Question

我在使用Google Colab训练GAN时遇到问题。我的训练数据集是MIDI文件的文件夹，该文件使用pypianoroll包转换为numpy数组。它一直用不上25 GB的可用内存，我认为这个数量绝对足够。

我尝试将批次大小减小为1，但这没有帮助。

class GANDrum(object):
    def __init__(self,buffer_size=60000,batch_size=1,epochs=20,noise_dim=100,num_examples_to_generate=1):
    self.train_files=[]
    self.buffer_size=buffer_size
    self.batch_size=batch_size
    self.load_data()

    self.generator_optimizer=tf.keras.optimizers.Adam(1e-4)
    self.discriminator_optimizer=tf.keras.optimizers.Adam(1e-4)

    self.generator=self.make_generator_model()
    self.discriminator=self.make_discriminator_model()

    self.checkpoint_prefix=os.path.join('./training_checkpoints',"ckpt")
    self.checkpoint=tf.train.Checkpoint(generator_optimizer=self.generator_optimizer,discriminator_optimizer=self.discriminator_optimizer,generator=self.generator,discriminator=self.discriminator)

    self.epochs=epochs
    self.noise_dim=noise_dim
    self.seed=tf.random.normal([num_examples_to_generate, noise_dim])


def load_data(self):
    i=0
    entries=os.listdir(r'/content/drive/My Drive/midi_d_processed/')
    for entry in entries:
        try:
            data=(pp.parse(r'/content/drive/My Drive/midi_d_processed/'+entry))
            mtx=tf.reshape(tensor=tf.convert_to_tensor(data.tracks[0].pianoroll),shape=(8,12,128))

            mtx=tf.cast(mtx, tf.float32)
            self.train_files.append(mtx)
        except:
            i=i+1
    print(i,'files were not in the correct format')
    self.train_dataset=tf.data.Dataset.from_tensor_slices(self.train_files).shuffle(self.buffer_size).batch(self.batch_size)

def make_generator_model(self):
    try:
        model=tf.keras.Sequential()


        model.add(layers.Dense(units=1536,input_shape=(100,)))
        model.add(layers.BatchNormalization())
        model.add(layers.LeakyReLU())

        model.add(layers.Dense(units=768))
        model.add(layers.BatchNormalization())
        model.add(layers.LeakyReLU())


        model.add(layers.Reshape((2,3,128)))


        model.add(layers.Conv2DTranspose(filters=1,strides=2,kernel_size=2))
        model.add(layers.BatchNormalization())
        model.add(layers.LeakyReLU())

        model.add(layers.Conv2DTranspose(filters=1,strides=2,kernel_size=2))
        model.add(layers.BatchNormalization())
        model.add(layers.LeakyReLU())

        model.add(layers.Conv2DTranspose(filters=128,strides=1,kernel_size=1,activation='tanh'))
        model.add(layers.BatchNormalization())
        model.add(layers.Softmax())

        return model
    except:
        print('make_generator_model')



def make_discriminator_model(self):
        model=tf.keras.Sequential()

        model.add(layers.Conv2D(filters=128,strides=1,kernel_size=1,input_shape=[8,12,128]))
        model.add(layers.LeakyReLU())
        model.add(layers.Conv2D(filters=1,strides=2,kernel_size=2,input_shape=[8,12,128]))
        model.add(layers.LeakyReLU())

        model.add(layers.Conv2D(filters=1,strides=2,kernel_size=2))
        model.add(layers.LeakyReLU())

        model.add(layers.Conv2D(filters=1,kernel_size=2))
        model.add(layers.LeakyReLU())
        model.add(layers.Flatten())
        model.add(layers.Dense(1))

        return model


def discriminator_loss(self,real_output,fake_output):
    cross_entropy=tf.keras.losses.BinaryCrossentropy(from_logits=True)

    real_loss=cross_entropy(tf.ones_like(real_output),real_output)
    fake_loss=cross_entropy(tf.zeros_like(fake_output),fake_output)
    total_loss=real_loss+fake_loss
    return total_loss


def generator_loss(self,fake_output):
    cross_entropy=tf.keras.losses.BinaryCrossentropy(from_logits=True)

    return cross_entropy(tf.ones_like(fake_output),fake_output)


def train_step(self,files):
    noise = tf.random.normal([self.batch_size,self.noise_dim])

    with tf.GradientTape() as gen_tape,tf.GradientTape() as disc_tape:
        generated_midi=self.generator(noise,training=True)

        real_output=self.discriminator(files,training=True)
        fake_output=self.discriminator(generated_midi,training=True)

        gen_loss=self.generator_loss(fake_output)
        disc_loss=self.discriminator_loss(real_output,fake_output)

    gradients_of_generator=gen_tape.gradient(gen_loss,self.generator.trainable_variables)
    gradients_of_discriminator=disc_tape.gradient(disc_loss,self.discriminator.trainable_variables)

    self.generator_optimizer.apply_gradients(zip(gradients_of_generator,self.generator.trainable_variables))
    self.discriminator_optimizer.apply_gradients(zip(gradients_of_discriminator,self.discriminator.trainable_variables))


def train(self,dataset,epochs):
    for self.epoch in range(epochs):
        start=time.time()

    for midi_batch in self.train_dataset:
        self.train_step(midi_batch)

    self.generate_and_save_midi(self.generator,self.epoch+1,self.seed)

    if(self.epoch+1)%15==0:
        checkpoint.save(file_prefix=self.checkpoint_prefix)

    print ('Time for epoch {} is {} sec'.format(self.epoch+1,time.time()-start))


def generate_and_save_midi(self,model,epoch,test_input):
        predictions=model(test_input,training=False)

        shaped_predictions=np.reshape(a=predictions.numpy(),newshape=(96,128))

        track=pp.Track(pianoroll=shaped_predictions,is_drum=True)

        multitrack=pp.Multitrack(tracks=[track])

        multitrack.write('./generated_midi')


def run(self):
    self.train(self.train_dataset,self.epochs)

我通常会收到一条错误消息，提示我内存不足，但有时会收到未知错误。可能是什么问题？

为什么在训练时我的创生对抗网络会不断耗尽内存？

0 个答案: