Hui Zhang
5 years ago
parent
dd3fb069a3
commit
6cc80c0aff
@ -1,2 +1,4 @@
|
||||
.DS_Store
|
||||
*.pyc
|
||||
tools/venv
|
||||
dataset
|
||||
|
||||
@ -0,0 +1,70 @@
|
||||
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
from yacs.config import CfgNode as CN
|
||||
|
||||
_C = CN()
|
||||
_C.data = CN(
|
||||
dict(
|
||||
train_manifest="",
|
||||
dev_manifest="",
|
||||
test_manifest="",
|
||||
vocab_filepath="",
|
||||
mean_std_filepath="",
|
||||
augmentation_config='{}',
|
||||
max_duration=float('inf'),
|
||||
min_duration=0.0,
|
||||
stride_ms=10.0, # ms
|
||||
window_ms=20.0, # ms
|
||||
n_fft=None, # fft points
|
||||
max_freq=None, # None for samplerate/2
|
||||
specgram_type='linear', # 'linear', 'mfcc'
|
||||
target_sample_rate=16000, # sample rate
|
||||
use_dB_normalization=True,
|
||||
target_dB=-20,
|
||||
random_seed=0,
|
||||
keep_transcription_text=False,
|
||||
batch_size=32, # batch size
|
||||
num_workers=0, # data loader workers
|
||||
sortagrad=False, # sorted in first epoch when True
|
||||
shuffle_method="batch_shuffle", # 'batch_shuffle', 'instance_shuffle'
|
||||
))
|
||||
|
||||
_C.model = CN(
|
||||
dict(
|
||||
num_conv_layers=2, #Number of stacking convolution layers.
|
||||
num_rnn_layers=3, #Number of stacking RNN layers.
|
||||
rnn_layer_size=1024, #RNN layer size (number of RNN cells).
|
||||
use_gru=True, #Use gru if set True. Use simple rnn if set False.
|
||||
share_rnn_weights=False #Whether to share input-hidden weights between forward and backward directional RNNs.Notice that for GRU, weight sharing is not supported.
|
||||
))
|
||||
|
||||
_C.training = CN(
|
||||
dict(
|
||||
lr=5e-4, # learning rate
|
||||
weight_decay=1e-6, # the coeff of weight decay
|
||||
global_grad_clip=400.0, # the global norm clip
|
||||
plot_interval=1000, # plot attention and spectrogram by step
|
||||
valid_interval=1000, # validation by step
|
||||
save_interval=1000, # checkpoint by step
|
||||
max_iteration=500000, # max iteration to train by step
|
||||
n_epoch=50, # train epochs
|
||||
))
|
||||
|
||||
|
||||
def get_cfg_defaults():
|
||||
"""Get a yacs CfgNode object with default values for my_project."""
|
||||
# Return a clone so that the defaults will not be altered
|
||||
# This is for the "local variable" use pattern
|
||||
return _C.clone()
|
||||
@ -0,0 +1,64 @@
|
||||
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
import argparse
|
||||
|
||||
|
||||
def default_argument_parser():
|
||||
r"""A simple yet genral argument parser for experiments with parakeet.
|
||||
|
||||
This is used in examples with parakeet. And it is intended to be used by
|
||||
other experiments with parakeet. It requires a minimal set of command line
|
||||
arguments to start a training script.
|
||||
|
||||
The ``--config`` and ``--opts`` are used for overwrite the deault
|
||||
configuration.
|
||||
|
||||
The ``--data`` and ``--output`` specifies the data path and output path.
|
||||
Resuming training from existing progress at the output directory is the
|
||||
intended default behavior.
|
||||
|
||||
The ``--checkpoint_path`` specifies the checkpoint to load from.
|
||||
|
||||
The ``--device`` and ``--nprocs`` specifies how to run the training.
|
||||
|
||||
|
||||
See Also
|
||||
--------
|
||||
parakeet.training.experiment
|
||||
Returns
|
||||
-------
|
||||
argparse.ArgumentParser
|
||||
the parser
|
||||
"""
|
||||
parser = argparse.ArgumentParser()
|
||||
|
||||
# yapf: disable
|
||||
# data and output
|
||||
parser.add_argument("--config", metavar="FILE", help="path of the config file to overwrite to default config with.")
|
||||
parser.add_argument("--data", metavar="DATA_DIR", help="path to the datatset.")
|
||||
parser.add_argument("--output", metavar="OUTPUT_DIR", help="path to save checkpoint and logs.")
|
||||
|
||||
# load from saved checkpoint
|
||||
parser.add_argument("--checkpoint_path", type=str, help="path of the checkpoint to load")
|
||||
|
||||
# running
|
||||
parser.add_argument("--device", type=str, choices=["cpu", "gpu"], help="device type to use, cpu and gpu are supported.")
|
||||
parser.add_argument("--nprocs", type=int, default=1, help="number of parallel processes to use.")
|
||||
|
||||
# overwrite extra config and default config
|
||||
parser.add_argument("--opts", nargs=argparse.REMAINDER, help="options to overwrite --config file and the default config, passing in KEY VALUE pairs")
|
||||
# yapd: enable
|
||||
|
||||
return parser
|
||||
Loading…
Reference in new issue