MCPcopy Create free account

hub / github.com/IIP-Sogang/olkavs-avspeech / functions

Functions234 in github.com/IIP-Sogang/olkavs-avspeech

⨍ Functions 234◇ Types & classes 60
↓ 8 callersFunctionget_residual_layer
( num_layer, in_channels, out_channels, kernel_size, dim=1, identity=False)
avsr/models/resnet/resnet.py:5
↓ 6 callersMethodget_no_spaced_text
(self, text)
avsr/utils/metric.py:242
↓ 4 callersFunctionget_metric
(vocab, log_path, unit:str='character', error_type:str='cer')
avsr/utils/getter.py:22
↓ 4 callersFunctionreset_count
()
avsr/utils/korean_funcs.py:83
↓ 4 callersFunctionreset_count
()
vocabulary/utils.py:83
↓ 4 callersMethodstep
Updates the learning rate by taking a step. :param current_step: Optionally specify what step to set the learning rate to.
avsr/scheduler/noam.py:62
↓ 4 callersFunctionvid_length_
(p)
dataset/dataset.py:394
↓ 3 callersMethod__init__
(self, vocab, log_path, unit='character', error_type='cer')
avsr/utils/metric.py:8
↓ 3 callersMethod__init__
(self, *args, **kwargs)
avsr/models/model.py:81
↓ 3 callersMethod__init__
(self, in_channels, out_channels, kernel_size, identity=True)
avsr/models/resnet/resnet.py:19
↓ 3 callersMethod__init__
(self, shape: tuple, contiguous: bool = False)
avsr/models/conformer/modules.py:55
↓ 3 callersMethod__init__
( self, in_channels: int, out_channels: int, kernel_size: int,
avsr/models/conformer/convolution.py:43
↓ 3 callersFunction_parse_audio
(signal, transform, normalize)
dataset/dataset.py:132
↓ 3 callersFunction_parse_transcript
(transcript, sos_id, eos_id)
dataset/dataset.py:140
↓ 3 callersFunction_parse_video
(video_path, is_raw=False)
dataset/dataset.py:111
↓ 3 callersFunctionprepare_dataset
( transcripts_path: str, vocab: Vocabulary, use_audio: bool = True, use_video: bool = True
dataset/dataset.py:54
↓ 2 callersMethod__init__
( self, blank_id = None, )
avsr/utils/loss.py:33
↓ 2 callersMethod__init__
( self, vocab_size : int, decoder_n_layer : int, decoder_d_model : int,
avsr/models/decoder.py:12
↓ 2 callersMethod__init__
( self, front_dim : int, encoder_n_layer : int, encoder_d_model : int, encoder_n_he
avsr/models/encoder.py:51
↓ 2 callersFunctionbuild_model
( vocab_size : int, pad_id : int, architecture : str = 'audio_visual', loss_fn : str = 'hybrid', fr
avsr/utils/model_builder.py:14
↓ 2 callersFunctioncal_rms
(amp)
dataset/augment.py:91
↓ 2 callersMethodget_space_tokenized_text
(self, text)
avsr/utils/metric.py:245
↓ 2 callersFunctiongrp2char
(sentence)
avsr/utils/korean_funcs.py:69
↓ 2 callersMethodjoint
Joint `encoder_outputs` and `decoder_outputs`. Args: encoder_outputs (torch.FloatTensor): A output sequence of encoder.
avsr/models/conformer/model.py:120
↓ 2 callersMethodlabel_to_string
(self, labels)
avsr/utils/korean_funcs.py:152
↓ 2 callersFunctionload_dataset
Provides dictionary of filename and labels Args: transcripts_path (str): path of transcripts Returns: target_dict -
dataset/dataset.py:28
↓ 2 callersMethodmetric
(self, *args, **kwargs)
avsr/utils/metric.py:117
↓ 2 callersFunctionsetup
world_size : number of processes rank : this should be a number between 0 and world_size-1
train.py:30
↓ 2 callersFunctionshow_description
(epoch, total_epoch, it, total_it, lr, loss, mean_loss, _time, ctc_loss=None, att_loss=None, ctc_mean_loss=Non
train.py:70
↓ 1 callersMethod__init__
(self, *args, **kwargs)
avsr/utils/korean_funcs.py:146
↓ 1 callersMethod__init__
( self, d_model: int = 512, num_heads: int = 16, dropout_p: fl
avsr/models/conformer/attention.py:46
↓ 1 callersMethod__init__
( self, encoder_dim: int = 512, num_attention_heads: int = 8,
avsr/models/conformer/encoder.py:56
↓ 1 callersMethod__init__
(self, dim: int)
avsr/models/conformer/activation.py:36
↓ 1 callersMethod__init__
(self, *args, **kwargs)
vocabulary/utils.py:145
↓ 1 callersMethod_beam_search
( self, features, current_hypothesis, length:int, batch_size:int,
avsr/utils/search.py:197
↓ 1 callersFunction_collate_fn
functions that pad to the maximum sequence length
dataset/dataset.py:387
↓ 1 callersFunction_crop_and_save_audio_segment
(path:str, array:ndarray, sr:int, start:float, end:float)
preprocess.py:101
↓ 1 callersFunction_crop_and_save_video_segment
( path:str, array:ndarray, fps:float, start:float, end:float, resize_shape:tuple=(224,224))
preprocess.py:106
↓ 1 callersMethod_get_distance
Provides total character distance between targets & y_hats Args: targets (torch.Tensor): set of ground truth
avsr/utils/metric.py:73
↓ 1 callersFunction_get_sentence_info
(label:dict)
preprocess.py:80
↓ 1 callersFunction_infer_collate_fn
functions that pad to the maximum sequence length
dataset/dataset.py:480
↓ 1 callersMethod_init_ctc
( self, features, )
avsr/utils/search.py:264
↓ 1 callersFunction_load_and_crop_video
(video_path:str, bboxes:List[List[int]], resize_shape=(224,224))
preprocess.py:89
↓ 1 callersFunction_parse_korean_transcript
(korean_transcript, sos_id, eos_id)
dataset/dataset.py:149
↓ 1 callersMethod_relative_shift
(self, pos_score: Tensor)
avsr/models/conformer/attention.py:118
↓ 1 callersMethod_search
( self, features = None, beam_size : int = 1, D_end : int = -4, M
avsr/utils/search.py:124
↓ 1 callersMethod_search_batch
( self, video_inputs, video_input_lengths, audio_inputs, audio_in
avsr/utils/search.py:103
↓ 1 callersMethod_text2vec
(self, texts, vocab2idx:Dict[str,int], pad_id:int = 0, return_length:boo
avsr/models/decoder.py:89
↓ 1 callersFunctioncal_adjusted_rms
(clean_rms, snr)
dataset/augment.py:79
↓ 1 callersFunctionchar2grp
(test_keyword)
vocabulary/utils.py:46
↓ 1 callersFunctioncleanup
()
train.py:51
↓ 1 callersFunctioncount_parameters
(model)
avsr/utils/model_builder.py:10
↓ 1 callersFunctionctc_label_scores
Compute the CTC label scores for each hypothesized character in a vectorized manner. Args: g: string representation of the previ
avsr/utils/search.py:326
↓ 1 callersMethoddecode
Decode `encoder_outputs`. Args: encoder_output (torch.FloatTensor): A output sequence of encoder. `FloatTensor` of size
avsr/models/conformer/model.py:174
↓ 1 callersFunctionend_detect
(hypothesis, length, M=12, threshold=-4, progressive=None)
avsr/utils/search.py:16
↓ 1 callersFunctionfindGraphemeTokens
(text)
avsr/utils/korean_funcs.py:33
↓ 1 callersFunctionfindGraphemeTokens
(text)
vocabulary/utils.py:33
↓ 1 callersMethodforward
(self, labels, inputs, pad_id=None, **kwargs)
avsr/models/decoder.py:57
↓ 1 callersMethodfusion
(self, visual_feature, audio_feature)
avsr/models/encoder.py:39
↓ 1 callersMethodgenerate_square_subsequent_mask
(self, sz)
avsr/models/decoder.py:133
↓ 1 callersFunctionget_args
()
train.py:365
↓ 1 callersFunctionget_args
()
preprocess.py:218
↓ 1 callersFunctionget_args
()
inference.py:281
↓ 1 callersMethodget_attn_pad_mask
(self, seq, pad)
avsr/models/decoder.py:138
↓ 1 callersFunctionget_criterion
( loss_fn : str = 'hybrid', ignore_index : int = None, label_smoothing : float = 0.0, blan
avsr/utils/getter.py:8
↓ 1 callersMethodget_lr
Gets a list of the current learning rates.
avsr/scheduler/noam.py:58
↓ 1 callersFunctionget_optimizer
( params, learning_rate, scheduler : str = 'noam', epochs : int = None, warmup : int
avsr/utils/getter.py:26
↓ 1 callersMethodget_scores
( self, features, seq_length:int, g, g_att:List[float]=None,
avsr/utils/search.py:287
↓ 1 callersMethodget_space_normalized_text
(self, ref, hyp)
avsr/utils/metric.py:183
↓ 1 callersFunctioninfer
(config, model, vocab, dataset, scores, sampler=None, device='cpu')
inference.py:83
↓ 1 callersFunctionjoin_key
(a: str, b: int)
avsr/utils/search.py:400
↓ 1 callersFunctionload_audio
(audio_path)
preprocess.py:84
↓ 1 callersFunctionload_checkpoint
(model, checkpoint_path, device='cpu')
inference.py:40
↓ 1 callersFunctionload_ddp_checkpoint
(rank, model, checkpoint_path, epoch)
train.py:62
↓ 1 callersFunctionload_label
(filepath, encoding='utf-8')
preprocess.py:136
↓ 1 callersMethodload_vocab
Provides char2id, id2char Args: label_path (str): csv file with character labels encoding (str): encoding
avsr/utils/korean_funcs.py:206
↓ 1 callersMethodload_vocab
Provides char2id, id2char Args: label_path (str): csv file with character labels encoding (str): encoding
vocabulary/utils.py:205
↓ 1 callersFunctionmain
(args)
train.py:295
↓ 1 callersFunctionmain
(args, loop=None)
inference.py:173
↓ 1 callersFunctionnum2kor
(num)
preprocess.py:169
↓ 1 callersMethodparse_audio
(self,audio_path: str, augment_method)
dataset/dataset.py:277
↓ 1 callersMethodparse_korean_transcripts
(self, korean_transcript)
dataset/dataset.py:297
↓ 1 callersMethodparse_transcript
(self, transcript)
dataset/dataset.py:293
↓ 1 callersMethodparse_video
(self, video_path: str)
dataset/dataset.py:289
↓ 1 callersFunctionpreprocess_source_data
( root_dir=".", save_dir="./save", transcription_label_path="./label.txt", vocabulary_
preprocess.py:20
↓ 1 callersFunctionrefine_transcription
(transcript:str)
preprocess.py:196
↓ 1 callersFunctionsave_checkpoint
(model, checkpoint_path, epoch)
train.py:55
↓ 1 callersFunctionsave_video_mp4
(save_path, video_frames, fps=30, resize_shape=(224,224))
preprocess.py:124
↓ 1 callersFunctionselect_search
( model, vocab_size, method:str='default', pad_id:int=0, sos_id:int=1, eos_id:in
avsr/utils/getter.py:52
↓ 1 callersFunctionsentence_to_target
(sentence, char2id, unit='grapheme')
preprocess.py:154
↓ 1 callersMethodset_epoch
(self, epoch:int)
dataset/sampler.py:15
↓ 1 callersFunctionshow_description
(it, total_it, ger, mean_ger, cer, mean_cer, wer, mean_wer, swer, mean_swer, _time)
inference.py:28
↓ 1 callersFunctionsynthesize
(clean_amp, noise_amp, snr)
dataset/augment.py:94
↓ 1 callersFunctionunzip_groups
(transcript)
preprocess.py:206
Method__call__
(self, *args, **kwargs)
avsr/scheduler/abstract.py:5
Method__call__
(self, targets, outputs, target_lengths=None, output_lengths=None, show=False, file_path=None)
avsr/utils/metric.py:38
Method__call__
Calculating character error rate
avsr/utils/metric.py:65
Method__call__
( self, video_inputs, video_input_lengths, audio_inputs, audio_in
avsr/utils/search.py:68
Method__call__
(self, signal)
dataset/feature.py:47
Method__call__
(self, signal)
dataset/feature.py:94
next →1–100 of 234, ranked by callers