def testBuildCudnnGraph(self):
self.config.hparams.use_cudnn = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'train',
self.config,
sequence_example_file_paths=[self._sequence_file.name])()
def testBuildCudnnGraphWithResidualConnections(self):
self.config.hparams.use_cudnn = True
self.config.hparams.residual_connections = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'train', self.config,
sequence_example_file_paths=[self._sequence_file.name])()
def testBuildGraphWithAttention(self):
self.config.hparams.attn_length = 10
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'train',
self.config,
sequence_example_file_paths=[self._sequence_file.name])()
def main(unused_argv):
tf.logging.set_verbosity(mt.FLAGS.log)
data_dir = tgt.OUTPUT_DIR
train_dir = os.path.join(tgt.MODEL_DIR, "logdir/train")
if not os.path.exists(train_dir):
tf.gfile.MakeDirs(train_dir)
config = mt.melody_rnn_config_flags.config_from_flags()
if not mt.FLAGS.eval:
train_file = tf.gfile.Glob(
os.path.join(data_dir, "training_melodies.tfrecord"))
tf.logging.info("Train dir: %s", train_dir)
with tf.gfile.Open(os.path.join(train_dir, "hparams"), mode="w") as f:
f.write("\t".join([mt.FLAGS.config, mt.FLAGS.hparams]))
graph = events_rnn_graph.get_build_graph_fn("train", config,
train_file)
events_rnn_train.run_training(
graph,
train_dir,
mt.FLAGS.num_training_steps,
mt.FLAGS.summary_frequency,
checkpoints_to_keep=mt.FLAGS.num_checkpoints)
else:
eval_file = tf.gfile.Glob(
os.path.join(data_dir, "eval_melodies.tfrecord"))
eval_dir = os.path.join(tgt.MODEL_DIR, "logdir/eval")
if not os.path.exists(eval_dir):
tf.gfile.MakeDirs(eval_dir)
tf.logging.info("Eval dir: %s", eval_dir)
examples = mt.FLAGS.num_eval_examples if mt.FLAGS.num_eval_examples else magenta.common.count_records(
eval_file)
if examples >= config.hparams.batch_size:
num_batches = examples // config.hparams.batch_size
else:
config.hparams.batch_size = examples
num_batches = 1
graph = events_rnn_graph.get_build_graph_fn("eval", config, eval_file)
events_rnn_train.run_eval(graph, train_dir, eval_dir, num_batches)
def main(unused_argv):
tf.logging.set_verbosity(FLAGS.log)
if not FLAGS.run_dir:
tf.logging.fatal('--run_dir required')
return
if not FLAGS.sequence_example_file:
tf.logging.fatal('--sequence_example_file required')
return
sequence_example_file_paths = tf.gfile.Glob(
os.path.expanduser(FLAGS.sequence_example_file))
run_dir = os.path.expanduser(FLAGS.run_dir)
config = drums_rnn_config_flags.config_from_flags()
mode = 'eval' if FLAGS.eval else 'train'
build_graph_fn = events_rnn_graph.get_build_graph_fn(
mode, config, sequence_example_file_paths)
train_dir = os.path.join(run_dir, 'train')
if not os.path.exists(train_dir):
tf.gfile.MakeDirs(train_dir)
tf.logging.info('Train dir: %s', train_dir)
if FLAGS.eval:
eval_dir = os.path.join(run_dir, 'eval')
if not os.path.exists(eval_dir):
tf.gfile.MakeDirs(eval_dir)
tf.logging.info('Eval dir: %s', eval_dir)
num_batches = (
(FLAGS.num_eval_examples
or magenta.common.count_records(sequence_example_file_paths)) //
config.hparams.batch_size)
events_rnn_train.run_eval(build_graph_fn, train_dir, eval_dir,
num_batches)
else:
events_rnn_train.run_training(
build_graph_fn,
train_dir,
FLAGS.num_training_steps,
FLAGS.summary_frequency,
checkpoints_to_keep=FLAGS.num_checkpoints)
def main(unused_argv):
tf.logging.set_verbosity(FLAGS.log)
if not FLAGS.run_dir:
tf.logging.fatal('--run_dir required')
return
if not FLAGS.sequence_example_file:
tf.logging.fatal('--sequence_example_file required')
return
sequence_example_file_paths = tf.gfile.Glob(
os.path.expanduser(FLAGS.sequence_example_file))
run_dir = os.path.expanduser(FLAGS.run_dir)
config = melody_rnn_config_flags.config_from_flags()
mode = 'eval' if FLAGS.eval else 'train'
build_graph_fn = events_rnn_graph.get_build_graph_fn(
mode, config, sequence_example_file_paths)
train_dir = os.path.join(run_dir, 'train')
if not os.path.exists(train_dir):
tf.gfile.MakeDirs(train_dir)
tf.logging.info('Train dir: %s', train_dir)
if FLAGS.eval:
eval_dir = os.path.join(run_dir, 'eval')
if not os.path.exists(eval_dir):
tf.gfile.MakeDirs(eval_dir)
tf.logging.info('Eval dir: %s', eval_dir)
num_batches = (
(FLAGS.num_eval_examples if FLAGS.num_eval_examples else
magenta.common.count_records(sequence_example_file_paths)) //
config.hparams.batch_size)
events_rnn_train.run_eval(build_graph_fn, train_dir, eval_dir, num_batches)
else:
events_rnn_train.run_training(build_graph_fn, train_dir,
FLAGS.num_training_steps,
FLAGS.summary_frequency,
checkpoints_to_keep=FLAGS.num_checkpoints)
def testBuildEvalGraph(self):
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'eval', self.config,
sequence_example_file_paths=[self._sequence_file.name])()
def _build_graph_for_generation(self):
events_rnn_graph.get_build_graph_fn('generate', self._config)()
def testBuildCudnnGenerateGraphWithResidualConnections(self):
self.config.hparams.use_cudnn = True
self.config.hparams.residual_connections = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildCudnnGenerateGraph(self):
self.config.hparams.use_cudnn = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildEvalGraph(self):
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'eval',
self.config,
sequence_example_file_paths=[self._sequence_file.name])()
# run_dir = os.path.join(work_dir, 'logdir')
run_dir = 'logdir'
print(run_dir)
tf.logging.set_verbosity('INFO')
sequence_example_file = 'data' + os.path.sep + 'training_melodies.tfrecord'
sequence_example_file_paths = tf.io.gfile.glob(
os.path.join(work_dir, sequence_example_file))
config = melody_rnn_model.default_configs['attention_rnn']
config.hparams.batch_size = 64
config.hparams.rnn_layer_sizes = [64, 64]
mode = 'train'
build_graph_fn = events_rnn_graph.get_build_graph_fn(
mode, config, sequence_example_file_paths)
train_dir = os.path.join(run_dir, 'train')
eval_dir = os.path.join(run_dir, 'eval')
print('Train directory: %s', train_dir)
print('Evaluate directory: %s', eval_dir)
print(config.hparams.batch_size)
print(magenta.common.count_records(sequence_example_file_paths))
# num_batches = magenta.common.count_records(sequence_example_file_paths) // config.hparams.batch_size
events_rnn_train.run_training(build_graph_fn, train_dir, 20000, 10, 3)
def testBuildCudnnGenerateGraphWithResidualConnections(self):
self.config.hparams.use_cudnn = True
self.config.hparams.residual_connections = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildCudnnGenerateGraph(self):
self.config.hparams.use_cudnn = True
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildGenerateGraph(self):
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildGenerateGraph(self):
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn('generate', self.config)()
def testBuildGraphWithAttention(self):
self.config.hparams.attn_length = 10
with tf.Graph().as_default():
events_rnn_graph.get_build_graph_fn(
'train', self.config,
sequence_example_file_paths=[self._sequence_file.name])()
def _build_graph_for_generation(self):
events_rnn_graph.get_build_graph_fn('generate', self._config)()
