Add ASR CTC streaming example (#15309)

* Single-epoch run

* Apply suggestions from code review
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Infinite dataset

* Trainer fix + distributed benchmark

* Benchmark fix

* unused import

* interleaved splits

* interleaved splits

* has_length util

* Move to research projects

* Leftover Sized checks

* Bump min version

* Unused import

* Revert trainer changes
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

examples/pytorch/speech-recognition/README.md

@@ -127,6 +127,62 @@ python -m torch.distributed.launch \
 On 8 V100 GPUs, this script should run in *ca.* 18 minutes and yield a CTC loss of **0.39** and word error rate
 of **0.36**.
 
+
+### Multi GPU CTC with Dataset Streaming
+
+The following command shows how to use [Dataset Streaming mode](https://huggingface.co/docs/datasets/dataset_streaming.html)
+to fine-tune [XLS-R](https://huggingface.co/transformers/master/model_doc/xls_r.html) 
+on [Common Voice](https://huggingface.co/datasets/common_voice) using 4 GPUs in half-precision.
+
+Streaming mode imposes several constraints on training:
+1. We need to construct a tokenizer beforehand and define it via `--tokenizer_name_or_path`.
+2. `--num_train_epochs` has to be replaced by `--max_steps`. Similarly, all other epoch-based arguments have to be 
+replaced by step-based ones.
+3. Full dataset shuffling on each epoch is not possible, since we don't have the whole dataset available at once. 
+However, the `--shuffle_buffer_size` argument controls how many examples we can pre-download before shuffling them.
+
+
+```bash
+**python -m torch.distributed.launch \
+	--nproc_per_node 4 run_speech_recognition_ctc_streaming.py \
+	--dataset_name="common_voice" \
+	--model_name_or_path="facebook/wav2vec2-xls-r-300m" \
+	--tokenizer_name_or_path="anton-l/wav2vec2-tokenizer-turkish" \
+	--dataset_config_name="tr" \
+	--train_split_name="train+validation" \
+	--eval_split_name="test" \
+	--output_dir="wav2vec2-xls-r-common_voice-tr-ft" \
+	--overwrite_output_dir \
+	--max_steps="5000" \
+	--per_device_train_batch_size="8" \
+	--gradient_accumulation_steps="2" \
+	--learning_rate="5e-4" \
+	--warmup_steps="500" \
+	--evaluation_strategy="steps" \
+	--text_column_name="sentence" \
+	--save_steps="500" \
+	--eval_steps="500" \
+	--logging_steps="1" \
+	--layerdrop="0.0" \
+	--eval_metrics wer cer \
+	--save_total_limit="1" \
+	--mask_time_prob="0.3" \
+	--mask_time_length="10" \
+	--mask_feature_prob="0.1" \
+	--mask_feature_length="64" \
+	--freeze_feature_encoder \
+	--chars_to_ignore , ? . ! - \; \: \" “ % ‘ ” � \
+	--max_duration_in_seconds="20" \
+	--shuffle_buffer_size="500" \
+	--fp16 \
+	--push_to_hub \
+	--do_train --do_eval \
+	--gradient_checkpointing**
+```
+
+On 4 V100 GPUs, this script should run in *ca.* 3h 31min and yield a CTC loss of **0.35** and word error rate
+of **0.29**.
+
 ### Examples CTC
 
 The following tables present a couple of example runs on the most popular speech-recognition datasets. 
@@ -175,6 +231,7 @@ they can serve as a baseline to improve upon.
 | [Common Voice](https://huggingface.co/datasets/common_voice)| `"tr"`  | [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) | 0.35 | - | 1 GPU V100   |  1h20min                      | [here](https://huggingface.co/patrickvonplaten/wav2vec2-common_voice-tr-demo)  | [run.sh](https://huggingface.co/patrickvonplaten/wav2vec2-common_voice-tr-demo/blob/main/run.sh) |
 | [Common Voice](https://huggingface.co/datasets/common_voice)| `"tr"`  | [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m)  | 0.31     | - | 8 GPU V100   |  1h05            | [here](https://huggingface.co/patrickvonplaten/wav2vec2-large-xls-r-300m-common_voice-tr-ft)      |  [run.sh](https://huggingface.co/patrickvonplaten/wav2vec2-large-xls-r-300m-common_voice-tr-ft/blob/main/run.sh) |
 | [Common Voice](https://huggingface.co/datasets/common_voice)| `"tr"`  | [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b)  | 0.21 | -  | 2 GPU Titan 24 GB RAM   |  15h10            | [here](https://huggingface.co/patrickvonplaten/wav2vec2-xls-r-1b-common_voice-tr-ft)      |  [run.sh](https://huggingface.co/patrickvonplaten/wav2vec2-large-xls-r-1b-common_voice-tr-ft/blob/main/run.sh) |
+| [Common Voice](https://huggingface.co/datasets/common_voice)| `"tr"` in streaming mode  | [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m)  | 0.29     | - | 4 GPU V100   |  3h31            | [here](https://huggingface.co/anton-l/wav2vec2-xls-r-common_voice-tr-ft-stream)      |  [run.sh](https://huggingface.co/anton-l/wav2vec2-xls-r-common_voice-tr-ft-stream/blob/main/run.sh) |
 
 
 #### Multilingual Librispeech CTC

examples/pytorch/speech-recognition/requirements.txt

-datasets >= 1.13.3
+datasets >= 1.18.0
 torch >= 1.5
 torchaudio
 librosa

examples/pytorch/speech-recognition/run_speech_recognition_ctc.py

@@ -51,7 +51,7 @@ from transformers.utils.versions import require_version
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
 check_min_version("4.17.0.dev0")
 
-require_version("datasets>=1.13.3", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
+require_version("datasets>=1.18.0", "To fix: pip install -r examples/pytorch/speech-recognition/requirements.txt")
 
 
 logger = logging.getLogger(__name__)
@@ -146,7 +146,8 @@ class DataTrainingArguments:
     train_split_name: str = field(
         default="train+validation",
         metadata={
-            "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train+validation'"
+            "help": "The name of the training data set split to use (via the datasets library). Defaults to "
+            "'train+validation'"
         },
     )
     eval_split_name: str = field(

examples/pytorch/speech-recognition/run_speech_recognition_seq2seq.py

@@ -49,7 +49,7 @@ from transformers.utils.versions import require_version
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
 check_min_version("4.17.0.dev0")
 
-require_version("datasets>=1.8.0", "To fix: pip install -r examples/pytorch/summarization/requirements.txt")
+require_version("datasets>=1.18.0", "To fix: pip install -r examples/pytorch/speech-recognition/requirements.txt")
 
 logger = logging.getLogger(__name__)
 

examples/research_projects/robust-speech-event/run_speech_recognition_ctc_streaming.py

+#!/usr/bin/env python
+# coding=utf-8
+# Copyright 2022 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
+
+""" Fine-tuning a 🤗 Transformers CTC model for automatic speech recognition in streaming mode"""
+
+import logging
+import os
+import re
+import sys
+import warnings
+from dataclasses import dataclass, field
+from typing import Dict, List, Optional, Union
+
+import datasets
+import numpy as np
+import torch
+from datasets import IterableDatasetDict, interleave_datasets, load_dataset, load_metric
+from torch.utils.data import IterableDataset
+
+import transformers
+from transformers import (
+    AutoConfig,
+    AutoFeatureExtractor,
+    AutoModelForCTC,
+    AutoProcessor,
+    AutoTokenizer,
+    HfArgumentParser,
+    Trainer,
+    TrainerCallback,
+    TrainingArguments,
+    Wav2Vec2Processor,
+    set_seed,
+)
+from transformers.trainer_pt_utils import IterableDatasetShard
+from transformers.trainer_utils import get_last_checkpoint, is_main_process
+from transformers.utils import check_min_version
+from transformers.utils.versions import require_version
+
+
+# Will error if the minimal version of Transformers is not installed. Remove at your own risk.
+check_min_version("4.17.0.dev0")
+
+require_version("datasets>=1.18.2", "To fix: pip install 'datasets>=1.18.2'")
+
+
+logger = logging.getLogger(__name__)
+
+
+def list_field(default=None, metadata=None):
+    return field(default_factory=lambda: default, metadata=metadata)
+
+
+@dataclass
+class ModelArguments:
+    """
+    Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
+    """
+
+    model_name_or_path: str = field(
+        metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
+    )
+    tokenizer_name_or_path: Optional[str] = field(
+        default=None,
+        metadata={"help": "Path to pretrained tokenizer or tokenizer identifier from huggingface.co/models"},
+    )
+    cache_dir: Optional[str] = field(
+        default=None,
+        metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
+    )
+    freeze_feature_encoder: bool = field(
+        default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
+    )
+    attention_dropout: float = field(
+        default=0.0, metadata={"help": "The dropout ratio for the attention probabilities."}
+    )
+    activation_dropout: float = field(
+        default=0.0, metadata={"help": "The dropout ratio for activations inside the fully connected layer."}
+    )
+    feat_proj_dropout: float = field(default=0.0, metadata={"help": "The dropout ratio for the projected features."})
+    hidden_dropout: float = field(
+        default=0.0,
+        metadata={
+            "help": "The dropout probability for all fully connected layers in the embeddings, encoder, and pooler."
+        },
+    )
+    final_dropout: float = field(
+        default=0.0,
+        metadata={"help": "The dropout probability for the final projection layer."},
+    )
+    mask_time_prob: float = field(
+        default=0.05,
+        metadata={
+            "help": "Probability of each feature vector along the time axis to be chosen as the start of the vector"
+            "span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"
+            "vectors will be masked along the time axis."
+        },
+    )
+    mask_time_length: int = field(
+        default=10,
+        metadata={"help": "Length of vector span to mask along the time axis."},
+    )
+    mask_feature_prob: float = field(
+        default=0.0,
+        metadata={
+            "help": "Probability of each feature vector along the feature axis to be chosen as the start of the vector"
+            "span to be masked. Approximately ``mask_feature_prob * sequence_length // mask_feature_length`` feature bins will be masked along the time axis."
+        },
+    )
+    mask_feature_length: int = field(
+        default=10,
+        metadata={"help": "Length of vector span to mask along the feature axis."},
+    )
+    layerdrop: float = field(default=0.0, metadata={"help": "The LayerDrop probability."})
+    ctc_loss_reduction: Optional[str] = field(
+        default="mean", metadata={"help": "The way the ctc loss should be reduced. Should be one of 'mean' or 'sum'."}
+    )
+
+
+@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: str = field(
+        metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
+    )
+    dataset_config_name: str = field(
+        default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
+    )
+    train_split_name: str = field(
+        default="train+validation",
+        metadata={
+            "help": "The name of the training data set split to use (via the datasets library). Defaults to "
+            "'train+validation'"
+        },
+    )
+    eval_split_name: str = field(
+        default="test",
+        metadata={
+            "help": "The name of the training data set split to use (via the datasets library). Defaults to 'test'"
+        },
+    )
+    audio_column_name: str = field(
+        default="audio",
+        metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
+    )
+    text_column_name: str = field(
+        default="text",
+        metadata={"help": "The name of the dataset column containing the text data. Defaults to 'text'"},
+    )
+    overwrite_cache: bool = field(
+        default=False, metadata={"help": "Overwrite the cached preprocessed datasets or not."}
+    )
+    preprocessing_num_workers: Optional[int] = field(
+        default=None,
+        metadata={"help": "The number of processes to use for the preprocessing."},
+    )
+    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 validation examples to this "
+            "value if set."
+        },
+    )
+    shuffle_buffer_size: Optional[int] = field(
+        default=500,
+        metadata={
+            "help": "The number of streamed examples to download before shuffling them. The large the buffer, "
+            "the closer it is to real offline shuffling."
+        },
+    )
+    chars_to_ignore: Optional[List[str]] = list_field(
+        default=None,
+        metadata={"help": "A list of characters to remove from the transcripts."},
+    )
+    eval_metrics: List[str] = list_field(
+        default=["wer"],
+        metadata={"help": "A list of metrics the model should be evaluated on. E.g. `'wer cer'`"},
+    )
+    max_duration_in_seconds: float = field(
+        default=20.0,
+        metadata={"help": "Filter audio files that are longer than `max_duration_in_seconds` seconds."},
+    )
+    preprocessing_only: bool = field(
+        default=False,
+        metadata={
+            "help": "Whether to only do data preprocessing and skip training. "
+            "This is especially useful when data preprocessing errors out in distributed training due to timeout. "
+            "In this case, one should run the preprocessing in a non-distributed setup with `preprocessing_only=True` "
+            "so that the cached datasets can consequently be loaded in distributed training"
+        },
+    )
+    use_auth_token: bool = field(
+        default=False,
+        metadata={
+            "help": "If :obj:`True`, will use the token generated when running"
+            ":obj:`transformers-cli login` as HTTP bearer authorization for remote files."
+        },
+    )
+    phoneme_language: Optional[str] = field(
+        default=None,
+        metadata={
+            "help": "The target language that should be used be"
+            " passed to the tokenizer for tokenization. Note that"
+            " this is only relevant if the model classifies the"
+            " input audio to a sequence of phoneme sequences."
+        },
+    )
+
+
+@dataclass
+class DataCollatorCTCWithPadding:
+    """
+    Data collator that will dynamically pad the inputs received.
+    Args:
+        processor (:class:`~transformers.AutoProcessor`)
+            The processor used for proccessing the data.
+        padding (:obj:`bool`, :obj:`str` or :class:`~transformers.tokenization_utils_base.PaddingStrategy`, `optional`, defaults to :obj:`True`):
+            Select a strategy to pad the returned sequences (according to the model's padding side and padding index)
+            among:
+            * :obj:`True` or :obj:`'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
+              sequence if provided).
+            * :obj:`'max_length'`: Pad to a maximum length specified with the argument :obj:`max_length` or to the
+              maximum acceptable input length for the model if that argument is not provided.
+            * :obj:`False` or :obj:`'do_not_pad'` (default): No padding (i.e., can output a batch with sequences of
+              different lengths).
+        max_length (:obj:`int`, `optional`):
+            Maximum length of the ``input_values`` of the returned list and optionally padding length (see above).
+        max_length_labels (:obj:`int`, `optional`):
+            Maximum length of the ``labels`` returned list and optionally padding length (see above).
+        pad_to_multiple_of (:obj:`int`, `optional`):
+            If set will pad the sequence to a multiple of the provided value.
+            This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >=
+            7.5 (Volta).
+    """
+
+    processor: AutoProcessor
+    padding: Union[bool, str] = "longest"
+    max_length: Optional[int] = None
+    pad_to_multiple_of: Optional[int] = None
+    pad_to_multiple_of_labels: Optional[int] = None
+
+    def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
+        # split inputs and labels since they have to be of different lenghts and need
+        # different padding methods
+        input_features = []
+        label_features = []
+        for feature in features:
+            if self.max_length and feature["input_values"].shape[-1] > self.max_length:
+                continue
+            input_features.append({"input_values": feature["input_values"]})
+            label_features.append({"input_ids": feature["labels"]})
+
+        batch = self.processor.pad(
+            input_features,
+            padding=self.padding,
+            pad_to_multiple_of=self.pad_to_multiple_of,
+            return_tensors="pt",
+        )
+
+        with self.processor.as_target_processor():
+            labels_batch = self.processor.pad(
+                label_features,
+                padding=self.padding,
+                pad_to_multiple_of=self.pad_to_multiple_of_labels,
+                return_tensors="pt",
+            )
+
+        # replace padding with -100 to ignore loss correctly
+        labels = labels_batch["input_ids"].masked_fill(labels_batch.attention_mask.ne(1), -100)
+
+        batch["labels"] = labels
+
+        return batch
+
+
+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()
+
+    # 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 overcome."
+            )
+        elif last_checkpoint is not 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."
+            )
+
+    # 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)],
+    )
+    logger.setLevel(logging.INFO if is_main_process(training_args.local_rank) else logging.WARN)
+
+    # 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}"
+    )
+    # Set the verbosity to info of the Transformers logger (on main process only):
+    if is_main_process(training_args.local_rank):
+        transformers.utils.logging.set_verbosity_info()
+    logger.info("Training/evaluation parameters %s", training_args)
+
+    # Set seed before initializing model.
+    set_seed(training_args.seed)
+
+    # 1. First, let's load the dataset
+    raw_datasets = IterableDatasetDict()
+    raw_column_names = {}
+
+    def load_streaming_dataset(split, sampling_rate, **kwargs):
+        if "+" in split:
+            dataset_splits = [load_dataset(split=split_name, **kwargs) for split_name in split.split("+")]
+            # `features` and `cast_column` won't be available after interleaving, so we'll use them here
+            features = dataset_splits[0].features
+            # make sure that the dataset decodes audio with a correct sampling rate
+            dataset_splits = [
+                dataset.cast_column(data_args.audio_column_name, datasets.features.Audio(sampling_rate=sampling_rate))
+                for dataset in dataset_splits
+            ]
+
+            interleaved_dataset = interleave_datasets(dataset_splits)
+            return interleaved_dataset, features
+        else:
+            dataset = load_dataset(split=split, **kwargs)
+            features = dataset.features
+            # make sure that the dataset decodes audio with a correct sampling rate
+            dataset = dataset.cast_column(
+                data_args.audio_column_name, datasets.features.Audio(sampling_rate=sampling_rate)
+            )
+            return dataset, features
+
+    # `datasets` takes care of automatically loading and resampling the audio,
+    # so we just need to set the correct target sampling rate and normalize the input
+    # via the `feature_extractor`
+    feature_extractor = AutoFeatureExtractor.from_pretrained(
+        model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
+    )
+
+    if training_args.do_train:
+        raw_datasets["train"], train_features = load_streaming_dataset(
+            path=data_args.dataset_name,
+            name=data_args.dataset_config_name,
+            split=data_args.train_split_name,
+            use_auth_token=data_args.use_auth_token,
+            streaming=True,
+            sampling_rate=feature_extractor.sampling_rate,
+        )
+        raw_column_names["train"] = list(train_features.keys())
+
+        if data_args.audio_column_name not in raw_column_names["train"]:
+            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_column_names['train'])}."
+            )
+
+        if data_args.text_column_name not in raw_column_names["train"]:
+            raise ValueError(
+                f"--text_column_name {data_args.text_column_name} not found in dataset '{data_args.dataset_name}'. "
+                "Make sure to set `--text_column_name` to the correct text column - one of "
+                f"{', '.join(raw_column_names['train'])}."
+            )
+
+        if data_args.max_train_samples is not None:
+            raw_datasets["train"] = raw_datasets["train"].take(range(data_args.max_train_samples))
+
+    if training_args.do_eval:
+        raw_datasets["eval"], eval_features = load_streaming_dataset(
+            path=data_args.dataset_name,
+            name=data_args.dataset_config_name,
+            split=data_args.eval_split_name,
+            use_auth_token=data_args.use_auth_token,
+            streaming=True,
+            sampling_rate=feature_extractor.sampling_rate,
+        )
+        raw_column_names["eval"] = list(eval_features.keys())
+
+        if data_args.max_eval_samples is not None:
+            raw_datasets["eval"] = raw_datasets["eval"].take(range(data_args.max_eval_samples))
+
+    # 2. We remove some special characters from the datasets
+    # that make training complicated and do not help in transcribing the speech
+    # E.g. characters, such as `,` and `.` do not really have an acoustic characteristic
+    # that could be easily picked up by the model
+    chars_to_ignore_regex = (
+        f'[{"".join(data_args.chars_to_ignore)}]' if data_args.chars_to_ignore is not None else None
+    )
+    text_column_name = data_args.text_column_name
+
+    def remove_special_characters(batch):
+        if chars_to_ignore_regex is not None:
+            batch["target_text"] = re.sub(chars_to_ignore_regex, "", batch[text_column_name]).lower() + " "
+        else:
+            batch["target_text"] = batch[text_column_name].lower() + " "
+        return batch
+
+    with training_args.main_process_first(desc="dataset map special characters removal"):
+        for split, dataset in raw_datasets.items():
+            raw_datasets[split] = dataset.map(
+                remove_special_characters,
+            ).remove_columns([text_column_name])
+
+    # 3. Next, let's load the config as we might need it to create
+    # the tokenizer
+    config = AutoConfig.from_pretrained(
+        model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_auth_token=data_args.use_auth_token
+    )
+
+    # 4. Now we can instantiate the tokenizer and model
+    # Note for distributed training, the .from_pretrained methods guarantee that only
+    # one local process can concurrently download model & vocab.
+
+    tokenizer_name_or_path = model_args.tokenizer_name_or_path
+    if tokenizer_name_or_path is None:
+        raise ValueError(
+            "Tokenizer has to be created before training in streaming mode. Please specify --tokenizer_name_or_path"
+        )
+    # load feature_extractor and tokenizer
+    tokenizer = AutoTokenizer.from_pretrained(
+        tokenizer_name_or_path,
+        config=config,
+        use_auth_token=data_args.use_auth_token,
+    )
+
+    # adapt config
+    config.update(
+        {
+            "feat_proj_dropout": model_args.feat_proj_dropout,
+            "attention_dropout": model_args.attention_dropout,
+            "hidden_dropout": model_args.hidden_dropout,
+            "final_dropout": model_args.final_dropout,
+            "mask_time_prob": model_args.mask_time_prob,
+            "mask_time_length": model_args.mask_time_length,
+            "mask_feature_prob": model_args.mask_feature_prob,
+            "mask_feature_length": model_args.mask_feature_length,
+            "gradient_checkpointing": training_args.gradient_checkpointing,
+            "layerdrop": model_args.layerdrop,
+            "ctc_loss_reduction": model_args.ctc_loss_reduction,
+            "pad_token_id": tokenizer.pad_token_id,
+            "vocab_size": len(tokenizer),
+            "activation_dropout": model_args.activation_dropout,
+        }
+    )
+
+    # create model
+    model = AutoModelForCTC.from_pretrained(
+        model_args.model_name_or_path,
+        cache_dir=model_args.cache_dir,
+        config=config,
+        use_auth_token=data_args.use_auth_token,
+    )
+
+    # freeze encoder
+    if model_args.freeze_feature_encoder:
+        model.freeze_feature_encoder()
+
+    # 5. Now we preprocess the datasets including loading the audio, resampling and normalization
+    audio_column_name = data_args.audio_column_name
+
+    # `phoneme_language` is only relevant if the model is fine-tuned on phoneme classification
+    phoneme_language = data_args.phoneme_language
+
+    # Preprocessing the datasets.
+    # We need to read the audio files as arrays and tokenize the targets.
+    def prepare_dataset(batch):
+        # load audio
+        sample = batch[audio_column_name]
+
+        inputs = feature_extractor(sample["array"], sampling_rate=sample["sampling_rate"])
+        batch["input_values"] = inputs.input_values[0]
+        batch["input_length"] = len(batch["input_values"])
+
+        # encode targets
+        additional_kwargs = {}
+        if phoneme_language is not None:
+            additional_kwargs["phonemizer_lang"] = phoneme_language
+
+        batch["labels"] = tokenizer(batch["target_text"], **additional_kwargs).input_ids
+        return batch
+
+    vectorized_datasets = IterableDatasetDict()
+    with training_args.main_process_first(desc="dataset map preprocessing"):
+        for split, dataset in raw_datasets.items():
+            vectorized_datasets[split] = (
+                dataset.map(prepare_dataset)
+                .remove_columns(raw_column_names[split] + ["target_text"])
+                .with_format("torch")
+            )
+            if split == "train":
+                vectorized_datasets[split] = vectorized_datasets[split].shuffle(
+                    buffer_size=data_args.shuffle_buffer_size,
+                    seed=training_args.seed,
+                )
+
+    # 6. Next, we can prepare the training.
+    # Let's use word error rate (WER) as our evaluation metric,
+    # instantiate a data collator and the trainer
+
+    # Define evaluation metrics during training, *i.e.* word error rate, character error rate
+    eval_metrics = {metric: load_metric(metric) for metric in data_args.eval_metrics}
+
+    def compute_metrics(pred):
+        pred_logits = pred.predictions
+        pred_ids = np.argmax(pred_logits, axis=-1)
+
+        pred.label_ids[pred.label_ids == -100] = tokenizer.pad_token_id
+
+        pred_str = tokenizer.batch_decode(pred_ids)
+        # we do not want to group tokens when computing the metrics
+        label_str = tokenizer.batch_decode(pred.label_ids, group_tokens=False)
+
+        metrics = {k: v.compute(predictions=pred_str, references=label_str) for k, v in eval_metrics.items()}
+
+        return metrics
+
+    # Now save everything to be able to create a single processor later
+    if is_main_process(training_args.local_rank):
+        # save feature extractor, tokenizer and config
+        feature_extractor.save_pretrained(training_args.output_dir)
+        tokenizer.save_pretrained(training_args.output_dir)
+        config.save_pretrained(training_args.output_dir)
+
+    try:
+        processor = AutoProcessor.from_pretrained(training_args.output_dir)
+    except (OSError, KeyError):
+        warnings.warn(
+            "Loading a processor from a feature extractor config that does not"
+            " include a `processor_class` attribute is deprecated and will be removed in v5. Please add the following "
+            " attribute to your `preprocessor_config.json` file to suppress this warning: "
+            " `'processor_class': 'Wav2Vec2Processor'`",
+            FutureWarning,
+        )
+        processor = Wav2Vec2Processor.from_pretrained(training_args.output_dir)
+
+    # Instantiate custom data collator
+    max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
+    data_collator = DataCollatorCTCWithPadding(processor=processor, max_length=max_input_length)
+
+    # trainer callback to reinitialize and reshuffle the streamable datasets at the beginning of each epoch
+    class ShuffleCallback(TrainerCallback):
+        def on_epoch_begin(self, args, state, control, train_dataloader, **kwargs):
+            if isinstance(train_dataloader.dataset, IterableDatasetShard):
+                pass  # set_epoch() is handled by the Trainer
+            elif isinstance(train_dataloader.dataset, IterableDataset):
+                train_dataloader.dataset.set_epoch(train_dataloader.dataset._epoch + 1)
+
+    # Initialize Trainer
+    trainer = Trainer(
+        model=model,
+        data_collator=data_collator,
+        args=training_args,
+        compute_metrics=compute_metrics,
+        train_dataset=vectorized_datasets["train"] if training_args.do_train else None,
+        eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
+        tokenizer=processor,
+        callbacks=[ShuffleCallback()],
+    )
+
+    # 7. Finally, we can start training
+
+    # Training
+    if training_args.do_train:
+
+        # use last checkpoint if exist
+        if last_checkpoint is not None:
+            checkpoint = last_checkpoint
+        elif os.path.isdir(model_args.model_name_or_path):
+            checkpoint = model_args.model_name_or_path
+        else:
+            checkpoint = None
+
+        train_result = trainer.train(resume_from_checkpoint=checkpoint)
+        trainer.save_model()
+
+        metrics = train_result.metrics
+        if data_args.max_train_samples:
+            metrics["train_samples"] = data_args.max_train_samples
+
+        trainer.log_metrics("train", metrics)
+        trainer.save_metrics("train", metrics)
+        trainer.save_state()
+
+    # Evaluation
+    results = {}
+    if training_args.do_eval:
+        logger.info("*** Evaluate ***")
+        metrics = trainer.evaluate()
+        if data_args.max_eval_samples:
+            metrics["eval_samples"] = data_args.max_eval_samples
+
+        trainer.log_metrics("eval", metrics)
+        trainer.save_metrics("eval", metrics)
+
+    # Write model card and (optionally) push to hub
+    config_name = data_args.dataset_config_name if data_args.dataset_config_name is not None else "na"
+    kwargs = {
+        "finetuned_from": model_args.model_name_or_path,
+        "tasks": "speech-recognition",
+        "tags": ["automatic-speech-recognition", data_args.dataset_name],
+        "dataset_args": f"Config: {config_name}, Training split: {data_args.train_split_name}, Eval split: {data_args.eval_split_name}",
+        "dataset": f"{data_args.dataset_name.upper()} - {config_name.upper()}",
+    }
+    if "common_voice" in data_args.dataset_name:
+        kwargs["language"] = config_name
+
+    if training_args.push_to_hub:
+        trainer.push_to_hub(**kwargs)
+    else:
+        trainer.create_model_card(**kwargs)
+
+    return results
+
+
+if __name__ == "__main__":
+    main()