使用StellarGraph的“监视您的步进”模型不适用于GPU。

Question

我正在尝试用WatchYourStep算法来训练一个使用StellarGraph的大图嵌入。

由于某种原因，模型只在一个上训练，而不使用GPU。

使用：

• TensorFlow-gpu 2.3.1
• 有2个GPU，Cuda10.1
• 运行在nvidia-码头容器中。(tf.debugging.set_log_device_placement(True))
• I试图在with tf.device('/GPU:0'):
• I下运行，已经尝试用tf.distribute.MirroredStrategy().
• Tried运行它来卸载tensorflow并重新安装tensorflow-gpu.

尽管如此，在运行nvidia时，我没有看到GPU上有任何活动，而且培训非常缓慢。

如何调试？

def watch_your_step_model():
    '''use the config to geenrate the WatchYourStep model'''
    cfg = load_config()
    generator           = generator_for_watch_your_step()
    num_walks           = cfg['num_walks']
    embedding_dimension = cfg['embedding_dimension']
    learning_rate       = cfg['learning_rate']
    
    wys = WatchYourStep(
        generator,
        num_walks=num_walks,
        embedding_dimension=embedding_dimension,
        attention_regularizer=regularizers.l2(0.5),
    )
    
    x_in, x_out = wys.in_out_tensors()
    model = Model(inputs=x_in, outputs=x_out)
    model.compile(loss=graph_log_likelihood, optimizer=optimizers.Adam(learning_rate))
    return model, generator, wys

def train_watch_your_step_model(epochs = 3000):
    cfg = load_config()
    batch_size      = cfg['batch_size']
    steps_per_epoch = cfg['steps_per_epoch']
    callbacks, checkpoint_file = watch_your_step_callbacks(cfg)
    
    # strategy = tf.distribute.MirroredStrategy()
    # print('Number of devices: {}'.format(strategy.num_replicas_in_sync))
    # with strategy.scope():
    
    model, generator, wys = watch_your_step_model()

    train_gen = generator.flow(batch_size=batch_size, num_parallel_calls=8)
    train_gen.prefetch(20480000)

    history = model.fit(
        train_gen, 
        epochs=epochs, 
        verbose=1, 
        steps_per_epoch=steps_per_epoch,
        callbacks = callbacks
    )
     
    copy_last_trained_wys_weights_to_data()
    
    return history, checkpoint_file

with tf.device('/GPU:0'):
    train_watch_your_step_model()

Answer 1

我只是按照下面的说明：https://github.com/stellargraph/stellargraph/issues/546。

对我起作用了。

基本上，您必须从stellargraph编辑文件setup.py，并删除tensorflow要求(第25和27行https://github.com/stellargraph/stellargraph/blob/develop/setup.py)。