File size: 17,575 Bytes
662b801
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
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
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
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
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
#!/usr/bin/env python
# coding=utf-8
# Copyright 2021 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import logging
import os
import sys
import warnings
from dataclasses import dataclass, field
from random import randint
from typing import Optional

import datasets
import evaluate
import numpy as np
from datasets import DatasetDict, load_dataset

import transformers
from transformers import (
    AutoConfig,
    AutoFeatureExtractor,
    AutoModelForAudioClassification,
    HfArgumentParser,
    Trainer,
    TrainingArguments,
    set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version, send_example_telemetry
from transformers.utils.versions import require_version


logger = logging.getLogger(__name__)

# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version("4.27.0.dev0")

require_version("datasets>=1.14.0", "To fix: pip install -r examples/pytorch/audio-classification/requirements.txt")


def random_subsample(wav: np.ndarray, max_length: float, sample_rate: int = 16000):
    """Randomly sample chunks of `max_length` seconds from the input audio"""
    sample_length = int(round(sample_rate * max_length))
    if len(wav) <= sample_length:
        return wav
    random_offset = randint(0, len(wav) - sample_length - 1)
    return wav[random_offset : random_offset + sample_length]


@dataclass
class DataTrainingArguments:
    """
    Arguments pertaining to what data we are going to input our model for training and eval.
    Using `HfArgumentParser` we can turn this class
    into argparse arguments to be able to specify them on
    the command line.
    """

    dataset_name: Optional[str] = field(default=None, metadata={"help": "Name of a dataset from the datasets package"})
    dataset_config_name: Optional[str] = field(
        default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
    )
    train_file: Optional[str] = field(
        default=None, metadata={"help": "A file containing the training audio paths and labels."}
    )
    eval_file: Optional[str] = field(
        default=None, metadata={"help": "A file containing the validation audio paths and labels."}
    )
    train_split_name: str = field(
        default="train",
        metadata={
            "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
        },
    )
    eval_split_name: str = field(
        default="validation",
        metadata={
            "help": (
                "The name of the training data set split to use (via the datasets library). Defaults to 'validation'"
            )
        },
    )
    audio_column_name: str = field(
        default="audio",
        metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
    )
    label_column_name: str = field(
        default="label", metadata={"help": "The name of the dataset column containing the labels. Defaults to 'label'"}
    )
    max_train_samples: Optional[int] = field(
        default=None,
        metadata={
            "help": (
                "For debugging purposes or quicker training, truncate the number of training examples to this "
                "value if set."
            )
        },
    )
    max_eval_samples: Optional[int] = field(
        default=None,
        metadata={
            "help": (
                "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
                "value if set."
            )
        },
    )
    max_length_seconds: float = field(
        default=20,
        metadata={"help": "Audio clips will be randomly cut to this length during training if the value is set."},
    )


@dataclass
class ModelArguments:
    """
    Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
    """

    model_name_or_path: str = field(
        default="facebook/wav2vec2-base",
        metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"},
    )
    config_name: Optional[str] = field(
        default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
    )
    cache_dir: Optional[str] = field(
        default=None, metadata={"help": "Where do you want to store the pretrained models downloaded from the Hub"}
    )
    model_revision: str = field(
        default="main",
        metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
    )
    feature_extractor_name: Optional[str] = field(
        default=None, metadata={"help": "Name or path of preprocessor config."}
    )
    freeze_feature_encoder: bool = field(
        default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
    )
    attention_mask: bool = field(
        default=True, metadata={"help": "Whether to generate an attention mask in the feature extractor."}
    )
    use_auth_token: bool = field(
        default=False,
        metadata={
            "help": (
                "Will use the token generated when running `huggingface-cli login` (necessary to use this script "
                "with private models)."
            )
        },
    )
    freeze_feature_extractor: Optional[bool] = field(
        default=None, metadata={"help": "Whether to freeze the feature extractor layers of the model."}
    )
    ignore_mismatched_sizes: bool = field(
        default=False,
        metadata={"help": "Will enable to load a pretrained model whose head dimensions are different."},
    )
    dropout: Optional[float] = field(
        default=0., metadata={"help": "Encoder dropout to apply."}
    )

    def __post_init__(self):
        if not self.freeze_feature_extractor and self.freeze_feature_encoder:
            warnings.warn(
                "The argument `--freeze_feature_extractor` is deprecated and "
                "will be removed in a future version. Use `--freeze_feature_encoder`"
                "instead. Setting `freeze_feature_encoder==True`.",
                FutureWarning,
            )
        if self.freeze_feature_extractor and not self.freeze_feature_encoder:
            raise ValueError(
                "The argument `--freeze_feature_extractor` is deprecated and "
                "should not be used in combination with `--freeze_feature_encoder`."
                "Only make use of `--freeze_feature_encoder`."
            )


def main():
    # See all possible arguments in src/transformers/training_args.py
    # or by passing the --help flag to this script.
    # We now keep distinct sets of args, for a cleaner separation of concerns.

    parser = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments))
    if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
        # If we pass only one argument to the script and it's the path to a json file,
        # let's parse it to get our arguments.
        model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
    else:
        model_args, data_args, training_args = parser.parse_args_into_dataclasses()

    # Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
    # information sent is the one passed as arguments along with your Python/PyTorch versions.
    send_example_telemetry("run_audio_classification", model_args, data_args)

    # Setup logging
    logging.basicConfig(
        format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
        datefmt="%m/%d/%Y %H:%M:%S",
        handlers=[logging.StreamHandler(sys.stdout)],
    )

    if training_args.should_log:
        # The default of training_args.log_level is passive, so we set log level at info here to have that default.
        transformers.utils.logging.set_verbosity_info()

    log_level = training_args.get_process_log_level()
    logger.setLevel(log_level)
    transformers.utils.logging.set_verbosity(log_level)
    transformers.utils.logging.enable_default_handler()
    transformers.utils.logging.enable_explicit_format()

    # Log on each process the small summary:
    logger.warning(
        f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu} "
        + f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
    )
    logger.info(f"Training/evaluation parameters {training_args}")

    # Set seed before initializing model.
    set_seed(training_args.seed)

    # Detecting last checkpoint.
    last_checkpoint = None
    if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
        last_checkpoint = get_last_checkpoint(training_args.output_dir)
        if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
            raise ValueError(
                f"Output directory ({training_args.output_dir}) already exists and is not empty. "
                "Use --overwrite_output_dir to train from scratch."
            )
        elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
            logger.info(
                f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
                "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
            )

    # Initialize our dataset and prepare it for the audio classification task.
    raw_datasets = DatasetDict()
    raw_datasets["train"] = load_dataset(
        data_args.dataset_name,
        data_args.dataset_config_name,
        split=data_args.train_split_name,
        use_auth_token=True if model_args.use_auth_token else None,
    )
    raw_datasets["eval"] = load_dataset(
        data_args.dataset_name,
        data_args.dataset_config_name,
        split=data_args.eval_split_name,
        use_auth_token=True if model_args.use_auth_token else None,
    )

    if data_args.audio_column_name not in raw_datasets["train"].column_names:
        raise ValueError(
            f"--audio_column_name {data_args.audio_column_name} not found in dataset '{data_args.dataset_name}'. "
            "Make sure to set `--audio_column_name` to the correct audio column - one of "
            f"{', '.join(raw_datasets['train'].column_names)}."
        )

    if data_args.label_column_name not in raw_datasets["train"].column_names:
        raise ValueError(
            f"--label_column_name {data_args.label_column_name} not found in dataset '{data_args.dataset_name}'. "
            "Make sure to set `--label_column_name` to the correct text column - one of "
            f"{', '.join(raw_datasets['train'].column_names)}."
        )

    # Setting `return_attention_mask=True` is the way to get a correctly masked mean-pooling over
    # transformer outputs in the classifier, but it doesn't always lead to better accuracy
    feature_extractor = AutoFeatureExtractor.from_pretrained(
        model_args.feature_extractor_name or model_args.model_name_or_path,
        return_attention_mask=model_args.attention_mask,
        cache_dir=model_args.cache_dir,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
    )

    # `datasets` takes care of automatically loading and resampling the audio,
    # so we just need to set the correct target sampling rate.
    raw_datasets = raw_datasets.cast_column(
        data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
    )

    model_input_name = feature_extractor.model_input_names[0]

    def train_transforms(batch):
        """Apply train_transforms across a batch."""
        subsampled_wavs = []
        for audio in batch[data_args.audio_column_name]:
            wav = random_subsample(
                audio["array"], max_length=data_args.max_length_seconds, sample_rate=feature_extractor.sampling_rate
            )
            subsampled_wavs.append(wav)
        inputs = feature_extractor(subsampled_wavs, sampling_rate=feature_extractor.sampling_rate)
        output_batch = {model_input_name: inputs.get(model_input_name)}
        output_batch["labels"] = list(batch[data_args.label_column_name])

        return output_batch

    def val_transforms(batch):
        """Apply val_transforms across a batch."""
        wavs = [audio["array"] for audio in batch[data_args.audio_column_name]]
        inputs = feature_extractor(wavs, sampling_rate=feature_extractor.sampling_rate)
        output_batch = {model_input_name: inputs.get(model_input_name)}
        output_batch["labels"] = list(batch[data_args.label_column_name])

        return output_batch

    # Prepare label mappings.
    # We'll include these in the model's config to get human readable labels in the Inference API.
    labels = raw_datasets["train"].features[data_args.label_column_name].names
    label2id, id2label = {}, {}
    for i, label in enumerate(labels):
        label2id[label] = str(i)
        id2label[str(i)] = label

    # Load the accuracy metric from the datasets package
    metric = evaluate.load("accuracy")

    # Define our compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with
    # `predictions` and `label_ids` fields) and has to return a dictionary string to float.
    def compute_metrics(eval_pred):
        """Computes accuracy on a batch of predictions"""
        predictions = np.argmax(eval_pred.predictions, axis=1)
        return metric.compute(predictions=predictions, references=eval_pred.label_ids)

    config = AutoConfig.from_pretrained(
        model_args.config_name or model_args.model_name_or_path,
        num_labels=len(labels),
        label2id=label2id,
        id2label=id2label,
        finetuning_task="audio-classification",
        cache_dir=model_args.cache_dir,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
    )
    config.activation_dropout = config.attention_dropout = config.dropout = model_args.dropout
    model = AutoModelForAudioClassification.from_pretrained(
        model_args.model_name_or_path,
        from_tf=bool(".ckpt" in model_args.model_name_or_path),
        config=config,
        cache_dir=model_args.cache_dir,
        revision=model_args.model_revision,
        use_auth_token=True if model_args.use_auth_token else None,
        ignore_mismatched_sizes=model_args.ignore_mismatched_sizes,
    )

    # freeze the convolutional waveform encoder
    if model_args.freeze_feature_encoder:
        model.freeze_feature_encoder()

    if training_args.do_train:
        if data_args.max_train_samples is not None:
            raw_datasets["train"] = (
                raw_datasets["train"].shuffle(seed=training_args.seed).select(range(data_args.max_train_samples))
            )
        # Set the training transforms
        raw_datasets["train"].set_transform(train_transforms, output_all_columns=False)

    if training_args.do_eval:
        if data_args.max_eval_samples is not None:
            raw_datasets["eval"] = (
                raw_datasets["eval"].shuffle(seed=training_args.seed).select(range(data_args.max_eval_samples))
            )
        # Set the validation transforms
        raw_datasets["eval"].set_transform(val_transforms, output_all_columns=False)

    # Initialize our trainer
    trainer = Trainer(
        model=model,
        args=training_args,
        train_dataset=raw_datasets["train"] if training_args.do_train else None,
        eval_dataset=raw_datasets["eval"] if training_args.do_eval else None,
        compute_metrics=compute_metrics,
        tokenizer=feature_extractor,
    )

    # Training
    if training_args.do_train:
        checkpoint = None
        if training_args.resume_from_checkpoint is not None:
            checkpoint = training_args.resume_from_checkpoint
        elif last_checkpoint is not None:
            checkpoint = last_checkpoint
        train_result = trainer.train(resume_from_checkpoint=checkpoint)
        trainer.save_model()
        trainer.log_metrics("train", train_result.metrics)
        trainer.save_metrics("train", train_result.metrics)
        trainer.save_state()

    # Evaluation
    if training_args.do_eval:
        metrics = trainer.evaluate()
        trainer.log_metrics("eval", metrics)
        trainer.save_metrics("eval", metrics)

    # Write model card and (optionally) push to hub
    kwargs = {
        "finetuned_from": model_args.model_name_or_path,
        "tasks": "audio-classification",
        "dataset": data_args.dataset_name,
        "tags": ["audio-classification"],
    }
    if training_args.push_to_hub:
        trainer.push_to_hub(**kwargs)
    else:
        trainer.create_model_card(**kwargs)


if __name__ == "__main__":
    main()