Remove old files

.gitattributes

@@ -1,27 +0,0 @@
-*.7z filter=lfs diff=lfs merge=lfs -text
-*.arrow filter=lfs diff=lfs merge=lfs -text
-*.bin filter=lfs diff=lfs merge=lfs -text
-*.bin.* filter=lfs diff=lfs merge=lfs -text
-*.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ftz filter=lfs diff=lfs merge=lfs -text
-*.gz filter=lfs diff=lfs merge=lfs -text
-*.h5 filter=lfs diff=lfs merge=lfs -text
-*.joblib filter=lfs diff=lfs merge=lfs -text
-*.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.model filter=lfs diff=lfs merge=lfs -text
-*.msgpack filter=lfs diff=lfs merge=lfs -text
-*.onnx filter=lfs diff=lfs merge=lfs -text
-*.ot filter=lfs diff=lfs merge=lfs -text
-*.parquet filter=lfs diff=lfs merge=lfs -text
-*.pb filter=lfs diff=lfs merge=lfs -text
-*.pt filter=lfs diff=lfs merge=lfs -text
-*.pth filter=lfs diff=lfs merge=lfs -text
-*.rar filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text 
-*.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tflite filter=lfs diff=lfs merge=lfs -text
-*.tgz filter=lfs diff=lfs merge=lfs -text
-*.xz filter=lfs diff=lfs merge=lfs -text
-*.zip filter=lfs diff=lfs merge=lfs -text
-*.zstandard filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -1,2 +0,0 @@
-checkpoint-*/
-.ipynb_checkpoints/*

README.md

@@ -1,69 +0,0 @@
----
-license: apache-2.0
-tags:
-- automatic-speech-recognition
-- openslr_SLR66
-- robust-speech-event
-- generated_from_trainer
-model-index:
-- name: ''
-  results: []
----
-
-<!-- This model card has been generated automatically according to the information the Trainer had access to. You
-should probably proofread and complete it, then remove this comment. -->
-
-# 
-
-This model is a fine-tuned version of [facebook/wav2vec2-xls-r-2b](https://huggingface.co/facebook/wav2vec2-xls-r-2b) on the OPENSLR_SLR66 - NA dataset.
-It achieves the following results on the evaluation set:
-- Loss: 0.9243
-- Wer: 0.9293
-
-## Model description
-
-More information needed
-
-## Intended uses & limitations
-
-More information needed
-
-## Training and evaluation data
-
-More information needed
-
-## Training procedure
-
-### Training hyperparameters
-
-The following hyperparameters were used during training:
-- learning_rate: 7.5e-05
-- train_batch_size: 4
-- eval_batch_size: 4
-- seed: 42
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 32
-- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
-- lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 2000
-- num_epochs: 60.0
-- mixed_precision_training: Native AMP
-
-### Training results
-
-| Training Loss | Epoch | Step | Validation Loss | Wer    |
-|:-------------:|:-----:|:----:|:---------------:|:------:|
-| 1.8141        | 9.61  | 1000 | 0.7609          | 0.7778 |
-| 2.2935        | 19.23 | 2000 | 1.1849          | 0.9657 |
-| 2.2848        | 28.84 | 3000 | 1.1789          | 0.9760 |
-| 2.3325        | 38.46 | 4000 | 1.0878          | 0.9595 |
-| 2.25          | 48.08 | 5000 | 1.0888          | 0.9586 |
-| 2.0757        | 57.69 | 6000 | 0.9504          | 0.9412 |
-
-
-### Framework versions
-
-- Transformers 4.16.0.dev0
-- Pytorch 1.10.1+cu102
-- Datasets 1.17.1.dev0
-- Tokenizers 0.11.0

added_tokens.json

@@ -1 +0,0 @@
-{"<s>": 77, "</s>": 78}

all_results.json

@@ -1,14 +0,0 @@
-{
-    "epoch": 60.0,
-    "eval_loss": 0.9243341088294983,
-    "eval_runtime": 80.7101,
-    "eval_samples": 1112,
-    "eval_samples_per_second": 13.778,
-    "eval_steps_per_second": 3.444,
-    "eval_wer": 0.9293273542600897,
-    "train_loss": 2.2000570346147588,
-    "train_runtime": 38994.8678,
-    "train_samples": 3336,
-    "train_samples_per_second": 5.133,
-    "train_steps_per_second": 0.16
-}

config.json

@@ -1,107 +0,0 @@
-{
-  "_name_or_path": "facebook/wav2vec2-xls-r-2b",
-  "activation_dropout": 0.1,
-  "adapter_kernel_size": 3,
-  "adapter_stride": 2,
-  "add_adapter": false,
-  "apply_spec_augment": true,
-  "architectures": [
-    "Wav2Vec2ForCTC"
-  ],
-  "attention_dropout": 0.0,
-  "bos_token_id": 1,
-  "classifier_proj_size": 256,
-  "codevector_dim": 1024,
-  "contrastive_logits_temperature": 0.1,
-  "conv_bias": true,
-  "conv_dim": [
-    512,
-    512,
-    512,
-    512,
-    512,
-    512,
-    512
-  ],
-  "conv_kernel": [
-    10,
-    3,
-    3,
-    3,
-    3,
-    2,
-    2
-  ],
-  "conv_stride": [
-    5,
-    2,
-    2,
-    2,
-    2,
-    2,
-    2
-  ],
-  "ctc_loss_reduction": "mean",
-  "ctc_zero_infinity": false,
-  "diversity_loss_weight": 0.1,
-  "do_stable_layer_norm": true,
-  "eos_token_id": 2,
-  "feat_extract_activation": "gelu",
-  "feat_extract_dropout": 0.0,
-  "feat_extract_norm": "layer",
-  "feat_proj_dropout": 0.0,
-  "feat_quantizer_dropout": 0.0,
-  "final_dropout": 0.0,
-  "hidden_act": "gelu",
-  "hidden_dropout": 0.0,
-  "hidden_size": 1920,
-  "initializer_range": 0.02,
-  "intermediate_size": 7680,
-  "layer_norm_eps": 1e-05,
-  "layerdrop": 0.0,
-  "mask_feature_length": 64,
-  "mask_feature_min_masks": 0,
-  "mask_feature_prob": 0.25,
-  "mask_time_length": 10,
-  "mask_time_min_masks": 2,
-  "mask_time_prob": 0.75,
-  "model_type": "wav2vec2",
-  "num_adapter_layers": 3,
-  "num_attention_heads": 16,
-  "num_codevector_groups": 2,
-  "num_codevectors_per_group": 320,
-  "num_conv_pos_embedding_groups": 16,
-  "num_conv_pos_embeddings": 128,
-  "num_feat_extract_layers": 7,
-  "num_hidden_layers": 48,
-  "num_negatives": 100,
-  "output_hidden_size": 1920,
-  "pad_token_id": 76,
-  "proj_codevector_dim": 1024,
-  "tdnn_dilation": [
-    1,
-    2,
-    3,
-    1,
-    1
-  ],
-  "tdnn_dim": [
-    512,
-    512,
-    512,
-    512,
-    1500
-  ],
-  "tdnn_kernel": [
-    5,
-    3,
-    3,
-    1,
-    1
-  ],
-  "torch_dtype": "float32",
-  "transformers_version": "4.16.0.dev0",
-  "use_weighted_layer_sum": false,
-  "vocab_size": 79,
-  "xvector_output_dim": 512
-}

eval_results.json

@@ -1,9 +0,0 @@
-{
-    "epoch": 60.0,
-    "eval_loss": 0.9243341088294983,
-    "eval_runtime": 80.7101,
-    "eval_samples": 1112,
-    "eval_samples_per_second": 13.778,
-    "eval_steps_per_second": 3.444,
-    "eval_wer": 0.9293273542600897
-}

preprocessor_config.json

@@ -1,9 +0,0 @@
-{
-  "do_normalize": true,
-  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
-  "feature_size": 1,
-  "padding_side": "right",
-  "padding_value": 0,
-  "return_attention_mask": true,
-  "sampling_rate": 16000
-}

pytorch_model.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:6447b9bfe1ae69ddefa68b10c300469609be3bdc873e1d729bcd1461962117a9
-size 8637973773

run.sh

@@ -1,35 +0,0 @@
-python run_speech_recognition_ctc.py \
---model_name_or_path="facebook/wav2vec2-xls-r-2b" \
---dataset_name="openslr_SLR66" \
---train_split_name="train" \
---preprocessing_num_workers="8" \
---output_dir="./" \
---overwrite_output_dir \
---num_train_epochs="5" \
---per_device_train_batch_size="4" \
---per_device_eval_batch_size="4" \
---gradient_accumulation_steps="8" \
---learning_rate="7.5e-5" \
---warmup_steps="2000" \
---length_column_name="input_length" \
---evaluation_strategy="steps" \
---text_column_name="sentence" \
---chars_to_ignore , ? . ! \- \; \: \" “ % ‘ ” � — ’ … – \
---save_steps="500" \
---eval_steps="500" \
---logging_steps="100" \
---layerdrop="0.0" \
---activation_dropout="0.1" \
---save_total_limit="3" \
---freeze_feature_encoder \
---feat_proj_dropout="0.0" \
---mask_time_prob="0.75" \
---mask_time_length="10" \
---mask_feature_prob="0.25" \
---mask_feature_length="64" \
---gradient_checkpointing \
---use_auth_token \
---fp16 \
---group_by_length \
---do_train --do_eval \
---push_to_hub

run_bnb.sh

@@ -1,34 +0,0 @@
-python run_speech_recognition_ctc_bnb.py \
---model_name_or_path="facebook/wav2vec2-xls-r-2b" \
---dataset_name="openslr_SLR66" \
---train_split_name="train" \
---preprocessing_num_workers="8" \
---output_dir="./" \
---overwrite_output_dir \
---num_train_epochs="60" \
---per_device_train_batch_size="4" \
---per_device_eval_batch_size="4" \
---gradient_accumulation_steps="8" \
---learning_rate="7.5e-5" \
---warmup_steps="2000" \
---length_column_name="input_length" \
---evaluation_strategy="steps" \
---text_column_name="sentence" \
---chars_to_ignore , ? . ! \- \; \: \" “ % ‘ ” � — ’ … – \
---save_steps="1000" \
---eval_steps="1000" \
---logging_steps="100" \
---layerdrop="0.0" \
---activation_dropout="0.1" \
---save_total_limit="3" \
---freeze_feature_encoder \
---feat_proj_dropout="0.0" \
---mask_time_prob="0.75" \
---mask_time_length="10" \
---mask_feature_prob="0.25" \
---mask_feature_length="64" \
---gradient_checkpointing \
---use_auth_token \
---fp16 \
---group_by_length \
---do_train --do_eval

run_speech_recognition_ctc.py

@@ -1,756 +0,0 @@
-#!/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
-
-""" Fine-tuning a 🤗 Transformers CTC model for automatic speech recognition"""
-
-import functools
-import json
-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 DatasetDict, load_dataset, load_metric
-
-import transformers
-from transformers import (
-    AutoConfig,
-    AutoFeatureExtractor,
-    AutoModelForCTC,
-    AutoProcessor,
-    AutoTokenizer,
-    HfArgumentParser,
-    Trainer,
-    TrainingArguments,
-    Wav2Vec2Processor,
-    set_seed,
-)
-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 risks.
-check_min_version("4.16.0.dev0")
-
-require_version("datasets>=1.13.3", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
-
-
-logger = logging.getLogger(__name__)
-
-
-def list_field(default=None, metadata=None):
-    return field(default_factory=lambda: default, metadata=metadata)
-
-def get_telugu_dataset(validation_split=False):
-    dataset = load_dataset('openslr', 'SLR66')
-
-    seed=1242
-    
-    if validation_split:
-        train_testvalid = dataset['train'].train_test_split(test_size=0.2, seed=seed)
-        # Split the 10% test + valid in half test, half valid
-        test_valid = train_testvalid['test'].train_test_split(test_size=0.33, seed=seed)
-        # gather everyone if you want to have a single DatasetDict
-        out_dataset = DatasetDict({
-            'train': train_testvalid['train'],
-            'test': test_valid['test'],
-            'valid': test_valid['train']})
-    else:
-        train_testvalid = dataset['train'].train_test_split(test_size=0.25, seed=seed)
-        out_dataset = DatasetDict({
-            'train': train_testvalid['train'],
-            'test': train_testvalid['test']})
-    return out_dataset
-
-
-@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'"
-        },
-    )
-    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 'train'"
-        },
-    )
-    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."
-        },
-    )
-    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 to 'max_duration_in_seconds`"
-        },
-    )
-    min_duration_in_seconds: float = field(
-        default=0.0, metadata={"help": "Filter audio files that are shorter than `min_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."
-        },
-    )
-    unk_token: str = field(
-        default="[UNK]",
-        metadata={"help": "The unk token for the tokenizer"},
-    )
-    pad_token: str = field(
-        default="[PAD]",
-        metadata={"help": "The padding token for the tokenizer"},
-    )
-    word_delimiter_token: str = field(
-        default="|",
-        metadata={"help": "The word delimiter token for the tokenizer"},
-    )
-    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"
-    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 = [{"input_values": feature["input_values"]} for feature in features]
-        label_features = [{"input_ids": feature["labels"]} for feature in features]
-
-        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 create_vocabulary_from_data(
-    datasets: DatasetDict,
-    word_delimiter_token: Optional[str] = None,
-    unk_token: Optional[str] = None,
-    pad_token: Optional[str] = None,
-):
-    # Given training and test labels create vocabulary
-    def extract_all_chars(batch):
-        all_text = " ".join(batch["target_text"])
-        vocab = list(set(all_text))
-        return {"vocab": [vocab], "all_text": [all_text]}
-
-    vocabs = datasets.map(
-        extract_all_chars,
-        batched=True,
-        batch_size=-1,
-        keep_in_memory=True,
-        remove_columns=datasets["train"].column_names,
-    )
-
-    # take union of all unique characters in each dataset
-    vocab_set = functools.reduce(
-        lambda vocab_1, vocab_2: set(vocab_1["vocab"][0]) | set(vocab_2["vocab"][0]), vocabs.values()
-    )
-
-    vocab_dict = {v: k for k, v in enumerate(sorted(list(vocab_set)))}
-
-    # replace white space with delimiter token
-    if word_delimiter_token is not None:
-        vocab_dict[word_delimiter_token] = vocab_dict[" "]
-        del vocab_dict[" "]
-
-    # add unk and pad token
-    if unk_token is not None:
-        vocab_dict[unk_token] = len(vocab_dict)
-
-    if pad_token is not None:
-        vocab_dict[pad_token] = len(vocab_dict)
-
-    return vocab_dict
-
-
-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()
-    
-    # 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)
-    
-    # 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."
-            )
-
-    # 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
-    te_dataset = get_telugu_dataset(validation_split=False)
-    def load_te_dataset(split):
-        return te_dataset[split]
-
-    raw_datasets = DatasetDict()
-
-    if training_args.do_train:
-        raw_datasets["train"] = load_te_dataset(
-                    split=data_args.train_split_name
-        )
-
-        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.text_column_name not in raw_datasets["train"].column_names:
-            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_datasets['train'].column_names)}."
-            )
-
-        if data_args.max_train_samples is not None:
-            raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
-
-    if training_args.do_eval:
-        raw_datasets["eval"] = load_te_dataset(
-                    split=data_args.eval_split_name
-        )
-
-        if data_args.max_eval_samples is not None:
-            raw_datasets["eval"] = raw_datasets["eval"].select(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"):
-        raw_datasets = raw_datasets.map(
-            remove_special_characters,
-            remove_columns=[text_column_name],
-            desc="remove special characters from datasets",
-        )
-
-    # save special tokens for tokenizer
-    word_delimiter_token = data_args.word_delimiter_token
-    unk_token = data_args.unk_token
-    pad_token = data_args.pad_token
-
-    # 3. Next, let's load the config as we might need it to create
-    # the tokenizer
-    # load config
-    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. Next, if no tokenizer file is defined,
-    # we create the vocabulary of the model by extracting all unique characters from
-    # the training and evaluation datasets
-    # We need to make sure that only first rank saves vocabulary
-    # make sure all processes wait until vocab is created
-    tokenizer_name_or_path = model_args.tokenizer_name_or_path
-    tokenizer_kwargs = {}
-    if tokenizer_name_or_path is None:
-        # save vocab in training output dir
-        tokenizer_name_or_path = training_args.output_dir
-
-        vocab_file = os.path.join(tokenizer_name_or_path, "vocab.json")
-
-        with training_args.main_process_first():
-            if training_args.overwrite_output_dir and os.path.isfile(vocab_file):
-                os.remove(vocab_file)
-
-        with training_args.main_process_first(desc="dataset map vocabulary creation"):
-            if not os.path.isfile(vocab_file):
-                os.makedirs(tokenizer_name_or_path, exist_ok=True)
-                vocab_dict = create_vocabulary_from_data(
-                    raw_datasets,
-                    word_delimiter_token=word_delimiter_token,
-                    unk_token=unk_token,
-                    pad_token=pad_token,
-                )
-
-                # save vocab dict to be loaded into tokenizer
-                with open(vocab_file, "w") as file:
-                    json.dump(vocab_dict, file)
-
-        # if tokenizer has just been created
-        # it is defined by `tokenizer_class` if present in config else by `model_type`
-        tokenizer_kwargs = {
-            "config": config if config.tokenizer_class is not None else None,
-            "tokenizer_type": config.model_type if config.tokenizer_class is None else None,
-            "unk_token": unk_token,
-            "pad_token": pad_token,
-            "word_delimiter_token": word_delimiter_token,
-        }
-
-    # 5. Now we can instantiate the feature extractor, tokenizer and model
-    # Note for distributed training, the .from_pretrained methods guarantee that only
-    # one local process can concurrently download model & vocab.
-
-    # load feature_extractor and tokenizer
-    tokenizer = AutoTokenizer.from_pretrained(
-        tokenizer_name_or_path,
-        use_auth_token=data_args.use_auth_token,
-        **tokenizer_kwargs,
-    )
-    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
-    )
-
-    # 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()
-
-    # 6. Now we preprocess the datasets including loading the audio, resampling and normalization
-    # Thankfully, `datasets` takes care of automatically loading and resampling the audio,
-    # so that we just need to set the correct target sampling rate and normalize the input
-    # via the `feature_extractor`
-
-    # make sure that dataset decodes audio with correct sampling rate
-    dataset_sampling_rate = next(iter(raw_datasets.values())).features[data_args.audio_column_name].sampling_rate
-    if dataset_sampling_rate != feature_extractor.sampling_rate:
-        raw_datasets = raw_datasets.cast_column(
-            data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
-        )
-
-    # derive max & min input length for sample rate & max duration
-    max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
-    min_input_length = data_args.min_duration_in_seconds * feature_extractor.sampling_rate
-    audio_column_name = data_args.audio_column_name
-    num_workers = data_args.preprocessing_num_workers
-
-    # `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
-
-    with training_args.main_process_first(desc="dataset map preprocessing"):
-        vectorized_datasets = raw_datasets.map(
-            prepare_dataset,
-            remove_columns=next(iter(raw_datasets.values())).column_names,
-            num_proc=num_workers,
-            desc="preprocess datasets",
-        )
-
-        def is_audio_in_length_range(length):
-            return length > min_input_length and length < max_input_length
-
-        # filter data that is shorter than min_input_length
-        vectorized_datasets = vectorized_datasets.filter(
-            is_audio_in_length_range,
-            num_proc=num_workers,
-            input_columns=["input_length"],
-        )
-
-    # 7. 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}
-
-    # for large datasets it is advised to run the preprocessing on a
-    # single machine first with ``args.preprocessing_only`` since there will mostly likely
-    # be a timeout when running the script in distributed mode.
-    # In a second step ``args.preprocessing_only`` can then be set to `False` to load the
-    # cached dataset
-    if data_args.preprocessing_only:
-        logger.info(f"Data preprocessing finished. Files cached at {vectorized_datasets.cache_files}")
-        return
-
-    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
-    data_collator = DataCollatorCTCWithPadding(processor=processor)
-
-    # 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=feature_extractor,
-    )
-
-    # 8. 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
-        max_train_samples = (
-            data_args.max_train_samples
-            if data_args.max_train_samples is not None
-            else len(vectorized_datasets["train"])
-        )
-        metrics["train_samples"] = min(max_train_samples, len(vectorized_datasets["train"]))
-
-        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()
-        max_eval_samples = (
-            data_args.max_eval_samples if data_args.max_eval_samples is not None else len(vectorized_datasets["eval"])
-        )
-        metrics["eval_samples"] = min(max_eval_samples, len(vectorized_datasets["eval"]))
-
-        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, "robust-speech-event"],
-        "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()

run_speech_recognition_ctc_bnb.py

@@ -1,780 +0,0 @@
-#!/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
-
-""" Fine-tuning a 🤗 Transformers CTC model for automatic speech recognition"""
-
-import functools
-import json
-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 DatasetDict, load_dataset, load_metric
-
-import bitsandbytes as bnb
-import transformers
-from transformers import (
-    AutoConfig,
-    AutoFeatureExtractor,
-    AutoModelForCTC,
-    AutoProcessor,
-    AutoTokenizer,
-    HfArgumentParser,
-    Trainer,
-    TrainingArguments,
-    Wav2Vec2Processor,
-    set_seed,
-)
-from transformers.trainer_pt_utils import get_parameter_names
-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 risks.
-check_min_version("4.16.0.dev0")
-
-require_version("datasets>=1.13.3", "To fix: pip install -r examples/pytorch/text-classification/requirements.txt")
-
-
-logger = logging.getLogger(__name__)
-
-
-def list_field(default=None, metadata=None):
-    return field(default_factory=lambda: default, metadata=metadata)
-
-def get_telugu_dataset(validation_split=False):
-    dataset = load_dataset('openslr', 'SLR66')
-
-    seed=1242
-    
-    if validation_split:
-        train_testvalid = dataset['train'].train_test_split(test_size=0.2, seed=seed)
-        # Split the 10% test + valid in half test, half valid
-        test_valid = train_testvalid['test'].train_test_split(test_size=0.33, seed=seed)
-        # gather everyone if you want to have a single DatasetDict
-        out_dataset = DatasetDict({
-            'train': train_testvalid['train'],
-            'test': test_valid['test'],
-            'valid': test_valid['train']})
-    else:
-        train_testvalid = dataset['train'].train_test_split(test_size=0.25, seed=seed)
-        out_dataset = DatasetDict({
-            'train': train_testvalid['train'],
-            'test': train_testvalid['test']})
-    return out_dataset
-
-
-@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'"
-        },
-    )
-    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 'train'"
-        },
-    )
-    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."
-        },
-    )
-    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 to 'max_duration_in_seconds`"
-        },
-    )
-    min_duration_in_seconds: float = field(
-        default=0.0, metadata={"help": "Filter audio files that are shorter than `min_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."
-        },
-    )
-    unk_token: str = field(
-        default="[UNK]",
-        metadata={"help": "The unk token for the tokenizer"},
-    )
-    pad_token: str = field(
-        default="[PAD]",
-        metadata={"help": "The padding token for the tokenizer"},
-    )
-    word_delimiter_token: str = field(
-        default="|",
-        metadata={"help": "The word delimiter token for the tokenizer"},
-    )
-    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"
-    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 = [{"input_values": feature["input_values"]} for feature in features]
-        label_features = [{"input_ids": feature["labels"]} for feature in features]
-
-        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 create_vocabulary_from_data(
-    datasets: DatasetDict,
-    word_delimiter_token: Optional[str] = None,
-    unk_token: Optional[str] = None,
-    pad_token: Optional[str] = None,
-):
-    # Given training and test labels create vocabulary
-    def extract_all_chars(batch):
-        all_text = " ".join(batch["target_text"])
-        vocab = list(set(all_text))
-        return {"vocab": [vocab], "all_text": [all_text]}
-
-    vocabs = datasets.map(
-        extract_all_chars,
-        batched=True,
-        batch_size=-1,
-        keep_in_memory=True,
-        remove_columns=datasets["train"].column_names,
-    )
-
-    # take union of all unique characters in each dataset
-    vocab_set = functools.reduce(
-        lambda vocab_1, vocab_2: set(vocab_1["vocab"][0]) | set(vocab_2["vocab"][0]), vocabs.values()
-    )
-
-    vocab_dict = {v: k for k, v in enumerate(sorted(list(vocab_set)))}
-
-    # replace white space with delimiter token
-    if word_delimiter_token is not None:
-        vocab_dict[word_delimiter_token] = vocab_dict[" "]
-        del vocab_dict[" "]
-
-    # add unk and pad token
-    if unk_token is not None:
-        vocab_dict[unk_token] = len(vocab_dict)
-
-    if pad_token is not None:
-        vocab_dict[pad_token] = len(vocab_dict)
-
-    return vocab_dict
-
-
-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()
-    
-    # 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)
-    
-    # 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."
-            )
-
-    
-    # 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
-    te_dataset = get_telugu_dataset(validation_split=False)
-    def load_te_dataset(split):
-        return te_dataset[split]
-
-    raw_datasets = DatasetDict()
-
-    if training_args.do_train:
-        raw_datasets["train"] = load_te_dataset(
-                    split=data_args.train_split_name
-        )
-
-        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.text_column_name not in raw_datasets["train"].column_names:
-            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_datasets['train'].column_names)}."
-            )
-
-        if data_args.max_train_samples is not None:
-            raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
-
-    if training_args.do_eval:
-        raw_datasets["eval"] = load_te_dataset(
-                    split=data_args.eval_split_name
-        )
-
-        if data_args.max_eval_samples is not None:
-            raw_datasets["eval"] = raw_datasets["eval"].select(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"):
-        raw_datasets = raw_datasets.map(
-            remove_special_characters,
-            remove_columns=[text_column_name],
-            desc="remove special characters from datasets",
-        )
-
-    # save special tokens for tokenizer
-    word_delimiter_token = data_args.word_delimiter_token
-    unk_token = data_args.unk_token
-    pad_token = data_args.pad_token
-
-    # 3. Next, let's load the config as we might need it to create
-    # the tokenizer
-    # load config
-    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. Next, if no tokenizer file is defined,
-    # we create the vocabulary of the model by extracting all unique characters from
-    # the training and evaluation datasets
-    # We need to make sure that only first rank saves vocabulary
-    # make sure all processes wait until vocab is created
-    tokenizer_name_or_path = model_args.tokenizer_name_or_path
-    tokenizer_kwargs = {}
-    if tokenizer_name_or_path is None:
-        # save vocab in training output dir
-        tokenizer_name_or_path = training_args.output_dir
-
-        vocab_file = os.path.join(tokenizer_name_or_path, "vocab.json")
-
-        with training_args.main_process_first():
-            if training_args.overwrite_output_dir and os.path.isfile(vocab_file):
-                os.remove(vocab_file)
-
-        with training_args.main_process_first(desc="dataset map vocabulary creation"):
-            if not os.path.isfile(vocab_file):
-                os.makedirs(tokenizer_name_or_path, exist_ok=True)
-                vocab_dict = create_vocabulary_from_data(
-                    raw_datasets,
-                    word_delimiter_token=word_delimiter_token,
-                    unk_token=unk_token,
-                    pad_token=pad_token,
-                )
-
-                # save vocab dict to be loaded into tokenizer
-                with open(vocab_file, "w") as file:
-                    json.dump(vocab_dict, file)
-
-        # if tokenizer has just been created
-        # it is defined by `tokenizer_class` if present in config else by `model_type`
-        tokenizer_kwargs = {
-            "config": config if config.tokenizer_class is not None else None,
-            "tokenizer_type": config.model_type if config.tokenizer_class is None else None,
-            "unk_token": unk_token,
-            "pad_token": pad_token,
-            "word_delimiter_token": word_delimiter_token,
-        }
-
-    # 5. Now we can instantiate the feature extractor, tokenizer and model
-    # Note for distributed training, the .from_pretrained methods guarantee that only
-    # one local process can concurrently download model & vocab.
-
-    # load feature_extractor and tokenizer
-    tokenizer = AutoTokenizer.from_pretrained(
-        tokenizer_name_or_path,
-        use_auth_token=data_args.use_auth_token,
-        **tokenizer_kwargs,
-    )
-    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
-    )
-
-    # 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()
-
-    # 6. Now we preprocess the datasets including loading the audio, resampling and normalization
-    # Thankfully, `datasets` takes care of automatically loading and resampling the audio,
-    # so that we just need to set the correct target sampling rate and normalize the input
-    # via the `feature_extractor`
-
-    # make sure that dataset decodes audio with correct sampling rate
-    dataset_sampling_rate = next(iter(raw_datasets.values())).features[data_args.audio_column_name].sampling_rate
-    if dataset_sampling_rate != feature_extractor.sampling_rate:
-        raw_datasets = raw_datasets.cast_column(
-            data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
-        )
-
-    # derive max & min input length for sample rate & max duration
-    max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
-    min_input_length = data_args.min_duration_in_seconds * feature_extractor.sampling_rate
-    audio_column_name = data_args.audio_column_name
-    num_workers = data_args.preprocessing_num_workers
-
-    # `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
-
-    with training_args.main_process_first(desc="dataset map preprocessing"):
-        vectorized_datasets = raw_datasets.map(
-            prepare_dataset,
-            remove_columns=next(iter(raw_datasets.values())).column_names,
-            num_proc=num_workers,
-            desc="preprocess datasets",
-        )
-
-        def is_audio_in_length_range(length):
-            return length > min_input_length and length < max_input_length
-
-        # filter data that is shorter than min_input_length
-        vectorized_datasets = vectorized_datasets.filter(
-            is_audio_in_length_range,
-            num_proc=num_workers,
-            input_columns=["input_length"],
-        )
-
-    # 7. 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}
-
-    # for large datasets it is advised to run the preprocessing on a
-    # single machine first with ``args.preprocessing_only`` since there will mostly likely
-    # be a timeout when running the script in distributed mode.
-    # In a second step ``args.preprocessing_only`` can then be set to `False` to load the
-    # cached dataset
-    if data_args.preprocessing_only:
-        logger.info(f"Data preprocessing finished. Files cached at {vectorized_datasets.cache_files}")
-        return
-
-    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
-    data_collator = DataCollatorCTCWithPadding(processor=processor)
-
-    decay_parameters = get_parameter_names(model, [torch.nn.LayerNorm])
-    decay_parameters = [name for name in decay_parameters if "bias" not in name]
-    optimizer_grouped_parameters = [
-        {
-            "params": [p for n, p in model.named_parameters() if n in decay_parameters],
-            "weight_decay": training_args.weight_decay,
-        },
-        {
-            "params": [p for n, p in model.named_parameters() if n not in decay_parameters],
-            "weight_decay": 0.0,
-        },
-    ]
-    optimizer = bnb.optim.Adam8bit(
-        params=optimizer_grouped_parameters,
-        betas=(training_args.adam_beta1, training_args.adam_beta2),
-        eps=training_args.adam_epsilon,
-    )
-
-    optimizers = (optimizer, None)
-
-    # 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=feature_extractor,
-        optimizers=optimizers,
-    )
-
-    # 8. 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
-        max_train_samples = (
-            data_args.max_train_samples
-            if data_args.max_train_samples is not None
-            else len(vectorized_datasets["train"])
-        )
-        metrics["train_samples"] = min(max_train_samples, len(vectorized_datasets["train"]))
-
-        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()
-        max_eval_samples = (
-            data_args.max_eval_samples if data_args.max_eval_samples is not None else len(vectorized_datasets["eval"])
-        )
-        metrics["eval_samples"] = min(max_eval_samples, len(vectorized_datasets["eval"]))
-
-        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, "robust-speech-event"],
-        "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()

special_tokens_map.json

@@ -1 +0,0 @@
-{"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}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"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

@@ -1 +0,0 @@
-{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer"}

train_results.json

@@ -1,8 +0,0 @@
-{
-    "epoch": 60.0,
-    "train_loss": 2.2000570346147588,
-    "train_runtime": 38994.8678,
-    "train_samples": 3336,
-    "train_samples_per_second": 5.133,
-    "train_steps_per_second": 0.16
-}

trainer_state.json

@@ -1,451 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 59.99760191846523,
-  "global_step": 6240,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.96,
-      "learning_rate": 4.8e-05,
-      "loss": 6.3756,
-      "step": 100
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 9.800000000000001e-05,
-      "loss": 1.5856,
-      "step": 200
-    },
-    {
-      "epoch": 2.88,
-      "learning_rate": 0.000148,
-      "loss": 1.5323,
-      "step": 300
-    },
-    {
-      "epoch": 3.84,
-      "learning_rate": 0.00019800000000000002,
-      "loss": 1.5582,
-      "step": 400
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 0.000248,
-      "loss": 1.6612,
-      "step": 500
-    },
-    {
-      "epoch": 5.77,
-      "learning_rate": 0.000298,
-      "loss": 1.6766,
-      "step": 600
-    },
-    {
-      "epoch": 6.73,
-      "learning_rate": 0.000348,
-      "loss": 1.7327,
-      "step": 700
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 0.000398,
-      "loss": 1.7573,
-      "step": 800
-    },
-    {
-      "epoch": 8.65,
-      "learning_rate": 0.000448,
-      "loss": 1.7926,
-      "step": 900
-    },
-    {
-      "epoch": 9.61,
-      "learning_rate": 0.000498,
-      "loss": 1.8141,
-      "step": 1000
-    },
-    {
-      "epoch": 9.61,
-      "eval_loss": 0.7609468102455139,
-      "eval_runtime": 79.8366,
-      "eval_samples_per_second": 13.928,
-      "eval_steps_per_second": 3.482,
-      "eval_wer": 0.7777578475336323,
-      "step": 1000
-    },
-    {
-      "epoch": 10.58,
-      "learning_rate": 0.0005480000000000001,
-      "loss": 1.8648,
-      "step": 1100
-    },
-    {
-      "epoch": 11.54,
-      "learning_rate": 0.000598,
-      "loss": 1.9241,
-      "step": 1200
-    },
-    {
-      "epoch": 12.5,
-      "learning_rate": 0.000648,
-      "loss": 1.9235,
-      "step": 1300
-    },
-    {
-      "epoch": 13.46,
-      "learning_rate": 0.0006979999999999999,
-      "loss": 1.9089,
-      "step": 1400
-    },
-    {
-      "epoch": 14.42,
-      "learning_rate": 0.000748,
-      "loss": 1.9364,
-      "step": 1500
-    },
-    {
-      "epoch": 15.38,
-      "learning_rate": 0.0007980000000000001,
-      "loss": 2.0406,
-      "step": 1600
-    },
-    {
-      "epoch": 16.35,
-      "learning_rate": 0.000848,
-      "loss": 2.1109,
-      "step": 1700
-    },
-    {
-      "epoch": 17.31,
-      "learning_rate": 0.000898,
-      "loss": 2.2393,
-      "step": 1800
-    },
-    {
-      "epoch": 18.27,
-      "learning_rate": 0.000948,
-      "loss": 2.2755,
-      "step": 1900
-    },
-    {
-      "epoch": 19.23,
-      "learning_rate": 0.000998,
-      "loss": 2.2935,
-      "step": 2000
-    },
-    {
-      "epoch": 19.23,
-      "eval_loss": 1.1848976612091064,
-      "eval_runtime": 80.3457,
-      "eval_samples_per_second": 13.84,
-      "eval_steps_per_second": 3.46,
-      "eval_wer": 0.9657399103139014,
-      "step": 2000
-    },
-    {
-      "epoch": 20.19,
-      "learning_rate": 0.0009773584905660377,
-      "loss": 2.3135,
-      "step": 2100
-    },
-    {
-      "epoch": 21.15,
-      "learning_rate": 0.0009537735849056604,
-      "loss": 2.2768,
-      "step": 2200
-    },
-    {
-      "epoch": 22.12,
-      "learning_rate": 0.000930188679245283,
-      "loss": 2.2858,
-      "step": 2300
-    },
-    {
-      "epoch": 23.08,
-      "learning_rate": 0.0009066037735849056,
-      "loss": 2.2804,
-      "step": 2400
-    },
-    {
-      "epoch": 24.04,
-      "learning_rate": 0.0008830188679245283,
-      "loss": 2.2852,
-      "step": 2500
-    },
-    {
-      "epoch": 25.0,
-      "learning_rate": 0.0008594339622641509,
-      "loss": 2.2629,
-      "step": 2600
-    },
-    {
-      "epoch": 25.96,
-      "learning_rate": 0.0008358490566037736,
-      "loss": 2.3056,
-      "step": 2700
-    },
-    {
-      "epoch": 26.92,
-      "learning_rate": 0.0008122641509433963,
-      "loss": 2.2643,
-      "step": 2800
-    },
-    {
-      "epoch": 27.88,
-      "learning_rate": 0.000788679245283019,
-      "loss": 2.3471,
-      "step": 2900
-    },
-    {
-      "epoch": 28.84,
-      "learning_rate": 0.0007650943396226416,
-      "loss": 2.2848,
-      "step": 3000
-    },
-    {
-      "epoch": 28.84,
-      "eval_loss": 1.178926944732666,
-      "eval_runtime": 80.2197,
-      "eval_samples_per_second": 13.862,
-      "eval_steps_per_second": 3.465,
-      "eval_wer": 0.9759641255605381,
-      "step": 3000
-    },
-    {
-      "epoch": 29.81,
-      "learning_rate": 0.0007415094339622641,
-      "loss": 2.3196,
-      "step": 3100
-    },
-    {
-      "epoch": 30.77,
-      "learning_rate": 0.0007179245283018868,
-      "loss": 2.3493,
-      "step": 3200
-    },
-    {
-      "epoch": 31.73,
-      "learning_rate": 0.0006943396226415094,
-      "loss": 2.4006,
-      "step": 3300
-    },
-    {
-      "epoch": 32.69,
-      "learning_rate": 0.0006707547169811321,
-      "loss": 2.3034,
-      "step": 3400
-    },
-    {
-      "epoch": 33.65,
-      "learning_rate": 0.0006474056603773585,
-      "loss": 2.3289,
-      "step": 3500
-    },
-    {
-      "epoch": 34.61,
-      "learning_rate": 0.0006240566037735848,
-      "loss": 2.2975,
-      "step": 3600
-    },
-    {
-      "epoch": 35.58,
-      "learning_rate": 0.0006004716981132075,
-      "loss": 2.2996,
-      "step": 3700
-    },
-    {
-      "epoch": 36.54,
-      "learning_rate": 0.0005768867924528301,
-      "loss": 2.327,
-      "step": 3800
-    },
-    {
-      "epoch": 37.5,
-      "learning_rate": 0.0005533018867924528,
-      "loss": 2.3041,
-      "step": 3900
-    },
-    {
-      "epoch": 38.46,
-      "learning_rate": 0.0005297169811320755,
-      "loss": 2.3325,
-      "step": 4000
-    },
-    {
-      "epoch": 38.46,
-      "eval_loss": 1.087767481803894,
-      "eval_runtime": 79.9787,
-      "eval_samples_per_second": 13.904,
-      "eval_steps_per_second": 3.476,
-      "eval_wer": 0.9594618834080717,
-      "step": 4000
-    },
-    {
-      "epoch": 39.42,
-      "learning_rate": 0.000506367924528302,
-      "loss": 2.3006,
-      "step": 4100
-    },
-    {
-      "epoch": 40.38,
-      "learning_rate": 0.0004827830188679245,
-      "loss": 2.3251,
-      "step": 4200
-    },
-    {
-      "epoch": 41.35,
-      "learning_rate": 0.0004591981132075472,
-      "loss": 2.2754,
-      "step": 4300
-    },
-    {
-      "epoch": 42.31,
-      "learning_rate": 0.00043561320754716983,
-      "loss": 2.2903,
-      "step": 4400
-    },
-    {
-      "epoch": 43.27,
-      "learning_rate": 0.00041202830188679247,
-      "loss": 2.2868,
-      "step": 4500
-    },
-    {
-      "epoch": 44.23,
-      "learning_rate": 0.0003884433962264151,
-      "loss": 2.2541,
-      "step": 4600
-    },
-    {
-      "epoch": 45.19,
-      "learning_rate": 0.00036485849056603774,
-      "loss": 2.2362,
-      "step": 4700
-    },
-    {
-      "epoch": 46.15,
-      "learning_rate": 0.00034127358490566043,
-      "loss": 2.2357,
-      "step": 4800
-    },
-    {
-      "epoch": 47.12,
-      "learning_rate": 0.00031768867924528306,
-      "loss": 2.2935,
-      "step": 4900
-    },
-    {
-      "epoch": 48.08,
-      "learning_rate": 0.00029410377358490565,
-      "loss": 2.25,
-      "step": 5000
-    },
-    {
-      "epoch": 48.08,
-      "eval_loss": 1.088808536529541,
-      "eval_runtime": 80.2673,
-      "eval_samples_per_second": 13.854,
-      "eval_steps_per_second": 3.463,
-      "eval_wer": 0.9585650224215246,
-      "step": 5000
-    },
-    {
-      "epoch": 49.04,
-      "learning_rate": 0.0002705188679245283,
-      "loss": 2.2023,
-      "step": 5100
-    },
-    {
-      "epoch": 50.0,
-      "learning_rate": 0.0002469339622641509,
-      "loss": 2.2132,
-      "step": 5200
-    },
-    {
-      "epoch": 50.96,
-      "learning_rate": 0.00022334905660377358,
-      "loss": 2.1842,
-      "step": 5300
-    },
-    {
-      "epoch": 51.92,
-      "learning_rate": 0.00019976415094339622,
-      "loss": 2.1644,
-      "step": 5400
-    },
-    {
-      "epoch": 52.88,
-      "learning_rate": 0.00017617924528301888,
-      "loss": 2.1449,
-      "step": 5500
-    },
-    {
-      "epoch": 53.84,
-      "learning_rate": 0.00015259433962264152,
-      "loss": 2.1539,
-      "step": 5600
-    },
-    {
-      "epoch": 54.81,
-      "learning_rate": 0.00012900943396226418,
-      "loss": 2.1193,
-      "step": 5700
-    },
-    {
-      "epoch": 55.77,
-      "learning_rate": 0.0001054245283018868,
-      "loss": 2.091,
-      "step": 5800
-    },
-    {
-      "epoch": 56.73,
-      "learning_rate": 8.183962264150944e-05,
-      "loss": 2.0871,
-      "step": 5900
-    },
-    {
-      "epoch": 57.69,
-      "learning_rate": 5.825471698113208e-05,
-      "loss": 2.0757,
-      "step": 6000
-    },
-    {
-      "epoch": 57.69,
-      "eval_loss": 0.950368344783783,
-      "eval_runtime": 79.8206,
-      "eval_samples_per_second": 13.931,
-      "eval_steps_per_second": 3.483,
-      "eval_wer": 0.9411659192825113,
-      "step": 6000
-    },
-    {
-      "epoch": 58.65,
-      "learning_rate": 3.4669811320754716e-05,
-      "loss": 2.056,
-      "step": 6100
-    },
-    {
-      "epoch": 59.61,
-      "learning_rate": 1.1084905660377358e-05,
-      "loss": 2.0488,
-      "step": 6200
-    },
-    {
-      "epoch": 60.0,
-      "step": 6240,
-      "total_flos": 1.9322773224714512e+20,
-      "train_loss": 2.2000570346147588,
-      "train_runtime": 38994.8678,
-      "train_samples_per_second": 5.133,
-      "train_steps_per_second": 0.16
-    }
-  ],
-  "max_steps": 6240,
-  "num_train_epochs": 60,
-  "total_flos": 1.9322773224714512e+20,
-  "trial_name": null,
-  "trial_params": null
-}

training_args.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:e671975c9fc29d251830e4bd90003705e709ea3d562726308fb5fc129d6750fe
-size 2991

vocab.json

@@ -1 +0,0 @@
-{"\\": 1, "_": 2, "e": 3, "g": 4, "l": 5, "n": 6, "p": 7, "r": 8, "s": 9, "t": 10, "ఁ": 11, "ం": 12, "ః": 13, "అ": 14, "ఆ": 15, "ఇ": 16, "ఈ": 17, "ఉ": 18, "ఊ": 19, "ఋ": 20, "ఎ": 21, "ఏ": 22, "ఐ": 23, "ఒ": 24, "ఓ": 25, "ఔ": 26, "క": 27, "ఖ": 28, "గ": 29, "ఘ": 30, "ఙ": 31, "చ": 32, "ఛ": 33, "జ": 34, "ఞ": 35, "ట": 36, "ఠ": 37, "డ": 38, "ఢ": 39, "ణ": 40, "త": 41, "థ": 42, "ద": 43, "ధ": 44, "న": 45, "ప": 46, "ఫ": 47, "బ": 48, "భ": 49, "మ": 50, "య": 51, "ర": 52, "ఱ": 53, "ల": 54, "ళ": 55, "వ": 56, "శ": 57, "ష": 58, "స": 59, "హ": 60, "ా": 61, "ి": 62, "ీ": 63, "ు": 64, "ూ": 65, "ృ": 66, "ె": 67, "ే": 68, "ై": 69, "ొ": 70, "ో": 71, "ౌ": 72, "్": 73, "‌": 74, "|": 0, "[UNK]": 75, "[PAD]": 76}