Rolv-Arild commited on
Commit
0614925
1 Parent(s): aa7bcfb

Training in progress, step 500

Browse files
.gitattributes CHANGED
@@ -29,3 +29,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
29
  *.zip filter=lfs diff=lfs merge=lfs -text
30
  *.zst filter=lfs diff=lfs merge=lfs -text
31
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
29
  *.zip filter=lfs diff=lfs merge=lfs -text
30
  *.zst filter=lfs diff=lfs merge=lfs -text
31
  *tfevents* filter=lfs diff=lfs merge=lfs -text
32
+ *.wandb filter=lfs diff=lfs merge=lfs -text
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ checkpoint-*/
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<s>": 39, "</s>": 40}
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-xls-r-1b",
3
+ "activation_dropout": 0.055,
4
+ "adapter_kernel_size": 3,
5
+ "adapter_stride": 2,
6
+ "add_adapter": false,
7
+ "apply_spec_augment": true,
8
+ "architectures": [
9
+ "Wav2Vec2ForCTC"
10
+ ],
11
+ "attention_dropout": 0.094,
12
+ "bos_token_id": 1,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 1024,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
17
+ "conv_dim": [
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512
25
+ ],
26
+ "conv_kernel": [
27
+ 10,
28
+ 3,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 2,
33
+ 2
34
+ ],
35
+ "conv_stride": [
36
+ 5,
37
+ 2,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2
43
+ ],
44
+ "ctc_loss_reduction": "mean",
45
+ "ctc_zero_infinity": true,
46
+ "diversity_loss_weight": 0.1,
47
+ "do_stable_layer_norm": true,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_dropout": 0.0,
51
+ "feat_extract_norm": "layer",
52
+ "feat_proj_dropout": 0.04,
53
+ "feat_quantizer_dropout": 0.0,
54
+ "final_dropout": 0.0,
55
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.047,
57
+ "hidden_size": 1280,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 5120,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.041,
62
+ "mask_feature_length": 64,
63
+ "mask_feature_min_masks": 0,
64
+ "mask_feature_prob": 0.25,
65
+ "mask_time_length": 10,
66
+ "mask_time_min_masks": 2,
67
+ "mask_time_prob": 0.082,
68
+ "model_type": "wav2vec2",
69
+ "num_adapter_layers": 3,
70
+ "num_attention_heads": 16,
71
+ "num_codevector_groups": 2,
72
+ "num_codevectors_per_group": 320,
73
+ "num_conv_pos_embedding_groups": 16,
74
+ "num_conv_pos_embeddings": 128,
75
+ "num_feat_extract_layers": 7,
76
+ "num_hidden_layers": 48,
77
+ "num_negatives": 100,
78
+ "output_hidden_size": 1280,
79
+ "pad_token_id": 38,
80
+ "proj_codevector_dim": 1024,
81
+ "tdnn_dilation": [
82
+ 1,
83
+ 2,
84
+ 3,
85
+ 1,
86
+ 1
87
+ ],
88
+ "tdnn_dim": [
89
+ 512,
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 1500
94
+ ],
95
+ "tdnn_kernel": [
96
+ 5,
97
+ 3,
98
+ 3,
99
+ 1,
100
+ 1
101
+ ],
102
+ "torch_dtype": "float32",
103
+ "transformers_version": "4.18.0",
104
+ "use_weighted_layer_sum": false,
105
+ "vocab_size": 41,
106
+ "xvector_output_dim": 512
107
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a7668aeda50b6bccf6c723f8337046bbe416bd5e5112f197586ade2af090293
3
+ size 3850475057
run.sh ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ WANDB_ENTITY=NbAiLab WANDB_PROJECT=wav2vec2 python run_speech_recognition_ctc.py \
2
+ --model_name_or_path="facebook/wav2vec2-xls-r-1b" \
3
+ --hub_model_id="NbAiLab/wav2vec2-1b-nst" \
4
+ --dataset_name="NbAiLab/NST" \
5
+ --dataset_config="no-close" \
6
+ --output_dir="./" \
7
+ --overwrite_output_dir \
8
+ --num_train_epochs="40" \
9
+ --per_device_train_batch_size="12" \
10
+ --per_device_eval_batch_size="12" \
11
+ --gradient_accumulation_steps="2" \
12
+ --learning_rate="2e-5" \
13
+ --warmup_steps="2000" \
14
+ --length_column_name="input_length" \
15
+ --evaluation_strategy="steps" \
16
+ --text_column_name="text" \
17
+ --save_steps="500" \
18
+ --eval_steps="500" \
19
+ --logging_steps="100" \
20
+ --layerdrop="0.041" \
21
+ --attention_dropout="0.094" \
22
+ --activation_dropout="0.055" \
23
+ --hidden_dropout="0.047" \
24
+ --save_total_limit="3" \
25
+ --freeze_feature_encoder \
26
+ --feat_proj_dropout="0.04" \
27
+ --mask_time_prob="0.082" \
28
+ --mask_time_length="10" \
29
+ --mask_feature_prob="0.25" \
30
+ --mask_feature_length="64" \
31
+ --gradient_checkpointing \
32
+ --min_duration_in_seconds="0.5" \
33
+ --max_duration_in_seconds="30.0" \
34
+ --use_auth_token \
35
+ --seed="42" \
36
+ --fp16 \
37
+ --group_by_length \
38
+ --do_train --do_eval \
39
+ --push_to_hub \
40
+ --preprocessing_num_workers="32" \
41
+ --ctc_zero_infinity
run_speech_recognition_ctc.py ADDED
@@ -0,0 +1,773 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+ # coding=utf-8
3
+ # Copyright 2021 The HuggingFace Inc. team. All rights reserved.
4
+ #
5
+ # Licensed under the Apache License, Version 2.0 (the "License");
6
+ # you may not use this file except in compliance with the License.
7
+ # You may obtain a copy of the License at
8
+ #
9
+ # http://www.apache.org/licenses/LICENSE-2.0
10
+ #
11
+ # Unless required by applicable law or agreed to in writing, software
12
+ # distributed under the License is distributed on an "AS IS" BASIS,
13
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14
+ # See the License for the specific language governing permissions and
15
+
16
+ """ Fine-tuning a 🤗 Transformers CTC model for automatic speech recognition"""
17
+
18
+ import functools
19
+ import json
20
+ import logging
21
+ import os
22
+ import re
23
+ import sys
24
+ import warnings
25
+ from dataclasses import dataclass, field
26
+ from typing import Dict, List, Optional, Union
27
+
28
+ import datasets
29
+ import numpy as np
30
+ import torch
31
+ from datasets import DatasetDict, load_dataset, load_metric
32
+
33
+ import transformers
34
+ from transformers import (
35
+ AutoConfig,
36
+ AutoFeatureExtractor,
37
+ AutoModelForCTC,
38
+ AutoProcessor,
39
+ AutoTokenizer,
40
+ HfArgumentParser,
41
+ Trainer,
42
+ TrainingArguments,
43
+ Wav2Vec2Processor,
44
+ set_seed,
45
+ )
46
+ from transformers.trainer_utils import get_last_checkpoint, is_main_process
47
+ from transformers.utils import check_min_version
48
+ from transformers.utils.versions import require_version
49
+
50
+
51
+ # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
52
+ check_min_version("4.16.0.dev0")
53
+
54
+ require_version("datasets>=1.13.3", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
55
+
56
+
57
+ logger = logging.getLogger(__name__)
58
+
59
+
60
+ def list_field(default=None, metadata=None):
61
+ return field(default_factory=lambda: default, metadata=metadata)
62
+
63
+
64
+ @dataclass
65
+ class ModelArguments:
66
+ """
67
+ Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
68
+ """
69
+
70
+ model_name_or_path: str = field(
71
+ metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
72
+ )
73
+ tokenizer_name_or_path: Optional[str] = field(
74
+ default=None,
75
+ metadata={"help": "Path to pretrained tokenizer or tokenizer identifier from huggingface.co/models"},
76
+ )
77
+ cache_dir: Optional[str] = field(
78
+ default=None,
79
+ metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
80
+ )
81
+ freeze_feature_encoder: bool = field(
82
+ default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
83
+ )
84
+ attention_dropout: float = field(
85
+ default=0.0, metadata={"help": "The dropout ratio for the attention probabilities."}
86
+ )
87
+ activation_dropout: float = field(
88
+ default=0.0, metadata={"help": "The dropout ratio for activations inside the fully connected layer."}
89
+ )
90
+ feat_proj_dropout: float = field(default=0.0, metadata={"help": "The dropout ratio for the projected features."})
91
+ hidden_dropout: float = field(
92
+ default=0.0,
93
+ metadata={
94
+ "help": "The dropout probability for all fully connected layers in the embeddings, encoder, and pooler."
95
+ },
96
+ )
97
+ final_dropout: float = field(
98
+ default=0.0,
99
+ metadata={"help": "The dropout probability for the final projection layer."},
100
+ )
101
+ mask_time_prob: float = field(
102
+ default=0.05,
103
+ metadata={
104
+ "help": "Probability of each feature vector along the time axis to be chosen as the start of the vector"
105
+ "span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"
106
+ "vectors will be masked along the time axis."
107
+ },
108
+ )
109
+ mask_time_length: int = field(
110
+ default=10,
111
+ metadata={"help": "Length of vector span to mask along the time axis."},
112
+ )
113
+ mask_feature_prob: float = field(
114
+ default=0.0,
115
+ metadata={
116
+ "help": "Probability of each feature vector along the feature axis to be chosen as the start of the vector"
117
+ "span to be masked. Approximately ``mask_feature_prob * sequence_length // mask_feature_length`` feature bins will be masked along the time axis."
118
+ },
119
+ )
120
+ mask_feature_length: int = field(
121
+ default=10,
122
+ metadata={"help": "Length of vector span to mask along the feature axis."},
123
+ )
124
+ layerdrop: float = field(default=0.0, metadata={"help": "The LayerDrop probability."})
125
+ ctc_loss_reduction: Optional[str] = field(
126
+ default="mean", metadata={"help": "The way the ctc loss should be reduced. Should be one of 'mean' or 'sum'."}
127
+ )
128
+ ctc_zero_infinity: Optional[bool] = field(
129
+ default=False, metadata={"help": "If True, will try yo aboud the CTC loss goinf to infinity."}
130
+ )
131
+
132
+ @dataclass
133
+ class DataTrainingArguments:
134
+ """
135
+ Arguments pertaining to what data we are going to input our model for training and eval.
136
+
137
+ Using `HfArgumentParser` we can turn this class
138
+ into argparse arguments to be able to specify them on
139
+ the command line.
140
+ """
141
+
142
+ dataset_name: str = field(
143
+ metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
144
+ )
145
+ dataset_config_name: str = field(
146
+ default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
147
+ )
148
+ train_split_name: str = field(
149
+ default="train",
150
+ metadata={
151
+ "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
152
+ },
153
+ )
154
+ eval_split_name: str = field(
155
+ default="test",
156
+ metadata={
157
+ "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
158
+ },
159
+ )
160
+ audio_column_name: str = field(
161
+ default="audio",
162
+ metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
163
+ )
164
+ text_column_name: str = field(
165
+ default="text",
166
+ metadata={"help": "The name of the dataset column containing the text data. Defaults to 'text'"},
167
+ )
168
+ overwrite_cache: bool = field(
169
+ default=False, metadata={"help": "Overwrite the cached preprocessed datasets or not."}
170
+ )
171
+ preprocessing_num_workers: Optional[int] = field(
172
+ default=None,
173
+ metadata={"help": "The number of processes to use for the preprocessing."},
174
+ )
175
+ max_train_samples: Optional[int] = field(
176
+ default=None,
177
+ metadata={
178
+ "help": "For debugging purposes or quicker training, truncate the number of training examples to this "
179
+ "value if set."
180
+ },
181
+ )
182
+ max_eval_samples: Optional[int] = field(
183
+ default=None,
184
+ metadata={
185
+ "help": "For debugging purposes or quicker training, truncate the number of validation examples to this "
186
+ "value if set."
187
+ },
188
+ )
189
+ chars_to_ignore: Optional[List[str]] = list_field(
190
+ default=None,
191
+ metadata={"help": "A list of characters to remove from the transcripts."},
192
+ )
193
+ eval_metrics: List[str] = list_field(
194
+ default=["wer"],
195
+ metadata={"help": "A list of metrics the model should be evaluated on. E.g. `'wer cer'`"},
196
+ )
197
+ max_duration_in_seconds: float = field(
198
+ default=20.0,
199
+ metadata={
200
+ "help": "Filter audio files that are longer than `max_duration_in_seconds` seconds to 'max_duration_in_seconds`"
201
+ },
202
+ )
203
+ min_duration_in_seconds: float = field(
204
+ default=0.0, metadata={"help": "Filter audio files that are shorter than `min_duration_in_seconds` seconds"}
205
+ )
206
+ preprocessing_only: bool = field(
207
+ default=False,
208
+ metadata={
209
+ "help": "Whether to only do data preprocessing and skip training. "
210
+ "This is especially useful when data preprocessing errors out in distributed training due to timeout. "
211
+ "In this case, one should run the preprocessing in a non-distributed setup with `preprocessing_only=True` "
212
+ "so that the cached datasets can consequently be loaded in distributed training"
213
+ },
214
+ )
215
+ use_auth_token: bool = field(
216
+ default=False,
217
+ metadata={
218
+ "help": "If :obj:`True`, will use the token generated when running"
219
+ ":obj:`transformers-cli login` as HTTP bearer authorization for remote files."
220
+ },
221
+ )
222
+ unk_token: str = field(
223
+ default="[UNK]",
224
+ metadata={"help": "The unk token for the tokenizer"},
225
+ )
226
+ pad_token: str = field(
227
+ default="[PAD]",
228
+ metadata={"help": "The padding token for the tokenizer"},
229
+ )
230
+ word_delimiter_token: str = field(
231
+ default="|",
232
+ metadata={"help": "The word delimiter token for the tokenizer"},
233
+ )
234
+ phoneme_language: Optional[str] = field(
235
+ default=None,
236
+ metadata={
237
+ "help": "The target language that should be used be"
238
+ " passed to the tokenizer for tokenization. Note that"
239
+ " this is only relevant if the model classifies the"
240
+ " input audio to a sequence of phoneme sequences."
241
+ },
242
+ )
243
+
244
+
245
+ @dataclass
246
+ class DataCollatorCTCWithPadding:
247
+ """
248
+ Data collator that will dynamically pad the inputs received.
249
+ Args:
250
+ processor (:class:`~transformers.AutoProcessor`)
251
+ The processor used for proccessing the data.
252
+ padding (:obj:`bool`, :obj:`str` or :class:`~transformers.tokenization_utils_base.PaddingStrategy`, `optional`, defaults to :obj:`True`):
253
+ Select a strategy to pad the returned sequences (according to the model's padding side and padding index)
254
+ among:
255
+ * :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
256
+ sequence if provided).
257
+ * :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
258
+ maximum acceptable input length for the model if that argument is not provided.
259
+ * :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
260
+ different lengths).
261
+ max_length (:obj:`int`, `optional`):
262
+ Maximum length of the ``input_values`` of the returned list and optionally padding length (see above).
263
+ max_length_labels (:obj:`int`, `optional`):
264
+ Maximum length of the ``labels`` returned list and optionally padding length (see above).
265
+ pad_to_multiple_of (:obj:`int`, `optional`):
266
+ If set will pad the sequence to a multiple of the provided value.
267
+ This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >=
268
+ 7.5 (Volta).
269
+ """
270
+
271
+ processor: AutoProcessor
272
+ padding: Union[bool, str] = "longest"
273
+ pad_to_multiple_of: Optional[int] = None
274
+ pad_to_multiple_of_labels: Optional[int] = None
275
+
276
+ def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
277
+ # split inputs and labels since they have to be of different lenghts and need
278
+ # different padding methods
279
+ input_features = [{"input_values": feature["input_values"]} for feature in features]
280
+ label_features = [{"input_ids": feature["labels"]} for feature in features]
281
+
282
+ batch = self.processor.pad(
283
+ input_features,
284
+ padding=self.padding,
285
+ pad_to_multiple_of=self.pad_to_multiple_of,
286
+ return_tensors="pt",
287
+ )
288
+
289
+ with self.processor.as_target_processor():
290
+ labels_batch = self.processor.pad(
291
+ label_features,
292
+ padding=self.padding,
293
+ pad_to_multiple_of=self.pad_to_multiple_of_labels,
294
+ return_tensors="pt",
295
+ )
296
+
297
+ # replace padding with -100 to ignore loss correctly
298
+ labels = labels_batch["input_ids"].masked_fill(labels_batch.attention_mask.ne(1), -100)
299
+
300
+ batch["labels"] = labels
301
+
302
+ return batch
303
+
304
+
305
+ def create_vocabulary_from_data(
306
+ datasets: DatasetDict,
307
+ word_delimiter_token: Optional[str] = None,
308
+ unk_token: Optional[str] = None,
309
+ pad_token: Optional[str] = None,
310
+ ):
311
+ # Given training and test labels create vocabulary
312
+ def extract_all_chars(batch):
313
+ all_text = " ".join(batch["target_text"])
314
+ vocab = list(set(all_text))
315
+ return {"vocab": [vocab], "all_text": [all_text]}
316
+
317
+ vocabs = datasets.map(
318
+ extract_all_chars,
319
+ batched=True,
320
+ batch_size=-1,
321
+ keep_in_memory=True,
322
+ remove_columns=datasets["train"].column_names,
323
+ )
324
+
325
+ # take union of all unique characters in each dataset
326
+ vocab_set = functools.reduce(
327
+ lambda vocab_1, vocab_2: set(vocab_1["vocab"][0]) | set(vocab_2["vocab"][0]), vocabs.values()
328
+ )
329
+
330
+ vocab_dict = {v: k for k, v in enumerate(sorted(list(vocab_set)))}
331
+
332
+ # replace white space with delimiter token
333
+ if word_delimiter_token is not None:
334
+ vocab_dict[word_delimiter_token] = vocab_dict[" "]
335
+ del vocab_dict[" "]
336
+
337
+ # add unk and pad token
338
+ if unk_token is not None:
339
+ vocab_dict[unk_token] = len(vocab_dict)
340
+
341
+ if pad_token is not None:
342
+ vocab_dict[pad_token] = len(vocab_dict)
343
+
344
+ return vocab_dict
345
+
346
+
347
+ def main():
348
+ # See all possible arguments in src/transformers/training_args.py
349
+ # or by passing the --help flag to this script.
350
+ # We now keep distinct sets of args, for a cleaner separation of concerns.
351
+
352
+ parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
353
+ if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
354
+ # If we pass only one argument to the script and it's the path to a json file,
355
+ # let's parse it to get our arguments.
356
+ model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
357
+ else:
358
+ model_args, data_args, training_args = parser.parse_args_into_dataclasses()
359
+
360
+ # Detecting last checkpoint.
361
+ last_checkpoint = None
362
+ if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
363
+ last_checkpoint = get_last_checkpoint(training_args.output_dir)
364
+ if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
365
+ raise ValueError(
366
+ f"Output directory ({training_args.output_dir}) already exists and is not empty. "
367
+ "Use --overwrite_output_dir to overcome."
368
+ )
369
+ elif last_checkpoint is not None:
370
+ logger.info(
371
+ f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
372
+ "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
373
+ )
374
+
375
+ # Setup logging
376
+ logging.basicConfig(
377
+ format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
378
+ datefmt="%m/%d/%Y %H:%M:%S",
379
+ handlers=[logging.StreamHandler(sys.stdout)],
380
+ )
381
+ logger.setLevel(logging.INFO if is_main_process(training_args.local_rank) else logging.WARN)
382
+
383
+ # Log on each process the small summary:
384
+ logger.warning(
385
+ f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
386
+ f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
387
+ )
388
+ # Set the verbosity to info of the Transformers logger (on main process only):
389
+ if is_main_process(training_args.local_rank):
390
+ transformers.utils.logging.set_verbosity_info()
391
+ logger.info("Training/evaluation parameters %s", training_args)
392
+
393
+ # Set seed before initializing model.
394
+ set_seed(training_args.seed)
395
+
396
+ # Pre-processing dataset
397
+ import re
398
+
399
+ def map_dataset(entry):
400
+ text = entry["text"].lower()
401
+ text = text.replace("(...Vær stille under dette opptaket...)", "")
402
+ text = re.sub('[áàâ]', 'a', text)
403
+ text = re.sub('[ä]', 'æ', text)
404
+ text = re.sub('[éèëê]', 'e', text)
405
+ text = re.sub('[íìïî]', 'i', text)
406
+ text = re.sub('[óòöô]', 'o', text)
407
+ text = re.sub('[ö]', 'ø', text)
408
+ text = re.sub('[ç]', 'c', text)
409
+ text = re.sub('[úùüû]', 'u', text)
410
+ # text = re.sub('\\(?=(Punktum|Komma|Utropstegn|Spørsmålstegn))', ' ', text)
411
+ text = re.sub('\s+', ' ', text)
412
+ return {"text": text}
413
+
414
+
415
+ def filter_dataset(entry):
416
+ if not (len(entry["text"]) <= len(entry["audio"]["array"]) // 320) and (len(entry["text"].strip()) >= 3):
417
+ return False # Too short
418
+ if re.match(entry["type"], "pIW|CA"):
419
+ return False # Spelling out words
420
+ return True
421
+
422
+ # 1. First, let's load the dataset
423
+ raw_datasets = DatasetDict()
424
+
425
+ if training_args.do_train:
426
+ raw_datasets["train"] = load_dataset(
427
+ data_args.dataset_name,
428
+ data_args.dataset_config_name,
429
+ split=data_args.train_split_name,
430
+ use_auth_token=data_args.use_auth_token,
431
+ ).shuffle()
432
+ raw_datasets["train"] = raw_datasets["train"].filter(filter_dataset)
433
+ raw_datasets["train"] = raw_datasets["train"].map(map_dataset)
434
+
435
+ if data_args.audio_column_name not in raw_datasets["train"].column_names:
436
+ raise ValueError(
437
+ f"--audio_column_name '{data_args.audio_column_name}' not found in dataset '{data_args.dataset_name}'. "
438
+ "Make sure to set `--audio_column_name` to the correct audio column - one of "
439
+ f"{', '.join(raw_datasets['train'].column_names)}."
440
+ )
441
+
442
+ if data_args.text_column_name not in raw_datasets["train"].column_names:
443
+ raise ValueError(
444
+ f"--text_column_name {data_args.text_column_name} not found in dataset '{data_args.dataset_name}'. "
445
+ "Make sure to set `--text_column_name` to the correct text column - one of "
446
+ f"{', '.join(raw_datasets['train'].column_names)}."
447
+ )
448
+
449
+ if data_args.max_train_samples is not None:
450
+ raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
451
+
452
+ if training_args.do_eval:
453
+ raw_datasets["eval"] = load_dataset(
454
+ data_args.dataset_name,
455
+ data_args.dataset_config_name,
456
+ split=data_args.eval_split_name,
457
+ use_auth_token=data_args.use_auth_token,
458
+ ).shuffle()
459
+ raw_datasets["eval"] = raw_datasets["eval"].filter(filter_dataset)
460
+ raw_datasets["eval"] = raw_datasets["eval"].map(map_dataset)
461
+
462
+ if data_args.max_eval_samples is not None:
463
+ raw_datasets["eval"] = raw_datasets["eval"].select(range(data_args.max_eval_samples))
464
+
465
+
466
+ # 2. We remove some special characters from the datasets
467
+ # that make training complicated and do not help in transcribing the speech
468
+ # E.g. characters, such as `,` and `.` do not really have an acoustic characteristic
469
+ # that could be easily picked up by the model
470
+ #chars_to_ignore_regex = (
471
+ # f'[{"".join(data_args.chars_to_ignore)}]' if data_args.chars_to_ignore is not None else None
472
+ #)
473
+ chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�\'\–\_\\\+\#\/]'
474
+
475
+ text_column_name = data_args.text_column_name
476
+
477
+ def remove_special_characters(batch):
478
+ if chars_to_ignore_regex is not None:
479
+ batch["target_text"] = re.sub(chars_to_ignore_regex, "", batch[text_column_name]).lower() + " "
480
+ else:
481
+ batch["target_text"] = batch[text_column_name].lower() + " "
482
+ return batch
483
+
484
+ with training_args.main_process_first(desc="dataset map special characters removal"):
485
+ raw_datasets = raw_datasets.map(
486
+ remove_special_characters,
487
+ remove_columns=[text_column_name],
488
+ desc="remove special characters from datasets",
489
+ )
490
+
491
+ # save special tokens for tokenizer
492
+ word_delimiter_token = data_args.word_delimiter_token
493
+ unk_token = data_args.unk_token
494
+ pad_token = data_args.pad_token
495
+
496
+ # 3. Next, let's load the config as we might need it to create
497
+ # the tokenizer
498
+ # load config
499
+ config = AutoConfig.from_pretrained(
500
+ model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
501
+ )
502
+
503
+ # 4. Next, if no tokenizer file is defined,
504
+ # we create the vocabulary of the model by extracting all unique characters from
505
+ # the training and evaluation datasets
506
+ # We need to make sure that only first rank saves vocabulary
507
+ # make sure all processes wait until vocab is created
508
+ tokenizer_name_or_path = model_args.tokenizer_name_or_path
509
+ tokenizer_kwargs = {}
510
+ if tokenizer_name_or_path is None:
511
+ # save vocab in training output dir
512
+ tokenizer_name_or_path = training_args.output_dir
513
+
514
+ vocab_file = os.path.join(tokenizer_name_or_path, "vocab.json")
515
+
516
+ with training_args.main_process_first():
517
+ if training_args.overwrite_output_dir and os.path.isfile(vocab_file):
518
+ os.remove(vocab_file)
519
+
520
+ with training_args.main_process_first(desc="dataset map vocabulary creation"):
521
+ if not os.path.isfile(vocab_file):
522
+ os.makedirs(tokenizer_name_or_path, exist_ok=True)
523
+ vocab_dict = create_vocabulary_from_data(
524
+ raw_datasets,
525
+ word_delimiter_token=word_delimiter_token,
526
+ unk_token=unk_token,
527
+ pad_token=pad_token,
528
+ )
529
+
530
+ # save vocab dict to be loaded into tokenizer
531
+ with open(vocab_file, "w") as file:
532
+ json.dump(vocab_dict, file)
533
+
534
+ # if tokenizer has just been created
535
+ # it is defined by `tokenizer_class` if present in config else by `model_type`
536
+ tokenizer_kwargs = {
537
+ "config": config if config.tokenizer_class is not None else None,
538
+ "tokenizer_type": config.model_type if config.tokenizer_class is None else None,
539
+ "unk_token": unk_token,
540
+ "pad_token": pad_token,
541
+ "word_delimiter_token": word_delimiter_token,
542
+ }
543
+
544
+ # 5. Now we can instantiate the feature extractor, tokenizer and model
545
+ # Note for distributed training, the .from_pretrained methods guarantee that only
546
+ # one local process can concurrently download model & vocab.
547
+
548
+ # load feature_extractor and tokenizer
549
+ tokenizer = AutoTokenizer.from_pretrained(
550
+ tokenizer_name_or_path,
551
+ use_auth_token=data_args.use_auth_token,
552
+ **tokenizer_kwargs,
553
+ )
554
+ feature_extractor = AutoFeatureExtractor.from_pretrained(
555
+ model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
556
+ )
557
+
558
+ # adapt config
559
+ config.update(
560
+ {
561
+ "feat_proj_dropout": model_args.feat_proj_dropout,
562
+ "attention_dropout": model_args.attention_dropout,
563
+ "hidden_dropout": model_args.hidden_dropout,
564
+ "final_dropout": model_args.final_dropout,
565
+ "mask_time_prob": model_args.mask_time_prob,
566
+ "mask_time_length": model_args.mask_time_length,
567
+ "mask_feature_prob": model_args.mask_feature_prob,
568
+ "mask_feature_length": model_args.mask_feature_length,
569
+ "gradient_checkpointing": training_args.gradient_checkpointing,
570
+ "layerdrop": model_args.layerdrop,
571
+ "ctc_loss_reduction": model_args.ctc_loss_reduction,
572
+ "ctc_zero_infinity": model_args.ctc_zero_infinity,
573
+ "pad_token_id": tokenizer.pad_token_id,
574
+ "vocab_size": len(tokenizer),
575
+ "activation_dropout": model_args.activation_dropout,
576
+ }
577
+ )
578
+
579
+ # create model
580
+ model = AutoModelForCTC.from_pretrained(
581
+ model_args.model_name_or_path,
582
+ cache_dir=model_args.cache_dir,
583
+ config=config,
584
+ use_auth_token=data_args.use_auth_token,
585
+ )
586
+
587
+ # freeze encoder
588
+ if model_args.freeze_feature_encoder:
589
+ model.freeze_feature_encoder()
590
+
591
+ # 6. Now we preprocess the datasets including loading the audio, resampling and normalization
592
+ # Thankfully, `datasets` takes care of automatically loading and resampling the audio,
593
+ # so that we just need to set the correct target sampling rate and normalize the input
594
+ # via the `feature_extractor`
595
+
596
+ # make sure that dataset decodes audio with correct sampling rate
597
+ dataset_sampling_rate = next(iter(raw_datasets.values())).features[data_args.audio_column_name].sampling_rate
598
+ if dataset_sampling_rate != feature_extractor.sampling_rate:
599
+ raw_datasets = raw_datasets.cast_column(
600
+ data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
601
+ )
602
+
603
+ # derive max & min input length for sample rate & max duration
604
+ max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
605
+ min_input_length = data_args.min_duration_in_seconds * feature_extractor.sampling_rate
606
+ audio_column_name = data_args.audio_column_name
607
+ num_workers = data_args.preprocessing_num_workers
608
+
609
+ # `phoneme_language` is only relevant if the model is fine-tuned on phoneme classification
610
+ phoneme_language = data_args.phoneme_language
611
+
612
+ # Preprocessing the datasets.
613
+ # We need to read the audio files as arrays and tokenize the targets.
614
+ def prepare_dataset(batch):
615
+ # load audio
616
+ sample = batch[audio_column_name]
617
+
618
+ inputs = feature_extractor(sample["array"], sampling_rate=sample["sampling_rate"])
619
+ batch["input_values"] = inputs.input_values[0]
620
+ batch["input_length"] = len(batch["input_values"])
621
+
622
+ # encode targets
623
+ additional_kwargs = {}
624
+ if phoneme_language is not None:
625
+ additional_kwargs["phonemizer_lang"] = phoneme_language
626
+
627
+ batch["labels"] = tokenizer(batch["target_text"], **additional_kwargs).input_ids
628
+ return batch
629
+
630
+ with training_args.main_process_first(desc="dataset map preprocessing"):
631
+ vectorized_datasets = raw_datasets.map(
632
+ prepare_dataset,
633
+ remove_columns=next(iter(raw_datasets.values())).column_names,
634
+ num_proc=num_workers,
635
+ desc="preprocess datasets",
636
+ )
637
+
638
+ def is_audio_in_length_range(length):
639
+ return length > min_input_length and length < max_input_length
640
+
641
+ # filter data that is shorter than min_input_length
642
+ vectorized_datasets = vectorized_datasets.filter(
643
+ is_audio_in_length_range,
644
+ num_proc=num_workers,
645
+ input_columns=["input_length"],
646
+ )
647
+
648
+ # 7. Next, we can prepare the training.
649
+ # Let's use word error rate (WER) as our evaluation metric,
650
+ # instantiate a data collator and the trainer
651
+
652
+ # Define evaluation metrics during training, *i.e.* word error rate, character error rate
653
+ eval_metrics = {metric: load_metric(metric) for metric in data_args.eval_metrics}
654
+
655
+ # for large datasets it is advised to run the preprocessing on a
656
+ # single machine first with ``args.preprocessing_only`` since there will mostly likely
657
+ # be a timeout when running the script in distributed mode.
658
+ # In a second step ``args.preprocessing_only`` can then be set to `False` to load the
659
+ # cached dataset
660
+ if data_args.preprocessing_only:
661
+ logger.info(f"Data preprocessing finished. Files cached at {vectorized_datasets.cache_files}")
662
+ return
663
+
664
+ def compute_metrics(pred):
665
+ pred_logits = pred.predictions
666
+ pred_ids = np.argmax(pred_logits, axis=-1)
667
+
668
+ pred.label_ids[pred.label_ids == -100] = tokenizer.pad_token_id
669
+
670
+ pred_str = tokenizer.batch_decode(pred_ids)
671
+ # we do not want to group tokens when computing the metrics
672
+ label_str = tokenizer.batch_decode(pred.label_ids, group_tokens=False)
673
+
674
+ metrics = {k: v.compute(predictions=pred_str, references=label_str) for k, v in eval_metrics.items()}
675
+
676
+ return metrics
677
+
678
+ # Now save everything to be able to create a single processor later
679
+ if is_main_process(training_args.local_rank):
680
+ # save feature extractor, tokenizer and config
681
+ feature_extractor.save_pretrained(training_args.output_dir)
682
+ tokenizer.save_pretrained(training_args.output_dir)
683
+ config.save_pretrained(training_args.output_dir)
684
+
685
+ try:
686
+ processor = AutoProcessor.from_pretrained(training_args.output_dir)
687
+ except (OSError, KeyError):
688
+ warnings.warn(
689
+ "Loading a processor from a feature extractor config that does not"
690
+ " include a `processor_class` attribute is deprecated and will be removed in v5. Please add the following "
691
+ " attribute to your `preprocessor_config.json` file to suppress this warning: "
692
+ " `'processor_class': 'Wav2Vec2Processor'`",
693
+ FutureWarning,
694
+ )
695
+ processor = Wav2Vec2Processor.from_pretrained(training_args.output_dir)
696
+
697
+ # Instantiate custom data collator
698
+ data_collator = DataCollatorCTCWithPadding(processor=processor)
699
+
700
+ # Initialize Trainer
701
+ trainer = Trainer(
702
+ model=model,
703
+ data_collator=data_collator,
704
+ args=training_args,
705
+ compute_metrics=compute_metrics,
706
+ train_dataset=vectorized_datasets["train"] if training_args.do_train else None,
707
+ eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
708
+ tokenizer=feature_extractor,
709
+ )
710
+
711
+ # 8. Finally, we can start training
712
+
713
+ # Training
714
+ if training_args.do_train:
715
+
716
+ # use last checkpoint if exist
717
+ if last_checkpoint is not None:
718
+ checkpoint = last_checkpoint
719
+ elif os.path.isdir(model_args.model_name_or_path):
720
+ checkpoint = model_args.model_name_or_path
721
+ else:
722
+ checkpoint = None
723
+
724
+ train_result = trainer.train(resume_from_checkpoint=checkpoint)
725
+ trainer.save_model()
726
+
727
+ metrics = train_result.metrics
728
+ max_train_samples = (
729
+ data_args.max_train_samples
730
+ if data_args.max_train_samples is not None
731
+ else len(vectorized_datasets["train"])
732
+ )
733
+ metrics["train_samples"] = min(max_train_samples, len(vectorized_datasets["train"]))
734
+
735
+ trainer.log_metrics("train", metrics)
736
+ trainer.save_metrics("train", metrics)
737
+ trainer.save_state()
738
+
739
+ # Evaluation
740
+ results = {}
741
+ if training_args.do_eval:
742
+ logger.info("*** Evaluate ***")
743
+ metrics = trainer.evaluate()
744
+ max_eval_samples = (
745
+ data_args.max_eval_samples if data_args.max_eval_samples is not None else len(vectorized_datasets["eval"])
746
+ )
747
+ metrics["eval_samples"] = min(max_eval_samples, len(vectorized_datasets["eval"]))
748
+
749
+ trainer.log_metrics("eval", metrics)
750
+ trainer.save_metrics("eval", metrics)
751
+
752
+ # Write model card and (optionally) push to hub
753
+ config_name = data_args.dataset_config_name if data_args.dataset_config_name is not None else "na"
754
+ kwargs = {
755
+ "finetuned_from": model_args.model_name_or_path,
756
+ "tasks": "speech-recognition",
757
+ "tags": ["automatic-speech-recognition", data_args.dataset_name],
758
+ "dataset_args": f"Config: {config_name}, Training split: {data_args.train_split_name}, Eval split: {data_args.eval_split_name}",
759
+ "dataset": f"{data_args.dataset_name.upper()} - {config_name.upper()}",
760
+ }
761
+ if "common_voice" in data_args.dataset_name:
762
+ kwargs["language"] = config_name
763
+
764
+ if training_args.push_to_hub:
765
+ trainer.push_to_hub(**kwargs)
766
+ else:
767
+ trainer.create_model_card(**kwargs)
768
+
769
+ return results
770
+
771
+
772
+ if __name__ == "__main__":
773
+ main()
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer"}
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9b58429fd47b7355babb68eb4ecb995c4e4472cb1b280d0d7290761bca41cc4
3
+ size 3055
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"(": 1, ")": 2, "0": 3, "3": 4, "7": 5, "8": 6, "9": 7, "a": 8, "b": 9, "c": 10, "d": 11, "e": 12, "f": 13, "g": 14, "h": 15, "i": 16, "j": 17, "k": 18, "l": 19, "m": 20, "n": 21, "o": 22, "p": 23, "q": 24, "r": 25, "s": 26, "t": 27, "u": 28, "v": 29, "w": 30, "x": 31, "y": 32, "z": 33, "å": 34, "æ": 35, "ø": 36, "|": 0, "[UNK]": 37, "[PAD]": 38}
wandb/debug-internal.log ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220830_110431-yvlr8ud4/logs/debug-internal.log
wandb/debug.log ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220830_110431-yvlr8ud4/logs/debug.log
wandb/latest-run ADDED
@@ -0,0 +1 @@
 
 
1
+ run-20220830_110431-yvlr8ud4
wandb/run-20220830_110431-yvlr8ud4/files/config.yaml ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220830_110431-yvlr8ud4/files/output.log ADDED
@@ -0,0 +1,2653 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+
3
+
4
+
5
+
6
+
7
+
8
+
9
+
10
+
11
+
12
+
13
+
14
+
15
+
16
+
17
+
18
+
19
+
20
+
21
+
22
+
23
+
24
+
25
+
26
+
27
+
28
+
29
+
30
+
31
+
32
+
33
+
34
+
35
+
36
+
37
+
38
+
39
+
40
+
41
+
42
+
43
+
44
+
45
+
46
+
47
+
48
+
49
+
50
+
51
+
52
+
53
+
54
+
55
+
56
+
57
+
58
+
59
+
60
+
61
+
62
+
63
+
64
+
65
+
66
+
67
+
68
+
69
+
70
+
71
+
72
+
73
+
74
+
75
+
76
+
77
+
78
+
79
+
80
+
81
+
82
+
83
+
84
+
85
+
86
+ 0%| | 100/483200 [03:34<128:24:34, 1.05it/s]
87
+
88
+
89
+
90
+
91
+
92
+
93
+
94
+
95
+
96
+
97
+
98
+
99
+
100
+
101
+
102
+
103
+
104
+
105
+
106
+
107
+
108
+
109
+
110
+
111
+
112
+
113
+
114
+
115
+
116
+
117
+
118
+
119
+
120
+
121
+
122
+
123
+
124
+
125
+
126
+
127
+
128
+
129
+
130
+
131
+
132
+
133
+
134
+
135
+
136
+
137
+
138
+
139
+
140
+
141
+
142
+
143
+
144
+
145
+
146
+
147
+
148
+
149
+
150
+
151
+
152
+
153
+
154
+
155
+
156
+
157
+
158
+
159
+
160
+
161
+
162
+
163
+
164
+
165
+
166
+
167
+
168
+
169
+
170
+ 0%| | 198/483200 [07:07<141:52:25, 1.06s/it]
171
+
172
+
173
+
174
+
175
+
176
+
177
+
178
+
179
+
180
+
181
+
182
+
183
+
184
+
185
+
186
+
187
+
188
+
189
+
190
+
191
+
192
+
193
+
194
+
195
+
196
+
197
+
198
+
199
+
200
+
201
+
202
+
203
+
204
+
205
+
206
+
207
+
208
+
209
+
210
+
211
+
212
+
213
+
214
+
215
+
216
+
217
+
218
+
219
+
220
+
221
+
222
+
223
+
224
+
225
+
226
+
227
+
228
+
229
+
230
+
231
+
232
+
233
+
234
+
235
+
236
+
237
+
238
+
239
+
240
+
241
+
242
+
243
+
244
+
245
+
246
+
247
+
248
+
249
+
250
+
251
+
252
+
253
+
254
+
255
+
256
+ 0%|▏ | 299/483200 [10:44<136:25:48, 1.02s/it]
257
+
258
+
259
+
260
+
261
+
262
+
263
+
264
+
265
+
266
+
267
+
268
+
269
+
270
+
271
+
272
+
273
+
274
+
275
+
276
+
277
+
278
+
279
+
280
+
281
+
282
+
283
+
284
+
285
+
286
+
287
+
288
+
289
+
290
+
291
+
292
+
293
+
294
+
295
+
296
+
297
+
298
+
299
+
300
+
301
+
302
+
303
+
304
+
305
+
306
+
307
+
308
+
309
+
310
+
311
+
312
+
313
+
314
+
315
+
316
+
317
+
318
+
319
+
320
+
321
+
322
+
323
+
324
+
325
+
326
+
327
+
328
+
329
+
330
+
331
+
332
+
333
+
334
+
335
+
336
+
337
+
338
+
339
+
340
+
341
+
342
+ 0%|▏ | 398/483200 [14:22<142:07:15, 1.06s/it]
343
+
344
+
345
+
346
+
347
+
348
+
349
+
350
+
351
+
352
+
353
+
354
+
355
+
356
+
357
+
358
+
359
+
360
+
361
+
362
+
363
+
364
+
365
+
366
+
367
+
368
+
369
+
370
+
371
+
372
+
373
+
374
+
375
+
376
+
377
+
378
+
379
+
380
+
381
+
382
+
383
+
384
+
385
+
386
+
387
+
388
+
389
+
390
+
391
+
392
+
393
+
394
+
395
+
396
+
397
+
398
+
399
+
400
+
401
+
402
+
403
+
404
+
405
+
406
+
407
+
408
+
409
+
410
+
411
+
412
+
413
+
414
+
415
+
416
+
417
+
418
+
419
+
420
+
421
+
422
+
423
+
424
+
425
+
426
+
427
+
428
+
429
+ 0%|▏ | 499/483200 [18:00<139:01:28, 1.04s/it]
430
+ 0%|▏ | 500/483200 [18:01<131:51:34, 1.02it/s]The following columns in the evaluation set don't have a corresponding argument in `Wav2Vec2ForCTC.forward` and have been ignored: input_length. If input_length are not expected by `Wav2Vec2ForCTC.forward`, you can safely ignore this message.
431
+ ***** Running Evaluation *****
432
+ Num examples = 75595
433
+ Batch size = 12
434
+
435
+
436
+
437
+
438
+
439
+
440
+
441
+
442
+
443
+
444
+
445
+
446
+
447
+
448
+
449
+
450
+
451
+
452
+
453
+
454
+
455
+
456
+
457
+
458
+
459
+
460
+
461
+
462
+
463
+
464
+
465
+
466
+
467
+
468
+
469
+
470
+
471
+
472
+
473
+
474
+
475
+
476
+
477
+
478
+
479
+
480
+
481
+
482
+
483
+
484
+
485
+
486
+
487
+
488
+
489
+
490
+
491
+
492
+
493
+
494
+
495
+
496
+
497
+
498
+
499
+
500
+
501
+
502
+
503
+
504
+
505
+
506
+
507
+
508
+
509
+
510
+
511
+
512
+
513
+
514
+
515
+
516
+
517
+
518
+
519
+
520
+
521
+
522
+
523
+
524
+
525
+
526
+
527
+
528
+
529
+
530
+
531
+
532
+
533
+
534
+
535
+
536
+
537
+
538
+
539
+
540
+
541
+
542
+
543
+
544
+
545
+
546
+
547
+
548
+
549
+
550
+
551
+
552
+
553
+
554
+
555
+
556
+
557
+
558
+
559
+
560
+
561
+
562
+
563
+
564
+
565
+
566
+
567
+
568
+
569
+
570
+
571
+
572
+
573
+
574
+
575
+
576
+
577
+
578
+
579
+
580
+
581
+
582
+
583
+
584
+
585
+
586
+
587
+
588
+
589
+
590
+
591
+
592
+
593
+
594
+
595
+
596
+
597
+
598
+
599
+
600
+
601
+
602
+
603
+
604
+
605
+
606
+
607
+
608
+
609
+
610
+
611
+
612
+
613
+
614
+
615
+
616
+
617
+
618
+
619
+
620
+
621
+
622
+
623
+
624
+
625
+
626
+
627
+
628
+
629
+
630
+
631
+
632
+
633
+
634
+
635
+
636
+
637
+
638
+
639
+
640
+
641
+
642
+
643
+
644
+
645
+
646
+
647
+
648
+
649
+
650
+
651
+
652
+
653
+
654
+
655
+
656
+
657
+
658
+
659
+
660
+
661
+
662
+
663
+
664
+
665
+
666
+
667
+
668
+
669
+
670
+
671
+
672
+
673
+
674
+
675
+
676
+
677
+
678
+
679
+
680
+
681
+
682
+
683
+
684
+
685
+
686
+
687
+
688
+
689
+
690
+
691
+
692
+
693
+
694
+
695
+
696
+
697
+
698
+
699
+
700
+
701
+
702
+
703
+
704
+
705
+
706
+
707
+
708
+
709
+
710
+
711
+
712
+
713
+
714
+
715
+
716
+
717
+
718
+
719
+
720
+
721
+
722
+
723
+
724
+
725
+
726
+
727
+
728
+
729
+
730
+
731
+
732
+
733
+
734
+
735
+
736
+
737
+
738
+
739
+
740
+
741
+
742
+
743
+
744
+
745
+
746
+
747
+
748
+
749
+
750
+
751
+
752
+
753
+
754
+
755
+
756
+
757
+
758
+
759
+
760
+
761
+
762
+
763
+
764
+
765
+
766
+
767
+
768
+
769
+
770
+
771
+
772
+
773
+
774
+
775
+
776
+
777
+
778
+
779
+
780
+
781
+
782
+
783
+
784
+
785
+
786
+
787
+
788
+
789
+
790
+
791
+
792
+
793
+
794
+
795
+
796
+
797
+
798
+
799
+
800
+
801
+
802
+
803
+
804
+
805
+
806
+
807
+
808
+
809
+
810
+
811
+
812
+
813
+
814
+
815
+
816
+
817
+
818
+
819
+
820
+
821
+
822
+
823
+
824
+
825
+
826
+
827
+
828
+
829
+
830
+
831
+
832
+
833
+
834
+
835
+
836
+
837
+
838
+
839
+
840
+
841
+
842
+
843
+
844
+
845
+
846
+
847
+
848
+
849
+
850
+
851
+
852
+
853
+
854
+
855
+
856
+
857
+
858
+
859
+
860
+
861
+
862
+
863
+
864
+
865
+
866
+
867
+
868
+
869
+
870
+
871
+
872
+
873
+
874
+
875
+
876
+
877
+
878
+
879
+
880
+
881
+
882
+
883
+
884
+
885
+
886
+
887
+
888
+
889
+
890
+
891
+
892
+
893
+
894
+
895
+
896
+
897
+
898
+
899
+
900
+
901
+
902
+
903
+
904
+
905
+
906
+
907
+
908
+
909
+
910
+
911
+
912
+
913
+
914
+
915
+
916
+
917
+
918
+
919
+
920
+
921
+
922
+
923
+
924
+
925
+
926
+
927
+
928
+
929
+
930
+
931
+
932
+
933
+
934
+
935
+
936
+
937
+
938
+
939
+
940
+
941
+
942
+
943
+
944
+
945
+
946
+
947
+
948
+
949
+
950
+
951
+
952
+
953
+
954
+
955
+
956
+
957
+
958
+
959
+
960
+
961
+
962
+
963
+
964
+
965
+
966
+
967
+
968
+
969
+
970
+
971
+
972
+
973
+
974
+
975
+
976
+
977
+
978
+
979
+
980
+
981
+
982
+
983
+
984
+
985
+
986
+
987
+
988
+
989
+
990
+
991
+
992
+
993
+
994
+
995
+
996
+
997
+
998
+
999
+
1000
+
1001
+
1002
+
1003
+
1004
+
1005
+
1006
+
1007
+
1008
+
1009
+
1010
+
1011
+
1012
+
1013
+
1014
+
1015
+
1016
+
1017
+
1018
+
1019
+
1020
+
1021
+
1022
+
1023
+
1024
+
1025
+
1026
+
1027
+
1028
+
1029
+
1030
+
1031
+
1032
+
1033
+
1034
+
1035
+
1036
+
1037
+
1038
+
1039
+
1040
+
1041
+
1042
+
1043
+
1044
+
1045
+
1046
+
1047
+
1048
+
1049
+
1050
+
1051
+
1052
+
1053
+
1054
+
1055
+
1056
+
1057
+
1058
+
1059
+
1060
+
1061
+
1062
+
1063
+
1064
+
1065
+
1066
+
1067
+
1068
+
1069
+
1070
+
1071
+
1072
+
1073
+
1074
+
1075
+
1076
+
1077
+
1078
+
1079
+
1080
+
1081
+
1082
+
1083
+
1084
+
1085
+
1086
+
1087
+
1088
+
1089
+
1090
+
1091
+
1092
+
1093
+
1094
+
1095
+
1096
+
1097
+
1098
+
1099
+
1100
+
1101
+
1102
+
1103
+
1104
+
1105
+
1106
+
1107
+
1108
+
1109
+
1110
+
1111
+
1112
+
1113
+
1114
+
1115
+
1116
+
1117
+
1118
+
1119
+
1120
+
1121
+
1122
+
1123
+
1124
+
1125
+
1126
+
1127
+
1128
+
1129
+
1130
+
1131
+
1132
+
1133
+
1134
+
1135
+
1136
+
1137
+
1138
+
1139
+
1140
+
1141
+
1142
+
1143
+
1144
+
1145
+
1146
+
1147
+
1148
+
1149
+
1150
+
1151
+
1152
+
1153
+
1154
+
1155
+
1156
+
1157
+
1158
+
1159
+
1160
+
1161
+
1162
+
1163
+
1164
+
1165
+
1166
+
1167
+
1168
+
1169
+
1170
+
1171
+
1172
+
1173
+
1174
+
1175
+
1176
+
1177
+
1178
+
1179
+
1180
+
1181
+
1182
+
1183
+
1184
+
1185
+
1186
+
1187
+
1188
+
1189
+
1190
+
1191
+
1192
+
1193
+
1194
+
1195
+
1196
+
1197
+
1198
+
1199
+
1200
+
1201
+
1202
+
1203
+
1204
+
1205
+
1206
+
1207
+
1208
+
1209
+
1210
+
1211
+
1212
+
1213
+
1214
+
1215
+
1216
+
1217
+
1218
+
1219
+
1220
+
1221
+
1222
+
1223
+
1224
+
1225
+
1226
+
1227
+
1228
+
1229
+
1230
+
1231
+
1232
+
1233
+
1234
+
1235
+
1236
+
1237
+
1238
+
1239
+
1240
+
1241
+
1242
+
1243
+
1244
+
1245
+
1246
+
1247
+
1248
+
1249
+
1250
+
1251
+
1252
+
1253
+
1254
+
1255
+
1256
+
1257
+
1258
+
1259
+
1260
+
1261
+
1262
+
1263
+
1264
+
1265
+
1266
+
1267
+
1268
+
1269
+
1270
+
1271
+
1272
+
1273
+
1274
+
1275
+
1276
+
1277
+
1278
+
1279
+
1280
+
1281
+
1282
+
1283
+
1284
+
1285
+
1286
+
1287
+
1288
+
1289
+
1290
+
1291
+
1292
+
1293
+
1294
+
1295
+
1296
+
1297
+
1298
+
1299
+
1300
+
1301
+
1302
+
1303
+
1304
+
1305
+
1306
+
1307
+
1308
+
1309
+
1310
+
1311
+
1312
+
1313
+
1314
+
1315
+
1316
+
1317
+
1318
+
1319
+
1320
+
1321
+
1322
+
1323
+
1324
+
1325
+
1326
+
1327
+
1328
+
1329
+
1330
+
1331
+
1332
+
1333
+
1334
+
1335
+
1336
+
1337
+
1338
+
1339
+
1340
+
1341
+
1342
+
1343
+
1344
+
1345
+
1346
+
1347
+
1348
+
1349
+
1350
+
1351
+
1352
+
1353
+
1354
+
1355
+
1356
+
1357
+
1358
+
1359
+
1360
+
1361
+
1362
+
1363
+
1364
+
1365
+
1366
+
1367
+
1368
+
1369
+
1370
+
1371
+
1372
+
1373
+
1374
+
1375
+
1376
+
1377
+
1378
+
1379
+
1380
+
1381
+
1382
+
1383
+
1384
+
1385
+
1386
+
1387
+
1388
+
1389
+
1390
+
1391
+
1392
+
1393
+
1394
+
1395
+
1396
+
1397
+
1398
+
1399
+
1400
+
1401
+
1402
+
1403
+
1404
+
1405
+
1406
+
1407
+
1408
+
1409
+
1410
+
1411
+
1412
+
1413
+
1414
+
1415
+
1416
+
1417
+
1418
+
1419
+
1420
+
1421
+
1422
+
1423
+
1424
+
1425
+
1426
+
1427
+
1428
+
1429
+
1430
+
1431
+
1432
+
1433
+
1434
+
1435
+
1436
+
1437
+
1438
+
1439
+
1440
+
1441
+
1442
+
1443
+
1444
+
1445
+
1446
+
1447
+
1448
+
1449
+
1450
+
1451
+
1452
+
1453
+
1454
+
1455
+
1456
+
1457
+
1458
+
1459
+
1460
+
1461
+
1462
+
1463
+
1464
+
1465
+
1466
+
1467
+
1468
+
1469
+
1470
+
1471
+
1472
+
1473
+
1474
+
1475
+
1476
+
1477
+
1478
+
1479
+
1480
+
1481
+
1482
+
1483
+
1484
+
1485
+
1486
+
1487
+
1488
+
1489
+
1490
+
1491
+
1492
+
1493
+
1494
+
1495
+
1496
+
1497
+
1498
+
1499
+
1500
+
1501
+
1502
+
1503
+
1504
+
1505
+
1506
+
1507
+
1508
+
1509
+
1510
+
1511
+
1512
+
1513
+
1514
+
1515
+
1516
+
1517
+
1518
+
1519
+
1520
+
1521
+
1522
+
1523
+
1524
+
1525
+
1526
+
1527
+
1528
+
1529
+
1530
+
1531
+
1532
+
1533
+
1534
+
1535
+
1536
+
1537
+
1538
+
1539
+
1540
+
1541
+
1542
+
1543
+
1544
+
1545
+
1546
+
1547
+
1548
+
1549
+
1550
+
1551
+
1552
+
1553
+
1554
+
1555
+
1556
+
1557
+
1558
+
1559
+
1560
+
1561
+
1562
+
1563
+
1564
+
1565
+
1566
+
1567
+
1568
+
1569
+
1570
+
1571
+
1572
+
1573
+
1574
+
1575
+
1576
+
1577
+
1578
+
1579
+
1580
+
1581
+
1582
+
1583
+
1584
+
1585
+
1586
+
1587
+
1588
+
1589
+
1590
+
1591
+
1592
+
1593
+
1594
+
1595
+
1596
+
1597
+
1598
+
1599
+
1600
+
1601
+
1602
+
1603
+
1604
+
1605
+
1606
+
1607
+
1608
+
1609
+
1610
+
1611
+
1612
+
1613
+
1614
+
1615
+
1616
+
1617
+
1618
+
1619
+
1620
+
1621
+
1622
+
1623
+
1624
+
1625
+
1626
+
1627
+
1628
+
1629
+
1630
+
1631
+
1632
+
1633
+
1634
+
1635
+
1636
+
1637
+
1638
+
1639
+
1640
+
1641
+
1642
+
1643
+
1644
+
1645
+
1646
+
1647
+
1648
+
1649
+
1650
+
1651
+
1652
+
1653
+
1654
+
1655
+
1656
+
1657
+
1658
+
1659
+
1660
+
1661
+
1662
+
1663
+
1664
+
1665
+
1666
+
1667
+
1668
+
1669
+
1670
+
1671
+
1672
+
1673
+
1674
+
1675
+
1676
+
1677
+
1678
+
1679
+
1680
+
1681
+
1682
+
1683
+
1684
+
1685
+
1686
+
1687
+
1688
+
1689
+
1690
+
1691
+
1692
+
1693
+
1694
+
1695
+
1696
+
1697
+
1698
+
1699
+
1700
+
1701
+
1702
+
1703
+
1704
+
1705
+
1706
+
1707
+
1708
+
1709
+
1710
+
1711
+
1712
+
1713
+
1714
+
1715
+
1716
+
1717
+
1718
+
1719
+
1720
+
1721
+
1722
+
1723
+
1724
+
1725
+
1726
+
1727
+
1728
+
1729
+
1730
+
1731
+
1732
+
1733
+
1734
+
1735
+
1736
+
1737
+
1738
+
1739
+
1740
+
1741
+
1742
+
1743
+
1744
+
1745
+
1746
+
1747
+
1748
+
1749
+
1750
+
1751
+
1752
+
1753
+
1754
+
1755
+
1756
+
1757
+
1758
+
1759
+
1760
+
1761
+
1762
+
1763
+
1764
+
1765
+
1766
+
1767
+
1768
+
1769
+
1770
+
1771
+
1772
+
1773
+
1774
+
1775
+
1776
+
1777
+
1778
+
1779
+
1780
+
1781
+
1782
+
1783
+
1784
+
1785
+
1786
+
1787
+
1788
+
1789
+
1790
+
1791
+
1792
+
1793
+
1794
+
1795
+
1796
+
1797
+
1798
+
1799
+
1800
+
1801
+
1802
+
1803
+
1804
+
1805
+
1806
+
1807
+
1808
+
1809
+
1810
+
1811
+
1812
+
1813
+
1814
+
1815
+
1816
+
1817
+
1818
+
1819
+
1820
+
1821
+
1822
+
1823
+
1824
+
1825
+
1826
+
1827
+
1828
+
1829
+
1830
+
1831
+
1832
+
1833
+
1834
+
1835
+
1836
+
1837
+
1838
+
1839
+
1840
+
1841
+
1842
+
1843
+
1844
+
1845
+
1846
+
1847
+
1848
+
1849
+
1850
+
1851
+
1852
+
1853
+
1854
+
1855
+
1856
+
1857
+
1858
+
1859
+
1860
+
1861
+
1862
+
1863
+
1864
+
1865
+
1866
+
1867
+
1868
+
1869
+
1870
+
1871
+
1872
+
1873
+
1874
+
1875
+
1876
+
1877
+
1878
+
1879
+
1880
+
1881
+
1882
+
1883
+
1884
+
1885
+
1886
+
1887
+
1888
+
1889
+
1890
+
1891
+
1892
+
1893
+
1894
+
1895
+
1896
+
1897
+
1898
+
1899
+
1900
+
1901
+
1902
+
1903
+
1904
+
1905
+
1906
+
1907
+
1908
+
1909
+
1910
+
1911
+
1912
+
1913
+
1914
+
1915
+
1916
+
1917
+
1918
+
1919
+
1920
+
1921
+
1922
+
1923
+
1924
+
1925
+
1926
+
1927
+
1928
+
1929
+
1930
+
1931
+
1932
+
1933
+
1934
+
1935
+
1936
+
1937
+
1938
+
1939
+
1940
+
1941
+
1942
+
1943
+
1944
+
1945
+
1946
+
1947
+
1948
+
1949
+
1950
+
1951
+
1952
+
1953
+
1954
+
1955
+
1956
+
1957
+
1958
+
1959
+
1960
+
1961
+
1962
+
1963
+
1964
+
1965
+
1966
+
1967
+
1968
+
1969
+
1970
+
1971
+
1972
+
1973
+
1974
+
1975
+
1976
+
1977
+
1978
+
1979
+
1980
+
1981
+
1982
+
1983
+
1984
+
1985
+
1986
+
1987
+
1988
+
1989
+
1990
+
1991
+
1992
+
1993
+
1994
+
1995
+
1996
+
1997
+
1998
+
1999
+
2000
+
2001
+
2002
+
2003
+
2004
+
2005
+
2006
+
2007
+
2008
+
2009
+
2010
+
2011
+
2012
+
2013
+
2014
+
2015
+
2016
+
2017
+
2018
+
2019
+
2020
+
2021
+
2022
+
2023
+
2024
+
2025
+
2026
+
2027
+
2028
+
2029
+
2030
+
2031
+
2032
+
2033
+
2034
+
2035
+
2036
+
2037
+
2038
+
2039
+
2040
+
2041
+
2042
+
2043
+
2044
+
2045
+
2046
+
2047
+
2048
+
2049
+
2050
+
2051
+
2052
+
2053
+
2054
+
2055
+
2056
+
2057
+
2058
+
2059
+
2060
+
2061
+
2062
+
2063
+
2064
+
2065
+
2066
+
2067
+
2068
+
2069
+
2070
+
2071
+
2072
+
2073
+
2074
+
2075
+
2076
+
2077
+
2078
+
2079
+
2080
+
2081
+
2082
+
2083
+
2084
+
2085
+
2086
+
2087
+
2088
+
2089
+
2090
+
2091
+
2092
+
2093
+
2094
+
2095
+
2096
+
2097
+
2098
+
2099
+
2100
+
2101
+
2102
+
2103
+
2104
+
2105
+
2106
+
2107
+
2108
+
2109
+
2110
+
2111
+
2112
+
2113
+
2114
+
2115
+
2116
+
2117
+
2118
+
2119
+
2120
+
2121
+
2122
+
2123
+
2124
+
2125
+
2126
+
2127
+
2128
+
2129
+
2130
+
2131
+
2132
+
2133
+
2134
+
2135
+
2136
+
2137
+
2138
+
2139
+
2140
+
2141
+
2142
+
2143
+
2144
+
2145
+
2146
+
2147
+
2148
+
2149
+
2150
+
2151
+
2152
+
2153
+
2154
+
2155
+
2156
+
2157
+
2158
+
2159
+
2160
+
2161
+
2162
+
2163
+
2164
+
2165
+
2166
+
2167
+
2168
+
2169
+
2170
+
2171
+
2172
+
2173
+
2174
+
2175
+
2176
+
2177
+
2178
+
2179
+
2180
+
2181
+
2182
+
2183
+
2184
+
2185
+
2186
+
2187
+
2188
+
2189
+
2190
+
2191
+
2192
+
2193
+
2194
+
2195
+
2196
+
2197
+
2198
+
2199
+
2200
+
2201
+
2202
+
2203
+
2204
+
2205
+
2206
+
2207
+
2208
+
2209
+
2210
+
2211
+
2212
+
2213
+
2214
+
2215
+
2216
+
2217
+
2218
+
2219
+
2220
+
2221
+
2222
+
2223
+
2224
+
2225
+
2226
+
2227
+
2228
+
2229
+
2230
+
2231
+
2232
+
2233
+
2234
+
2235
+
2236
+
2237
+
2238
+
2239
+
2240
+
2241
+
2242
+
2243
+
2244
+
2245
+
2246
+
2247
+
2248
+
2249
+
2250
+
2251
+
2252
+
2253
+
2254
+
2255
+
2256
+
2257
+
2258
+
2259
+
2260
+
2261
+
2262
+
2263
+
2264
+
2265
+
2266
+
2267
+
2268
+
2269
+
2270
+
2271
+
2272
+
2273
+
2274
+
2275
+
2276
+
2277
+
2278
+
2279
+
2280
+
2281
+
2282
+
2283
+
2284
+
2285
+
2286
+
2287
+
2288
+
2289
+
2290
+
2291
+
2292
+
2293
+
2294
+
2295
+
2296
+
2297
+
2298
+
2299
+
2300
+
2301
+
2302
+
2303
+
2304
+
2305
+
2306
+
2307
+
2308
+
2309
+
2310
+
2311
+
2312
+
2313
+
2314
+
2315
+
2316
+
2317
+
2318
+
2319
+
2320
+
2321
+
2322
+
2323
+
2324
+
2325
+
2326
+
2327
+
2328
+
2329
+
2330
+
2331
+
2332
+
2333
+
2334
+
2335
+
2336
+
2337
+
2338
+
2339
+
2340
+
2341
+
2342
+
2343
+
2344
+
2345
+
2346
+
2347
+
2348
+
2349
+
2350
+
2351
+
2352
+
2353
+
2354
+
2355
+
2356
+
2357
+
2358
+
2359
+
2360
+
2361
+
2362
+
2363
+
2364
+
2365
+
2366
+
2367
+
2368
+
2369
+
2370
+
2371
+
2372
+
2373
+
2374
+
2375
+
2376
+
2377
+
2378
+
2379
+
2380
+
2381
+
2382
+
2383
+
2384
+
2385
+
2386
+
2387
+
2388
+
2389
+
2390
+
2391
+
2392
+
2393
+
2394
+
2395
+
2396
+
2397
+
2398
+
2399
+
2400
+
2401
+
2402
+
2403
+
2404
+
2405
+
2406
+
2407
+
2408
+
2409
+
2410
+
2411
+
2412
+
2413
+
2414
+
2415
+
2416
+
2417
+
2418
+
2419
+
2420
+
2421
+
2422
+
2423
+
2424
+
2425
+
2426
+
2427
+
2428
+
2429
+
2430
+
2431
+
2432
+
2433
+
2434
+
2435
+
2436
+
2437
+
2438
+
2439
+
2440
+
2441
+
2442
+
2443
+
2444
+
2445
+
2446
+
2447
+
2448
+
2449
+
2450
+
2451
+
2452
+
2453
+
2454
+
2455
+
2456
+
2457
+
2458
+
2459
+
2460
+
2461
+
2462
+
2463
+
2464
+
2465
+
2466
+
2467
+
2468
+
2469
+
2470
+
2471
+
2472
+
2473
+
2474
+
2475
+
2476
+
2477
+
2478
+
2479
+
2480
+
2481
+
2482
+
2483
+
2484
+
2485
+
2486
+
2487
+
2488
+
2489
+
2490
+
2491
+
2492
+
2493
+
2494
+
2495
+
2496
+
2497
+
2498
+
2499
+
2500
+
2501
+
2502
+
2503
+
2504
+
2505
+
2506
+
2507
+
2508
+
2509
+
2510
+
2511
+
2512
+
2513
+
2514
+
2515
+
2516
+
2517
+
2518
+
2519
+
2520
+
2521
+
2522
+
2523
+
2524
+
2525
+
2526
+
2527
+
2528
+
2529
+
2530
+
2531
+
2532
+
2533
+
2534
+
2535
+
2536
+
2537
+
2538
+
2539
+
2540
+
2541
+
2542
+
2543
+
2544
+
2545
+
2546
+
2547
+
2548
+
2549
+
2550
+
2551
+
2552
+
2553
+
2554
+
2555
+
2556
+
2557
+
2558
+
2559
+
2560
+
2561
+
2562
+
2563
+
2564
+
2565
+
2566
+
2567
+
2568
+
2569
+
2570
+
2571
+
2572
+
2573
+
2574
+
2575
+
2576
+
2577
+
2578
+
2579
+
2580
+
2581
+
2582
+
2583
+
2584
+
2585
+
2586
+
2587
+
2588
+
2589
+
2590
+
2591
+
2592
+
2593
+
2594
+
2595
+
2596
+
2597
+
2598
+
2599
+
2600
+
2601
+
2602
+
2603
+
2604
+
2605
+
2606
+
2607
+
2608
+
2609
+
2610
+
2611
+
2612
+
2613
+
2614
+
2615
+
2616
+
2617
+
2618
+
2619
+
2620
+
2621
+
2622
+
2623
+
2624
+
2625
+
2626
+
2627
+
2628
+
2629
+
2630
+
2631
+
2632
+
2633
+
2634
+
2635
+
2636
+
2637
+
2638
+
2639
+
2640
+
2641
+
2642
+
2643
+
2644
+
2645
+
2646
+
2647
+
2648
+
2649
+
2650
+ Configuration saved in ./checkpoint-500/config.json
2651
+ {'eval_loss': 3.416260242462158, 'eval_wer': 0.9938169325893345, 'eval_runtime': 4566.4565, 'eval_samples_per_second': 16.554, 'eval_steps_per_second': 1.38, 'epoch': 0.04}
2652
+ Model weights saved in ./checkpoint-500/pytorch_model.bin
2653
+ Feature extractor saved in ./checkpoint-500/preprocessor_config.json
wandb/run-20220830_110431-yvlr8ud4/files/requirements.txt ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ aiohttp==3.8.1
2
+ aiosignal==1.2.0
3
+ appdirs==1.4.4
4
+ async-timeout==4.0.2
5
+ attrs==21.4.0
6
+ audioread==2.1.9
7
+ certifi==2021.10.8
8
+ cffi==1.15.0
9
+ charset-normalizer==2.0.12
10
+ click==8.1.2
11
+ datasets==2.1.0
12
+ decorator==5.1.1
13
+ dill==0.3.4
14
+ docker-pycreds==0.4.0
15
+ filelock==3.6.0
16
+ frozenlist==1.3.0
17
+ fsspec==2022.3.0
18
+ gitdb==4.0.9
19
+ gitpython==3.1.27
20
+ huggingface-hub==0.5.1
21
+ hypothesis==6.46.5
22
+ idna==3.3
23
+ jiwer==2.3.0
24
+ joblib==1.1.0
25
+ kenlm==0.0.0
26
+ librosa==0.9.1
27
+ llvmlite==0.38.0
28
+ multidict==6.0.2
29
+ multiprocess==0.70.12.2
30
+ numba==0.55.1
31
+ numpy==1.21.6
32
+ packaging==21.3
33
+ pandas==1.4.2
34
+ pathtools==0.1.2
35
+ pillow==9.1.0
36
+ pip==20.3.4
37
+ pkg-resources==0.0.0
38
+ pooch==1.6.0
39
+ promise==2.3
40
+ protobuf==3.20.1
41
+ psutil==5.9.0
42
+ pyarrow==7.0.0
43
+ pycparser==2.21
44
+ pyctcdecode==0.3.0
45
+ pygtrie==2.4.2
46
+ pyparsing==3.0.8
47
+ python-dateutil==2.8.2
48
+ python-levenshtein==0.12.2
49
+ pytz==2022.1
50
+ pyyaml==6.0
51
+ regex==2022.4.24
52
+ requests==2.27.1
53
+ resampy==0.2.2
54
+ responses==0.18.0
55
+ sacremoses==0.0.49
56
+ scikit-learn==1.0.2
57
+ scipy==1.8.0
58
+ sentry-sdk==1.5.10
59
+ setproctitle==1.2.3
60
+ setuptools==44.1.1
61
+ shortuuid==1.0.8
62
+ six==1.16.0
63
+ smmap==5.0.0
64
+ sortedcontainers==2.4.0
65
+ soundfile==0.10.3.post1
66
+ threadpoolctl==3.1.0
67
+ tokenizers==0.12.1
68
+ torch==1.11.0+cu113
69
+ torchaudio==0.11.0+cu113
70
+ torchvision==0.12.0+cu113
71
+ tqdm==4.64.0
72
+ transformers==4.18.0
73
+ typing-extensions==4.2.0
74
+ urllib3==1.26.9
75
+ wandb==0.12.15
76
+ xxhash==3.0.0
77
+ yarl==1.7.2
wandb/run-20220830_110431-yvlr8ud4/files/wandb-metadata.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.13.0-40-generic-x86_64-with-glibc2.34",
3
+ "python": "3.9.7",
4
+ "heartbeatAt": "2022-08-30T09:04:32.564832",
5
+ "startedAt": "2022-08-30T09:04:31.465309",
6
+ "docker": null,
7
+ "cpu_count": 96,
8
+ "cuda": null,
9
+ "args": [
10
+ "--model_name_or_path=facebook/wav2vec2-xls-r-1b",
11
+ "--hub_model_id=NbAiLab/wav2vec2-1b-nst",
12
+ "--dataset_name=NbAiLab/NST",
13
+ "--dataset_config=no-close",
14
+ "--output_dir=./",
15
+ "--overwrite_output_dir",
16
+ "--num_train_epochs=40",
17
+ "--per_device_train_batch_size=12",
18
+ "--per_device_eval_batch_size=12",
19
+ "--gradient_accumulation_steps=2",
20
+ "--learning_rate=2e-5",
21
+ "--warmup_steps=2000",
22
+ "--length_column_name=input_length",
23
+ "--evaluation_strategy=steps",
24
+ "--text_column_name=text",
25
+ "--save_steps=500",
26
+ "--eval_steps=500",
27
+ "--logging_steps=100",
28
+ "--layerdrop=0.041",
29
+ "--attention_dropout=0.094",
30
+ "--activation_dropout=0.055",
31
+ "--hidden_dropout=0.047",
32
+ "--save_total_limit=3",
33
+ "--freeze_feature_encoder",
34
+ "--feat_proj_dropout=0.04",
35
+ "--mask_time_prob=0.082",
36
+ "--mask_time_length=10",
37
+ "--mask_feature_prob=0.25",
38
+ "--mask_feature_length=64",
39
+ "--gradient_checkpointing",
40
+ "--min_duration_in_seconds=0.5",
41
+ "--max_duration_in_seconds=30.0",
42
+ "--use_auth_token",
43
+ "--seed=42",
44
+ "--fp16",
45
+ "--group_by_length",
46
+ "--do_train",
47
+ "--do_eval",
48
+ "--push_to_hub",
49
+ "--preprocessing_num_workers=32",
50
+ "--ctc_zero_infinity"
51
+ ],
52
+ "state": "running",
53
+ "program": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst/run_speech_recognition_ctc.py",
54
+ "codePath": "run_speech_recognition_ctc.py",
55
+ "git": {
56
+ "remote": "https://huggingface.co/NbAiLab/wav2vec2-1b-nst",
57
+ "commit": "aa7bcfb7473f662ac8f42c246a56419e5900d9c6"
58
+ },
59
+ "email": "[email protected]",
60
+ "root": "/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst",
61
+ "host": "dante",
62
+ "username": "rolvb",
63
+ "executable": "/mnt/lv_ai_1_dante/ml/rolvb/venv/bin/python"
64
+ }
wandb/run-20220830_110431-yvlr8ud4/files/wandb-summary.json ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220830_110431-yvlr8ud4/logs/debug-internal.log ADDED
The diff for this file is too large to render. See raw diff
 
wandb/run-20220830_110431-yvlr8ud4/logs/debug.log ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_setup.py:_flush():75] Loading settings from /home/rolvb/.config/wandb/settings
2
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_setup.py:_flush():75] Loading settings from /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst/wandb/settings
3
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_setup.py:_flush():75] Loading settings from environment variables: {'project': 'wav2vec2', 'entity': 'NbAiLab'}
4
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_setup.py:_flush():75] Inferring run settings from compute environment: {'program_relpath': 'run_speech_recognition_ctc.py', 'program': '/mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst/run_speech_recognition_ctc.py'}
5
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_init.py:_log_setup():437] Logging user logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst/wandb/run-20220830_110431-yvlr8ud4/logs/debug.log
6
+ 2022-08-30 11:04:31,468 INFO MainThread:3759610 [wandb_init.py:_log_setup():438] Logging internal logs to /mnt/lv_ai_1_dante/ml/models/wav2vec2-1b-nst/wandb/run-20220830_110431-yvlr8ud4/logs/debug-internal.log
7
+ 2022-08-30 11:04:31,469 INFO MainThread:3759610 [wandb_init.py:init():471] calling init triggers
8
+ 2022-08-30 11:04:31,469 INFO MainThread:3759610 [wandb_init.py:init():474] wandb.init called with sweep_config: {}
9
+ config: {}
10
+ 2022-08-30 11:04:31,469 INFO MainThread:3759610 [wandb_init.py:init():524] starting backend
11
+ 2022-08-30 11:04:31,469 INFO MainThread:3759610 [backend.py:_multiprocessing_setup():97] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
12
+ 2022-08-30 11:04:31,559 INFO MainThread:3759610 [backend.py:ensure_launched():217] starting backend process...
13
+ 2022-08-30 11:04:31,631 INFO MainThread:3759610 [backend.py:ensure_launched():222] started backend process with pid: 3760704
14
+ 2022-08-30 11:04:31,633 INFO MainThread:3759610 [wandb_init.py:init():533] backend started and connected
15
+ 2022-08-30 11:04:31,642 INFO MainThread:3759610 [wandb_init.py:init():597] updated telemetry
16
+ 2022-08-30 11:04:31,820 INFO MainThread:3759610 [wandb_init.py:init():628] communicating run to backend with 30 second timeout
17
+ 2022-08-30 11:04:32,385 INFO MainThread:3759610 [wandb_run.py:_on_init():1923] communicating current version
18
+ 2022-08-30 11:04:32,542 INFO MainThread:3759610 [wandb_run.py:_on_init():1927] got version response upgrade_message: "wandb version 0.13.2 is available! To upgrade, please run:\n $ pip install wandb --upgrade"
19
+
20
+ 2022-08-30 11:04:32,542 INFO MainThread:3759610 [wandb_init.py:init():659] starting run threads in backend
21
+ 2022-08-30 11:04:32,598 INFO MainThread:3759610 [wandb_run.py:_console_start():1897] atexit reg
22
+ 2022-08-30 11:04:32,599 INFO MainThread:3759610 [wandb_run.py:_redirect():1770] redirect: SettingsConsole.REDIRECT
23
+ 2022-08-30 11:04:32,600 INFO MainThread:3759610 [wandb_run.py:_redirect():1775] Redirecting console.
24
+ 2022-08-30 11:04:32,602 INFO MainThread:3759610 [wandb_run.py:_redirect():1831] Redirects installed.
25
+ 2022-08-30 11:04:32,602 INFO MainThread:3759610 [wandb_init.py:init():684] run started, returning control to user process
26
+ 2022-08-30 11:04:32,630 INFO MainThread:3759610 [wandb_run.py:_config_callback():1131] config_cb None None {'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'float32', 'use_bfloat16': False, 'pruned_heads': {}, 'tie_word_embeddings': True, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'chunk_size_feed_forward': 0, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'architectures': ['Wav2Vec2ForPreTraining'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': 38, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'facebook/wav2vec2-xls-r-1b', 'transformers_version': '4.18.0', 'feat_extract_dropout': 0.0, 'model_type': 'wav2vec2', 'num_feat_extract_layers': 7, 'hidden_size': 1280, 'feat_extract_norm': 'layer', 'feat_extract_activation': 'gelu', 'conv_dim': [512, 512, 512, 512, 512, 512, 512], 'conv_stride': [5, 2, 2, 2, 2, 2, 2], 'conv_kernel': [10, 3, 3, 3, 3, 2, 2], 'conv_bias': True, 'num_conv_pos_embeddings': 128, 'num_conv_pos_embedding_groups': 16, 'num_hidden_layers': 48, 'intermediate_size': 5120, 'hidden_act': 'gelu', 'num_attention_heads': 16, 'hidden_dropout': 0.047, 'attention_dropout': 0.094, 'activation_dropout': 0.055, 'feat_proj_dropout': 0.04, 'final_dropout': 0.0, 'layerdrop': 0.041, 'layer_norm_eps': 1e-05, 'initializer_range': 0.02, 'vocab_size': 41, 'do_stable_layer_norm': True, 'use_weighted_layer_sum': False, 'apply_spec_augment': True, 'mask_time_prob': 0.082, 'mask_time_length': 10, 'mask_time_min_masks': 2, 'mask_feature_prob': 0.25, 'mask_feature_length': 64, 'mask_feature_min_masks': 0, 'num_codevectors_per_group': 320, 'num_codevector_groups': 2, 'contrastive_logits_temperature': 0.1, 'feat_quantizer_dropout': 0.0, 'num_negatives': 100, 'codevector_dim': 1024, 'proj_codevector_dim': 1024, 'diversity_loss_weight': 0.1, 'ctc_loss_reduction': 'mean', 'ctc_zero_infinity': True, 'add_adapter': False, 'adapter_kernel_size': 3, 'adapter_stride': 2, 'num_adapter_layers': 3, 'output_hidden_size': 1280, 'classifier_proj_size': 256, 'tdnn_dim': [512, 512, 512, 512, 1500], 'tdnn_kernel': [5, 3, 3, 1, 1], 'tdnn_dilation': [1, 2, 3, 1, 1], 'xvector_output_dim': 512, 'output_dir': './', 'overwrite_output_dir': True, 'do_train': True, 'do_eval': True, 'do_predict': False, 'evaluation_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 12, 'per_device_eval_batch_size': 12, 'per_gpu_train_batch_size': 'None', 'per_gpu_eval_batch_size': 'None', 'gradient_accumulation_steps': 2, 'eval_accumulation_steps': 'None', 'eval_delay': 0, 'learning_rate': 2e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 40.0, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'warmup_ratio': 0.0, 'warmup_steps': 2000, 'log_level': -1, 'log_level_replica': -1, 'log_on_each_node': True, 'logging_dir': './runs/Aug30_11-03-31_dante', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 100, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 500, 'save_total_limit': 3, 'save_on_each_node': False, 'no_cuda': False, 'seed': 42, 'data_seed': 'None', 'bf16': False, 'fp16': True, 'fp16_opt_level': 'O1', 'half_precision_backend': 'amp', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': 'None', 'local_rank': -1, 'xpu_backend': 'None', 'tpu_num_cores': 'None', 'tpu_metrics_debug': False, 'debug': '[]', 'dataloader_drop_last': False, 'eval_steps': 500, 'dataloader_num_workers': 0, 'past_index': -1, 'run_name': './', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': 'None', 'load_best_model_at_end': False, 'metric_for_best_model': 'None', 'greater_is_better': 'None', 'ignore_data_skip': False, 'sharded_ddp': '[]', 'deepspeed': 'None', 'label_smoothing_factor': 0.0, 'optim': 'adamw_hf', 'adafactor': False, 'group_by_length': True, 'length_column_name': 'input_length', 'report_to': "['wandb']", 'ddp_find_unused_parameters': 'None', 'ddp_bucket_cap_mb': 'None', 'dataloader_pin_memory': True, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': True, 'resume_from_checkpoint': 'None', 'hub_model_id': 'NbAiLab/wav2vec2-1b-nst', 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'gradient_checkpointing': True, 'fp16_backend': 'auto', 'push_to_hub_model_id': 'None', 'push_to_hub_organization': 'None', 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', '_n_gpu': 1, 'mp_parameters': '', 'train_batch_size': 12, 'eval_batch_size': 12}
27
+ 2022-08-30 11:04:32,633 INFO MainThread:3759610 [wandb_watch.py:watch():47] Watching
wandb/run-20220830_110431-yvlr8ud4/run-yvlr8ud4.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:259022a5c62fb30a5d197a39fe99951a7c9f97219d04c793dce9fcceb827a9c8
3
+ size 8649498