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

Fix typo (#25966)

Browse Source 
* Update feature_extraction_clap.py

* changed all lenght to length
This commit is contained in:
Susnato Dhar
2023-09-05 13:42:25 +05:30
committed by GitHub
parent d8e13b3e04
commit 404ff8fc17
20 changed files with 35 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/legacy/seq2seq/seq2seq_training_args.py
View File

@@ -31,7 +31,7 @@ class Seq2SeqTrainingArguments(TrainingArguments):
label_smoothing (:obj:`float`, `optional`, defaults to 0):
The label smoothing epsilon to apply (if not zero).
sortish_sampler (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to SortishSamler or not. It sorts the inputs according to lenghts in-order to minimizing the padding size.
Whether to SortishSamler or not. It sorts the inputs according to lengths in-order to minimizing the padding size.
predict_with_generate (:obj:`bool`, `optional`, defaults to :obj:`False`):
Whether to use generate to calculate generative metrics (ROUGE, BLEU).
"""

2
examples/pytorch/speech-recognition/run_speech_recognition_ctc.py
View File

@@ -311,7 +311,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]
label_features = [{"input_ids": feature["labels"]} for feature in features]

2
examples/pytorch/speech-recognition/run_speech_recognition_ctc_adapter.py
View File

@@ -307,7 +307,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]
label_features = [{"input_ids": feature["labels"]} for feature in features]

2
examples/research_projects/robust-speech-event/run_speech_recognition_ctc_bnb.py
View File

@@ -292,7 +292,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]
label_features = [{"input_ids": feature["labels"]} for feature in features]

2
examples/research_projects/robust-speech-event/run_speech_recognition_ctc_streaming.py
View File

@@ -284,7 +284,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = []
label_features = []

2
examples/research_projects/wav2vec2/run_asr.py
View File

@@ -254,7 +254,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]
label_features = [{"input_ids": feature["labels"]} for feature in features]

2
examples/research_projects/wav2vec2/run_common_voice.py
View File

@@ -173,7 +173,7 @@ class DataCollatorCTCWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]
label_features = [{"input_ids": feature["labels"]} for feature in features]

2
examples/research_projects/xtreme-s/run_xtreme_s.py
View File

@@ -335,7 +335,7 @@ class SpeechDataCollatorWithPadding:
pad_to_multiple_of_labels: Optional[int] = None
def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# split inputs and labels since they have to be of different lengths and need
# different padding methods
input_features = [{"input_values": feature["input_values"]} for feature in features]

2
src/transformers/models/clap/feature_extraction_clap.py
View File

@@ -51,7 +51,7 @@ class ClapFeatureExtractor(SequenceFeatureExtractor):
Length of the overlaping windows for the STFT used to obtain the Mel Spectrogram. The audio will be split
in smaller `frames` with a step of `hop_length` between each frame.
max_length_s (`int`, defaults to 10):
The maximum input lenght of the model in seconds. This is used to pad the audio.
The maximum input length of the model in seconds. This is used to pad the audio.
fft_window_size (`int`, defaults to 1024):
Size of the window (in samples) on which the Fourier transform is applied. This controls the frequency
resolution of the spectrogram. 400 means that the fourrier transform is computed on windows of 400 samples.

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

@@ -283,8 +283,8 @@ class Data2VecAudioModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -252,8 +252,8 @@ class HubertModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -222,8 +222,8 @@ class SEWModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -243,8 +243,8 @@ class SEWDModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

8
tests/models/speecht5/test_feature_extraction_speecht5.py
View File

@@ -340,7 +340,7 @@ class SpeechT5FeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest
feat_dict["return_attention_mask"] = True
feat_extract = self.feature_extraction_class(**feat_dict)
speech_inputs = self.feat_extract_tester.prepare_inputs_for_target()
input_lenghts = [len(x) for x in speech_inputs]
input_lengths = [len(x) for x in speech_inputs]
input_name = feat_extract.model_input_names[0]
processed = BatchFeature({input_name: speech_inputs})
@@ -350,18 +350,18 @@ class SpeechT5FeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest
processed = feat_extract.pad(processed, padding="longest", return_tensors="np")
self.assertIn("attention_mask", processed)
self.assertListEqual(list(processed.attention_mask.shape), list(processed[input_name].shape[:2]))
self.assertListEqual(processed.attention_mask.sum(-1).tolist(), input_lenghts)
self.assertListEqual(processed.attention_mask.sum(-1).tolist(), input_lengths)
def test_attention_mask_with_truncation_target(self):
feat_dict = self.feat_extract_dict
feat_dict["return_attention_mask"] = True
feat_extract = self.feature_extraction_class(**feat_dict)
speech_inputs = self.feat_extract_tester.prepare_inputs_for_target()
input_lenghts = [len(x) for x in speech_inputs]
input_lengths = [len(x) for x in speech_inputs]
input_name = feat_extract.model_input_names[0]
processed = BatchFeature({input_name: speech_inputs})
max_length = min(input_lenghts)
max_length = min(input_lengths)
feat_extract.feature_size = feat_extract.num_mel_bins # hack!

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

@@ -245,8 +245,8 @@ class UniSpeechModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -265,8 +265,8 @@ class UniSpeechSatModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -404,8 +404,8 @@ class Wav2Vec2ModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -314,8 +314,8 @@ class Wav2Vec2ConformerModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

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

@@ -256,8 +256,8 @@ class WavLMModelTester:
input_values[i, input_lengths[i] :] = 0.0
if max_length_labels[i] < labels.shape[-1]:
# it's important that we make sure that target lenghts are at least
# one shorter than logit lenghts to prevent -inf
# it's important that we make sure that target lengths are at least
# one shorter than logit lengths to prevent -inf
labels[i, max_length_labels[i] - 1 :] = -100
loss = model(input_values, labels=labels).loss

8
tests/test_sequence_feature_extraction_common.py
View File

@@ -391,7 +391,7 @@ class SequenceFeatureExtractionTestMixin(FeatureExtractionSavingTestMixin):
feat_dict["return_attention_mask"] = True
feat_extract = self.feature_extraction_class(**feat_dict)
speech_inputs = self.feat_extract_tester.prepare_inputs_for_common()
input_lenghts = [len(x) for x in speech_inputs]
input_lengths = [len(x) for x in speech_inputs]
input_name = feat_extract.model_input_names[0]
processed = BatchFeature({input_name: speech_inputs})
@@ -399,18 +399,18 @@ class SequenceFeatureExtractionTestMixin(FeatureExtractionSavingTestMixin):
processed = feat_extract.pad(processed, padding="longest", return_tensors="np")
self.assertIn("attention_mask", processed)
self.assertListEqual(list(processed.attention_mask.shape), list(processed[input_name].shape[:2]))
self.assertListEqual(processed.attention_mask.sum(-1).tolist(), input_lenghts)
self.assertListEqual(processed.attention_mask.sum(-1).tolist(), input_lengths)
def test_attention_mask_with_truncation(self):
feat_dict = self.feat_extract_dict
feat_dict["return_attention_mask"] = True
feat_extract = self.feature_extraction_class(**feat_dict)
speech_inputs = self.feat_extract_tester.prepare_inputs_for_common()
input_lenghts = [len(x) for x in speech_inputs]
input_lengths = [len(x) for x in speech_inputs]
input_name = feat_extract.model_input_names[0]
processed = BatchFeature({input_name: speech_inputs})
max_length = min(input_lenghts)
max_length = min(input_lengths)
processed_pad = feat_extract.pad(
processed, padding="max_length", max_length=max_length, truncation=True, return_tensors="np"