experiments/STT
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request #2038 from mozilla/split-dev-test-epochs

Browse Source 
Perform separate validation and test epochs per dataset when multiple files are specified (Fixes #1634 and #2043)
This commit is contained in:
Reuben Morais 2019-04-16 15:23:16 +00:00 committed by GitHub
commit 1e601d5c4a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 126 additions and 92 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
DeepSpeech.py
View File

@ -14,6 +14,7 @@ import progressbar
import shutil import shutil
import tensorflow as tf import tensorflow as tf
from datetime import datetime
from ds_ctcdecoder import ctc_beam_search_decoder, Scorer from ds_ctcdecoder import ctc_beam_search_decoder, Scorer
from evaluate import evaluate from evaluate import evaluate
from six.moves import zip, range from six.moves import zip, range
@ -21,7 +22,7 @@ from tensorflow.python.tools import freeze_graph
from util.config import Config, initialize_globals from util.config import Config, initialize_globals
from util.feeding import create_dataset, samples_to_mfccs, audiofile_to_features from util.feeding import create_dataset, samples_to_mfccs, audiofile_to_features
from util.flags import create_flags, FLAGS from util.flags import create_flags, FLAGS
from util.logging import log_info, log_error, log_debug from util.logging import log_info, log_error, log_debug, log_progress, create_progressbar
# Graph Creation # Graph Creation
@ -366,7 +367,7 @@ def train():
# Create training and validation datasets # Create training and validation datasets
train_set = create_dataset(FLAGS.train_files.split(','), train_set = create_dataset(FLAGS.train_files.split(','),
batch_size=FLAGS.train_batch_size, batch_size=FLAGS.train_batch_size,
cache_path=FLAGS.train_cached_features_path) cache_path=FLAGS.feature_cache)
iterator = tf.data.Iterator.from_structure(train_set.output_types, iterator = tf.data.Iterator.from_structure(train_set.output_types,
train_set.output_shapes, train_set.output_shapes,
@ -376,10 +377,9 @@ def train():
train_init_op = iterator.make_initializer(train_set) train_init_op = iterator.make_initializer(train_set)
if FLAGS.dev_files: if FLAGS.dev_files:
dev_set = create_dataset(FLAGS.dev_files.split(','), dev_csvs = FLAGS.dev_files.split(',')
batch_size=FLAGS.dev_batch_size, dev_sets = [create_dataset([csv], batch_size=FLAGS.dev_batch_size) for csv in dev_csvs]
cache_path=FLAGS.dev_cached_features_path) dev_init_ops = [iterator.make_initializer(dev_set) for dev_set in dev_sets]
dev_init_op = iterator.make_initializer(dev_set)
# Dropout # Dropout
dropout_rates = [tf.placeholder(tf.float32, name='dropout_{}'.format(i)) for i in range(6)] dropout_rates = [tf.placeholder(tf.float32, name='dropout_{}'.format(i)) for i in range(6)]
@ -445,7 +445,7 @@ def train():
' - consider using load option "auto" or "init".' % FLAGS.load) ' - consider using load option "auto" or "init".' % FLAGS.load)
sys.exit(1) sys.exit(1)
def run_set(set_name, init_op): def run_set(set_name, epoch, init_op, dataset=None):
is_train = set_name == 'train' is_train = set_name == 'train'
train_op = apply_gradient_op if is_train else [] train_op = apply_gradient_op if is_train else []
feed_dict = dropout_feed_dict if is_train else no_dropout_feed_dict feed_dict = dropout_feed_dict if is_train else no_dropout_feed_dict
@ -456,6 +456,7 @@ def train():
step_summary_writer = step_summary_writers.get(set_name) step_summary_writer = step_summary_writers.get(set_name)
checkpoint_time = time.time() checkpoint_time = time.time()
# Setup progress bar
class LossWidget(progressbar.widgets.FormatLabel): class LossWidget(progressbar.widgets.FormatLabel):
def __init__(self): def __init__(self):
progressbar.widgets.FormatLabel.__init__(self, format='Loss: %(mean_loss)f') progressbar.widgets.FormatLabel.__init__(self, format='Loss: %(mean_loss)f')
@ -464,12 +465,12 @@ def train():
data['mean_loss'] = total_loss / step_count if step_count else 0.0 data['mean_loss'] = total_loss / step_count if step_count else 0.0
return progressbar.widgets.FormatLabel.__call__(self, progress, data, **kwargs) return progressbar.widgets.FormatLabel.__call__(self, progress, data, **kwargs)
if FLAGS.show_progressbar: prefix = 'Epoch {} | {:>10}'.format(epoch, 'Training' if is_train else 'Validation')
pbar = progressbar.ProgressBar(widgets=['Epoch {}'.format(epoch), widgets = [' | ', progressbar.widgets.Timer(),
' | ', progressbar.widgets.Timer(), ' | Steps: ', progressbar.widgets.Counter(),
' | Steps: ', progressbar.widgets.Counter(), ' | ', LossWidget()]
' | ', LossWidget()]) suffix = ' | Dataset: {}'.format(dataset) if dataset else None
pbar.start() pbar = create_progressbar(prefix=prefix, widgets=widgets, suffix=suffix).start()
# Initialize iterator to the appropriate dataset # Initialize iterator to the appropriate dataset
session.run(init_op) session.run(init_op)
@ -486,8 +487,7 @@ def train():
total_loss += batch_loss total_loss += batch_loss
step_count += 1 step_count += 1
if FLAGS.show_progressbar: pbar.update(step_count)
pbar.update(step_count)
step_summary_writer.add_summary(step_summary, current_step) step_summary_writer.add_summary(step_summary, current_step)
@ -495,31 +495,34 @@ def train():
checkpoint_saver.save(session, checkpoint_path, global_step=current_step) checkpoint_saver.save(session, checkpoint_path, global_step=current_step)
checkpoint_time = time.time() checkpoint_time = time.time()
if FLAGS.show_progressbar: pbar.finish()
pbar.finish() mean_loss = total_loss / step_count if step_count > 0 else 0.0
return mean_loss, step_count
return total_loss / step_count
log_info('STARTING Optimization') log_info('STARTING Optimization')
train_start_time = datetime.utcnow()
best_dev_loss = float('inf') best_dev_loss = float('inf')
dev_losses = [] dev_losses = []
try: try:
for epoch in range(FLAGS.epochs): for epoch in range(FLAGS.epochs):
# Training # Training
if not FLAGS.show_progressbar: log_progress('Training epoch %d...' % epoch)
log_info('Training epoch %d...' % epoch) train_loss, _ = run_set('train', epoch, train_init_op)
train_loss = run_set('train', train_init_op) log_progress('Finished training epoch %d - loss: %f' % (epoch, train_loss))
if not FLAGS.show_progressbar:
log_info('Finished training epoch %d - loss: %f' % (epoch, train_loss))
checkpoint_saver.save(session, checkpoint_path, global_step=global_step) checkpoint_saver.save(session, checkpoint_path, global_step=global_step)
if FLAGS.dev_files: if FLAGS.dev_files:
# Validation # Validation
if not FLAGS.show_progressbar: dev_loss = 0.0
log_info('Validating epoch %d...' % epoch) total_steps = 0
dev_loss = run_set('dev', dev_init_op) for csv, init_op in zip(dev_csvs, dev_init_ops):
if not FLAGS.show_progressbar: log_progress('Validating epoch %d on %s...' % (epoch, csv))
log_info('Finished validating epoch %d - loss: %f' % (epoch, dev_loss)) set_loss, steps = run_set('dev', epoch, init_op, dataset=csv)
dev_loss += set_loss * steps
total_steps += steps
log_progress('Finished validating epoch %d on %s - loss: %f' % (epoch, csv, set_loss))
dev_loss = dev_loss / total_steps
dev_losses.append(dev_loss) dev_losses.append(dev_loss)
if dev_loss < best_dev_loss: if dev_loss < best_dev_loss:
@ -543,6 +546,7 @@ def train():
break break
except KeyboardInterrupt: except KeyboardInterrupt:
pass pass
log_info('FINISHED optimization in {}'.format(datetime.utcnow() - train_start_time))
log_debug('Session closed.') log_debug('Session closed.')

2
bin/run-tc-ldc93s1_new.sh
View File

@ -14,7 +14,7 @@ fi;
python -u DeepSpeech.py --noshow_progressbar --noearly_stop \ python -u DeepSpeech.py --noshow_progressbar --noearly_stop \
--train_files ${ldc93s1_csv} --train_batch_size 1 \ --train_files ${ldc93s1_csv} --train_batch_size 1 \
--train_cached_features_path '/tmp/ldc93s1_cache' \ --feature_cache '/tmp/ldc93s1_cache' \
--dev_files ${ldc93s1_csv} --dev_batch_size 1 \ --dev_files ${ldc93s1_csv} --dev_batch_size 1 \
--test_files ${ldc93s1_csv} --test_batch_size 1 \ --test_files ${ldc93s1_csv} --test_batch_size 1 \
--n_hidden 100 --epochs $epoch_count \ --n_hidden 100 --epochs $epoch_count \

129
evaluate.py
View File

@ -18,7 +18,7 @@ from util.config import Config, initialize_globals
from util.evaluate_tools import calculate_report from util.evaluate_tools import calculate_report
from util.feeding import create_dataset from util.feeding import create_dataset
from util.flags import create_flags, FLAGS from util.flags import create_flags, FLAGS
from util.logging import log_error from util.logging import log_error, log_progress, create_progressbar
from util.text import levenshtein from util.text import levenshtein
@ -45,12 +45,14 @@ def evaluate(test_csvs, create_model, try_loading):
FLAGS.lm_binary_path, FLAGS.lm_trie_path, FLAGS.lm_binary_path, FLAGS.lm_trie_path,
Config.alphabet) Config.alphabet)
test_set = create_dataset(test_csvs, test_csvs = FLAGS.test_files.split(',')
batch_size=FLAGS.test_batch_size, test_sets = [create_dataset([csv], batch_size=FLAGS.test_batch_size) for csv in test_csvs]
cache_path=FLAGS.test_cached_features_path) iterator = tf.data.Iterator.from_structure(test_sets[0].output_types,
it = test_set.make_one_shot_iterator() test_sets[0].output_shapes,
output_classes=test_sets[0].output_classes)
test_init_ops = [iterator.make_initializer(test_set) for test_set in test_sets]
(batch_x, batch_x_len), batch_y = it.get_next() (batch_x, batch_x_len), batch_y = iterator.get_next()
# One rate per layer # One rate per layer
no_dropout = [None] * 6 no_dropout = [None] * 6
@ -67,10 +69,16 @@ def evaluate(test_csvs, create_model, try_loading):
tf.train.get_or_create_global_step() tf.train.get_or_create_global_step()
with tf.Session(config=Config.session_config) as session: # Get number of accessible CPU cores for this process
# Create a saver using variables from the above newly created graph try:
saver = tf.train.Saver() num_processes = cpu_count()
except NotImplementedError:
num_processes = 1
# Create a saver using variables from the above newly created graph
saver = tf.train.Saver()
with tf.Session(config=Config.session_config) as session:
# Restore variables from training checkpoint # Restore variables from training checkpoint
loaded = try_loading(session, saver, 'best_dev_checkpoint', 'best validation') loaded = try_loading(session, saver, 'best_dev_checkpoint', 'best validation')
if not loaded: if not loaded:
@ -79,70 +87,75 @@ def evaluate(test_csvs, create_model, try_loading):
log_error('Checkpoint directory ({}) does not contain a valid checkpoint state.'.format(FLAGS.checkpoint_dir)) log_error('Checkpoint directory ({}) does not contain a valid checkpoint state.'.format(FLAGS.checkpoint_dir))
exit(1) exit(1)
logitses = [] def run_test(init_op, dataset):
losses = [] logitses = []
seq_lengths = [] losses = []
ground_truths = [] seq_lengths = []
ground_truths = []
print('Computing acoustic model predictions...') bar = create_progressbar(prefix='Computing acoustic model predictions | ',
bar = progressbar.ProgressBar(widgets=['Steps: ', progressbar.Counter(), ' | ', progressbar.Timer()]) widgets=['Steps: ', progressbar.Counter(), ' | ', progressbar.Timer()]).start()
log_progress('Computing acoustic model predictions...')
step_count = 0 step_count = 0
# First pass, compute losses and transposed logits for decoding # Initialize iterator to the appropriate dataset
while True: session.run(init_op)
try:
logits, loss_, lengths, transcripts = session.run([transposed, loss, batch_x_len, batch_y])
except tf.errors.OutOfRangeError:
break
step_count += 1 # First pass, compute losses and transposed logits for decoding
bar.update(step_count) while True:
try:
logits, loss_, lengths, transcripts = session.run([transposed, loss, batch_x_len, batch_y])
except tf.errors.OutOfRangeError:
break
logitses.append(logits) step_count += 1
losses.extend(loss_) bar.update(step_count)
seq_lengths.append(lengths)
ground_truths.extend(sparse_tensor_value_to_texts(transcripts, Config.alphabet))
bar.finish() logitses.append(logits)
losses.extend(loss_)
seq_lengths.append(lengths)
ground_truths.extend(sparse_tensor_value_to_texts(transcripts, Config.alphabet))
predictions = [] bar.finish()
# Get number of accessible CPU cores for this process predictions = []
try:
num_processes = cpu_count()
except NotImplementedError:
num_processes = 1
print('Decoding predictions...') bar = create_progressbar(max_value=step_count,
bar = progressbar.ProgressBar(max_value=step_count, prefix='Decoding predictions | ').start()
widget=progressbar.AdaptiveETA) log_progress('Decoding predictions...')
# Second pass, decode logits and compute WER and edit distance metrics # Second pass, decode logits and compute WER and edit distance metrics
for logits, seq_length in bar(zip(logitses, seq_lengths)): for logits, seq_length in bar(zip(logitses, seq_lengths)):
decoded = ctc_beam_search_decoder_batch(logits, seq_length, Config.alphabet, FLAGS.beam_width, decoded = ctc_beam_search_decoder_batch(logits, seq_length, Config.alphabet, FLAGS.beam_width,
num_processes=num_processes, scorer=scorer) num_processes=num_processes, scorer=scorer)
predictions.extend(d[0][1] for d in decoded) predictions.extend(d[0][1] for d in decoded)
distances = [levenshtein(a, b) for a, b in zip(ground_truths, predictions)] distances = [levenshtein(a, b) for a, b in zip(ground_truths, predictions)]
wer, cer, samples = calculate_report(ground_truths, predictions, distances, losses) wer, cer, samples = calculate_report(ground_truths, predictions, distances, losses)
mean_loss = np.mean(losses) mean_loss = np.mean(losses)
# Take only the first report_count items # Take only the first report_count items
report_samples = itertools.islice(samples, FLAGS.report_count) report_samples = itertools.islice(samples, FLAGS.report_count)
print('Test - WER: %f, CER: %f, loss: %f' % print('Test on %s - WER: %f, CER: %f, loss: %f' %
(wer, cer, mean_loss)) (dataset, wer, cer, mean_loss))
print('-' * 80) print('-' * 80)
for sample in report_samples: for sample in report_samples:
print('WER: %f, CER: %f, loss: %f' % print('WER: %f, CER: %f, loss: %f' %
(sample.wer, sample.distance, sample.loss)) (sample.wer, sample.distance, sample.loss))
print(' - src: "%s"' % sample.src) print(' - src: "%s"' % sample.src)
print(' - res: "%s"' % sample.res) print(' - res: "%s"' % sample.res)
print('-' * 80) print('-' * 80)
return samples return samples
samples = []
for csv, init_op in zip(test_csvs, test_init_ops):
print('Testing model on {}'.format(csv))
samples.extend(run_test(init_op, dataset=csv))
return samples
def main(_): def main(_):

2
util/feeding.py
View File

@ -63,7 +63,7 @@ def to_sparse_tuple(sequence):
return indices, sequence, shape return indices, sequence, shape
def create_dataset(csvs, batch_size, cache_path): def create_dataset(csvs, batch_size, cache_path=''):
df = read_csvs(csvs) df = read_csvs(csvs)
df.sort_values(by='wav_filesize', inplace=True) df.sort_values(by='wav_filesize', inplace=True)

4
util/flags.py
View File

@ -16,9 +16,7 @@ def create_flags():
f.DEFINE_string('dev_files', '', 'comma separated list of files specifying the dataset used for validation. Multiple files will get merged. If empty, validation will not be run.') f.DEFINE_string('dev_files', '', 'comma separated list of files specifying the dataset used for validation. Multiple files will get merged. If empty, validation will not be run.')
f.DEFINE_string('test_files', '', 'comma separated list of files specifying the dataset used for testing. Multiple files will get merged. If empty, the model will not be tested.') f.DEFINE_string('test_files', '', 'comma separated list of files specifying the dataset used for testing. Multiple files will get merged. If empty, the model will not be tested.')
f.DEFINE_string('train_cached_features_path', '', 'comma separated list of files specifying the dataset used for training. multiple files will get merged') f.DEFINE_string('feature_cache', '', 'path where cached features extracted from --train_files will be saved. If empty, caching will be done in memory and no files will be written.')
f.DEFINE_string('dev_cached_features_path', '', 'comma separated list of files specifying the dataset used for validation. multiple files will get merged')
f.DEFINE_string('test_cached_features_path', '', 'comma separated list of files specifying the dataset used for testing. multiple files will get merged')
f.DEFINE_integer('feature_win_len', 32, 'feature extraction audio window length in milliseconds') f.DEFINE_integer('feature_win_len', 32, 'feature extraction audio window length in milliseconds')
f.DEFINE_integer('feature_win_step', 20, 'feature extraction window step length in milliseconds') f.DEFINE_integer('feature_win_step', 20, 'feature extraction window step length in milliseconds')

19
util/logging.py
View File

@ -1,5 +1,8 @@
from __future__ import print_function from __future__ import print_function
import progressbar
import sys
from util.flags import FLAGS from util.flags import FLAGS
@ -28,3 +31,19 @@ def log_warn(message):
def log_error(message): def log_error(message):
if FLAGS.log_level <= 3: if FLAGS.log_level <= 3:
prefix_print('E ', message) prefix_print('E ', message)
def create_progressbar(*args, **kwargs):
# Progress bars in stdout by default
if 'fd' not in kwargs:
kwargs['fd'] = sys.stdout
if FLAGS.show_progressbar:
return progressbar.ProgressBar(*args, **kwargs)
return progressbar.NullBar(*args, **kwargs)
def log_progress(message):
if not FLAGS.show_progressbar:
log_info(message)