u107/coqui

coqui-ai

Models

• v51
  • language: uk
  • Epoch: 9
  • Loss: 5.813825
  • Test: WER: 0.163878, CER: 0.033506, loss: 4.564916
• v50
  • language: uk-RU
  • Epoch: 9
  • Loss: 8.292612
• v49
  • language: uk-RU
  • Epoch: 8
  • Loss: 30.810781
  • Test: WER: 0.400063, CER: 0.189420, loss: 53.694748
• v48
  • language: uk
  • Epoch: 25
  • Loss: 26.365822
  • Test: WER: 0.239203, CER: 0.123846, loss: 32.882496

Використання:

docker run --gpus all --ulimit memlock=-1 --ulimit stack=67108864 --shm-size=18gb -it --rm -v ${PWD}/ds:/ds -v ${PWD}/../../datasets:/datasets ghcr.io/coqui-ai/stt-train:latest

Приклад тренування:

#!/usr/bin/env python
import os
from coqui_stt_training.util.config import initialize_globals_from_args
from coqui_stt_training.train import train
from coqui_stt_training.evaluate import test

# only one GPU for only one training sample
# os.environ["CUDA_VISIBLE_DEVICES"] = "0"
# os.environ["TF_ENABLE_DEPRECATION_WARNINGS"] = "1"

# TF_ENABLE_DEPRECATION_WARNINGS=1 python /ds/train_uk.py
# CUDA_VISIBLE_DEVICES=0 python /ds/train_uk.py

initialize_globals_from_args(
    # drop_source_layers=1,
    # load_train='last',
    use_allow_growth=True,
    bytes_output_mode=True,
    force_initialize_learning_rate=False,

    # load_train="auto",
    # load_evaluate="auto",
    #
    load_cudnn=False,
    train_cudnn=True,
    n_hidden=1024,
    epochs=5,
    learning_rate=0.01,
    dropout_rate=0.5,

    # shuffle_batches=True,

    show_progressbar=True,
    early_stop=False,

    # alphabet_config_path="/ds/alphabet.txt",
    train_files=['/ds/train_uk.sdb'],
    train_batch_size=32,
    dev_files=['/ds/dev_uk.sdb'],
    dev_batch_size=32,
    test_files=['/ds/test_uk.sdb'],
    test_batch_size=12,

    feature_cache='/ds/tmp/feature_cache_sdb',

    checkpoint_dir='/ds/ckpt_sdb',
    # load_checkpoint_dir='/ds/coqui-stt-1.4.0-checkpoint',
    # save_checkpoint_dir='/ds/ckpt_sdb',

    scorer_path='',

    export_dir='/ds/export_train_sdb',
    export_tflite=False,
    export_language='Ukraine (uk-UA)',
    export_author_id="Vadim",
    export_model_name="ds_uk",
    export_model_version="0.0.1",
    # audio_sample_rate=16000,
    # max_to_keep=5,
    # plateau_epochs=3,
    # checkpoint_secs=1200,
    # use_allow_growth=True,
)

train()

test()