SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Remove trust_remote_code when loading Libri Dummy (#31748)

Browse Source 
* [whisper integration] use parquet dataset for testing

* propagate to others

* more propagation

* last one
This commit is contained in:
Sanchit Gandhi
2024-07-23 14:54:38 +08:00
committed by GitHub
parent 3aefb4ec7f
commit f83c6f1d02
56 changed files with 110 additions and 254 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/models/audio_spectrogram_transformer/test_feature_extraction_audio_spectrogram_transformer.py
View File

@@ -153,9 +153,7 @@ class ASTFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.Test
def _load_datasamples(self, num_samples):
from datasets import load_dataset
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/clap/test_feature_extraction_clap.py
View File

@@ -164,9 +164,7 @@ class ClapFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.Tes
# Copied from tests.models.whisper.test_feature_extraction_whisper.WhisperFeatureExtractionTest._load_datasamples
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

16
tests/models/clap/test_modeling_clap.py
View File

@@ -665,9 +665,7 @@ class ClapModelIntegrationTest(unittest.TestCase):
"repeat": 0.0023,
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
audio_sample = librispeech_dummy[-1]
model_id = "laion/clap-htsat-unfused"
@@ -694,9 +692,7 @@ class ClapModelIntegrationTest(unittest.TestCase):
"pad": -0.000379,
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
audio_sample = librispeech_dummy[-1]
model_id = "laion/clap-htsat-fused"
@@ -723,9 +719,7 @@ class ClapModelIntegrationTest(unittest.TestCase):
"pad": 0.0006,
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
audio_samples = [sample["array"] for sample in librispeech_dummy[0:4]["audio"]]
model_id = "laion/clap-htsat-fused"
@@ -752,9 +746,7 @@ class ClapModelIntegrationTest(unittest.TestCase):
"pad": 0.0019,
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
audio_samples = [sample["array"] for sample in librispeech_dummy[0:4]["audio"]]
model_id = "laion/clap-htsat-unfused"

4
tests/models/clvp/test_feature_extraction_clvp.py
View File

@@ -209,9 +209,7 @@ class ClvpFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.Tes
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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
ds = ds.cast_column("audio", Audio(sampling_rate=22050))
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

8
tests/models/clvp/test_modeling_clvp.py
View File

@@ -371,9 +371,7 @@ class ClvpModelForConditionalGenerationTester:
def prepare_config_and_inputs(self):
_, input_ids, attention_mask = self.clvp_encoder_tester.prepare_config_and_inputs()
ds = datasets.load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = datasets.load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
ds = ds.cast_column("audio", datasets.Audio(sampling_rate=22050))
_, audio, sr = ds.sort("id").select(range(1))[:1]["audio"][0].values()
@@ -555,9 +553,7 @@ class ClvpModelForConditionalGenerationTest(ModelTesterMixin, unittest.TestCase)
class ClvpIntegrationTest(unittest.TestCase):
def setUp(self):
self.text = "This is an example text."
ds = datasets.load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = datasets.load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
ds = ds.cast_column("audio", datasets.Audio(sampling_rate=22050))
_, self.speech_samples, self.sr = ds.sort("id").select(range(1))[:1]["audio"][0].values()

4
tests/models/data2vec/test_modeling_data2vec_audio.py
View File

@@ -694,9 +694,7 @@ class Data2VecAudioUtilsTest(unittest.TestCase):
@slow
class Data2VecAudioModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/encodec/test_feature_extraction_encodec.py
View File

@@ -138,9 +138,7 @@ class EnCodecFeatureExtractionTest(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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
audio_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

12
tests/models/encodec/test_modeling_encodec.py
View File

@@ -461,9 +461,7 @@ class EncodecIntegrationTest(unittest.TestCase):
"1.5": [371955],
"24.0": [6659962],
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
model_id = "facebook/encodec_24khz"
model = EncodecModel.from_pretrained(model_id).to(torch_device)
@@ -517,9 +515,7 @@ class EncodecIntegrationTest(unittest.TestCase):
"3.0": [144259, 146765, 156435, 176871, 161971],
"24.0": [1568553, 1294948, 1306190, 1464747, 1663150],
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
model_id = "facebook/encodec_48khz"
model = EncodecModel.from_pretrained(model_id).to(torch_device)
@@ -581,9 +577,7 @@ class EncodecIntegrationTest(unittest.TestCase):
[85561, 81870, 76953, 48967, 79315, 85442, 81479, 107241],
],
}
librispeech_dummy = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
librispeech_dummy = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
model_id = "facebook/encodec_48khz"
model = EncodecModel.from_pretrained(model_id).to(torch_device)

4
tests/models/hubert/test_modeling_hubert.py
View File

@@ -753,9 +753,7 @@ class HubertModelIntegrationTest(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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/hubert/test_modeling_tf_hubert.py
View File

@@ -609,9 +609,7 @@ class TFHubertModelIntegrationTest(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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/pop2piano/test_feature_extraction_pop2piano.py
View File

@@ -136,9 +136,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
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]

4
tests/models/pop2piano/test_processor_pop2piano.py
View File

@@ -111,9 +111,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
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]

4
tests/models/seamless_m4t/test_feature_extraction_seamless_m4t.py
View File

@@ -258,9 +258,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_sample = ds.sort("id")[id]["audio"]["array"]

4
tests/models/sew/test_modeling_sew.py
View File

@@ -494,9 +494,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/sew_d/test_modeling_sew_d.py
View File

@@ -508,9 +508,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/speech_to_text/test_feature_extraction_speech_to_text.py
View File

@@ -259,9 +259,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/speech_to_text/test_modeling_speech_to_text.py
View File

@@ -793,9 +793,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/speech_to_text/test_modeling_tf_speech_to_text.py
View File

@@ -587,9 +587,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/speecht5/test_feature_extraction_speecht5.py
View File

@@ -380,9 +380,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

8
tests/models/speecht5/test_modeling_speecht5.py
View File

@@ -744,9 +744,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
@@ -1771,9 +1769,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/unispeech/test_modeling_unispeech.py
View File

@@ -549,9 +549,7 @@ class UniSpeechRobustModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.T
@slow
class UniSpeechModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/unispeech_sat/test_modeling_unispeech_sat.py
View File

@@ -806,9 +806,7 @@ class UniSpeechSatRobustModelTest(ModelTesterMixin, unittest.TestCase):
@slow
class UniSpeechSatModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/univnet/test_feature_extraction_univnet.py
View File

@@ -327,9 +327,7 @@ class UnivNetFeatureExtractionTest(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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
ds = ds.cast_column("audio", Audio(sampling_rate=self.feat_extract_tester.sampling_rate))
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/univnet/test_modeling_univnet.py
View File

@@ -216,9 +216,7 @@ class UnivNetModelIntegrationTests(unittest.TestCase):
torch.cuda.empty_cache()
def _load_datasamples(self, num_samples, sampling_rate=24000):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
ds = ds.cast_column("audio", Audio(sampling_rate=sampling_rate))
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/wav2vec2/test_modeling_flax_wav2vec2.py
View File

@@ -489,9 +489,7 @@ class FlaxWav2Vec2UtilsTest(unittest.TestCase):
@slow
class FlaxWav2Vec2ModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/wav2vec2/test_modeling_tf_wav2vec2.py
View File

@@ -716,9 +716,7 @@ class TFWav2Vec2ModelIntegrationTest(unittest.TestCase):
gc.collect()
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/wav2vec2/test_modeling_wav2vec2.py
View File

@@ -1464,9 +1464,7 @@ class Wav2Vec2ModelIntegrationTest(unittest.TestCase):
backend_empty_cache(torch_device)
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/wav2vec2_bert/test_modeling_wav2vec2_bert.py
View File

@@ -855,9 +855,7 @@ class Wav2Vec2BertUtilsTest(unittest.TestCase):
@slow
class Wav2Vec2BertModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)])
speech_samples = speech_samples[:num_samples]["audio"]

4
tests/models/wav2vec2_conformer/test_modeling_wav2vec2_conformer.py
View File

@@ -863,9 +863,7 @@ class Wav2Vec2ConformerUtilsTest(unittest.TestCase):
@slow
class Wav2Vec2ConformerModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)])
speech_samples = speech_samples[:num_samples]["audio"]

4
tests/models/wavlm/test_modeling_wavlm.py
View File

@@ -491,9 +491,7 @@ class WavLMModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
@slow
class WavLMModelIntegrationTest(unittest.TestCase):
def _load_datasamples(self, num_samples):
ds = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# 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)]

4
tests/models/whisper/test_feature_extraction_whisper.py
View File

@@ -215,9 +215,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

4
tests/models/whisper/test_modeling_flax_whisper.py
View File

@@ -410,9 +410,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

2
tests/models/whisper/test_modeling_tf_whisper.py
View File

@@ -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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]

26
tests/models/whisper/test_modeling_whisper.py
View File

@@ -1835,9 +1835,7 @@ 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", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# automatic decoding with librispeech
speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
@@ -2718,9 +2716,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
)
assistant_model.to(torch_device)
dataset = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
sample = dataset[0]["audio"]
input_features = processor(sample["array"], return_tensors="pt", sampling_rate=16_000).input_features
@@ -2769,9 +2765,7 @@ class WhisperModelIntegrationTests(unittest.TestCase):
)
assistant_model.to(torch_device)
dataset = load_dataset(
"hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True
)
dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
sample = dataset[0]["audio"]
input_features = processor(sample["array"], return_tensors="pt", sampling_rate=16_000).input_features
@@ -2812,7 +2806,7 @@ 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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
input_features = processor(
@@ -2848,9 +2842,7 @@ 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]", trust_remote_code=True
)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation[:-1]")
one_audio = np.concatenate([x["array"] for x in ds["audio"]], dtype=np.float32)
first_text = ds[0]["text"].lower()
@@ -2901,7 +2893,7 @@ 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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
input_features = processor(
@@ -2983,7 +2975,7 @@ 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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
input_features = processor(
@@ -3025,7 +3017,7 @@ 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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
audios = []
audios.append(one_audio[110000:])
@@ -3079,7 +3071,7 @@ 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", trust_remote_code=True)
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean")
one_audio = np.concatenate([x["array"] for x in ds["validation"]["audio"]], dtype=np.float32)
audios = []
audios.append(one_audio[110000:])