasets = load_dataset("nielsr/funsd", trust_remote_code=True)
         processor = UdopProcessor.from_pretrained("microsoft/udop-large", apply_ocr=False)
 
         def preprocess_data(examples):
@@ -223,7 +223,7 @@ class UdopProcessorIntegrationTests(unittest.TestCase):
         # we verify our implementation on 2 document images from the DocVQA dataset
         from datasets import load_dataset
 
-        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test")
+        ds = load_dataset("hf-internal-testing/fixtures_docvqa", split="test", trust_remote_code=True)
 
         image_1 = Image.open(ds[0]["file"]).convert("RGB")
         image_2 = Image.open(ds[1]["file"]).convert("RGB")