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

Merge pull request #3276 from lissyx/pr3256

Browse Source 
Pr3256
This commit is contained in:
lissyx 2020-08-25 22:39:50 +02:00 committed by GitHub
commit 9daa708047
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
5 changed files with 34 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
taskcluster/.build.yml
View File

@ -26,7 +26,7 @@ build:
nc_asset_name: 'native_client.tar.xz' nc_asset_name: 'native_client.tar.xz'
args: args:
tests_cmdline: '' tests_cmdline: ''
tensorflow_git_desc: 'TensorFlow: v2.3.0-4-g4336a5b' tensorflow_git_desc: 'TensorFlow: v2.3.0-6-g23ad988'
test_model_task: '' test_model_task: ''
homebrew: homebrew:
url: '' url: ''

44
taskcluster/.shared.yml
View File

@ -140,38 +140,38 @@ system:
namespace: "project.mozilla-voice-stt.swig.win.amd64.1a4c14945012f1282c2eddc174fb7674d5295de8.0" namespace: "project.mozilla-voice-stt.swig.win.amd64.1a4c14945012f1282c2eddc174fb7674d5295de8.0"
tensorflow: tensorflow:
linux_amd64_cpu: linux_amd64_cpu:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.cpu/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.cpu/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.cpu" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.cpu"
linux_amd64_cuda: linux_amd64_cuda:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.cuda/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.cuda/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.cuda" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.cuda"
linux_armv7: linux_armv7:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.arm/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.arm/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.arm" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.arm"
linux_arm64: linux_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.arm64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.arm64/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.arm64" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.arm64"
darwin_amd64: darwin_amd64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.1.osx/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.osx/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.1.osx" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.osx"
android_arm64: android_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.android-arm64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.android-arm64/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.android-arm64" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.android-arm64"
android_armv7: android_armv7:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.android-armv7/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.android-armv7/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.android-armv7" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.android-armv7"
win_amd64_cpu: win_amd64_cpu:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.win/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.win/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.win" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.win"
win_amd64_cuda: win_amd64_cuda:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.win-cuda/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.win-cuda/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.win-cuda" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.win-cuda"
ios_arm64: ios_arm64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.ios_arm64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.ios_arm64/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.ios_arm64" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.ios_arm64"
ios_x86_64: ios_x86_64:
url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.ios_x86_64/artifacts/public/home.tar.xz" url: "https://community-tc.services.mozilla.com/api/index/v1/task/project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.ios_x86_64/artifacts/public/home.tar.xz"
namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.4336a5b49fa6d650e24dbdba55bcef9581535244.0.ios_x86_64" namespace: "project.mozilla-voice-stt.tensorflow.pip.r2.3.23ad988fcde60fb01f9533e95004bbc4877a9143.0.ios_x86_64"
username: 'build-user' username: 'build-user'
homedir: homedir:
linux: '/home/build-user' linux: '/home/build-user'

2
tensorflow

@ -1 +1 @@
Subproject commit 4336a5b49fa6d650e24dbdba55bcef9581535244 Subproject commit 23ad988fcde60fb01f9533e95004bbc4877a9143

14
training/mozilla_voice_stt_training/train.py
View File

@ -74,7 +74,7 @@ def create_overlapping_windows(batch_x):
return batch_x return batch_x
def dense(name, x, units, dropout_rate=None, relu=True): def dense(name, x, units, dropout_rate=None, relu=True, layer_norm=False):
with tfv1.variable_scope(name): with tfv1.variable_scope(name):
bias = variable_on_cpu('bias', [units], tfv1.zeros_initializer()) bias = variable_on_cpu('bias', [units], tfv1.zeros_initializer())
weights = variable_on_cpu('weights', [x.shape[-1], units], tfv1.keras.initializers.VarianceScaling(scale=1.0, mode="fan_avg", distribution="uniform")) weights = variable_on_cpu('weights', [x.shape[-1], units], tfv1.keras.initializers.VarianceScaling(scale=1.0, mode="fan_avg", distribution="uniform"))
@ -84,6 +84,10 @@ def dense(name, x, units, dropout_rate=None, relu=True):
if relu: if relu:
output = tf.minimum(tf.nn.relu(output), FLAGS.relu_clip) output = tf.minimum(tf.nn.relu(output), FLAGS.relu_clip)
if layer_norm:
with tfv1.variable_scope(name):
output = tf.contrib.layers.layer_norm(output)
if dropout_rate is not None: if dropout_rate is not None:
output = tf.nn.dropout(output, rate=dropout_rate) output = tf.nn.dropout(output, rate=dropout_rate)
@ -177,9 +181,9 @@ def create_model(batch_x, seq_length, dropout, reuse=False, batch_size=None, pre
# The next three blocks will pass `batch_x` through three hidden layers with # The next three blocks will pass `batch_x` through three hidden layers with
# clipped RELU activation and dropout. # clipped RELU activation and dropout.
layers['layer_1'] = layer_1 = dense('layer_1', batch_x, Config.n_hidden_1, dropout_rate=dropout[0]) layers['layer_1'] = layer_1 = dense('layer_1', batch_x, Config.n_hidden_1, dropout_rate=dropout[0], layer_norm=FLAGS.layer_norm)
layers['layer_2'] = layer_2 = dense('layer_2', layer_1, Config.n_hidden_2, dropout_rate=dropout[1]) layers['layer_2'] = layer_2 = dense('layer_2', layer_1, Config.n_hidden_2, dropout_rate=dropout[1], layer_norm=FLAGS.layer_norm)
layers['layer_3'] = layer_3 = dense('layer_3', layer_2, Config.n_hidden_3, dropout_rate=dropout[2]) layers['layer_3'] = layer_3 = dense('layer_3', layer_2, Config.n_hidden_3, dropout_rate=dropout[2], layer_norm=FLAGS.layer_norm)
# `layer_3` is now reshaped into `[n_steps, batch_size, 2*n_cell_dim]`, # `layer_3` is now reshaped into `[n_steps, batch_size, 2*n_cell_dim]`,
# as the LSTM RNN expects its input to be of shape `[max_time, batch_size, input_size]`. # as the LSTM RNN expects its input to be of shape `[max_time, batch_size, input_size]`.
@ -196,7 +200,7 @@ def create_model(batch_x, seq_length, dropout, reuse=False, batch_size=None, pre
layers['rnn_output_state'] = output_state layers['rnn_output_state'] = output_state
# Now we feed `output` to the fifth hidden layer with clipped RELU activation # Now we feed `output` to the fifth hidden layer with clipped RELU activation
layers['layer_5'] = layer_5 = dense('layer_5', output, Config.n_hidden_5, dropout_rate=dropout[5]) layers['layer_5'] = layer_5 = dense('layer_5', output, Config.n_hidden_5, dropout_rate=dropout[5], layer_norm=FLAGS.layer_norm)
# Now we apply a final linear layer creating `n_classes` dimensional vectors, the logits. # Now we apply a final linear layer creating `n_classes` dimensional vectors, the logits.
layers['layer_6'] = layer_6 = dense('layer_6', layer_5, Config.n_hidden_6, relu=False) layers['layer_6'] = layer_6 = dense('layer_6', layer_5, Config.n_hidden_6, relu=False)

1
training/mozilla_voice_stt_training/util/flags.py
View File

@ -135,6 +135,7 @@ def create_flags():
# Geometry # Geometry
f.DEFINE_integer('n_hidden', 2048, 'layer width to use when initialising layers') f.DEFINE_integer('n_hidden', 2048, 'layer width to use when initialising layers')
f.DEFINE_boolean('layer_norm', False, 'wether to use layer-normalization after each fully-connected layer (except the last one)')
# Initialization # Initialization