如何在tensorflow中使用带有tf.distribute.MirroredStrategy的keras.utils.Sequence数据生成器进行多gpu模型训练？

Question

我想使用TensorFlow2.0在几个GPU上训练一个模型。在分布式培训(https://www.tensorflow.org/guide/distributed_training)的tensorflow教程中，tf.data数据生成器转换为分布式数据集，如下所示：

dist_dataset = mirrored_strategy.experimental_distribute_dataset(dataset)

但是，我希望使用自己的自定义数据生成器(例如，keras.utils.Sequence数据生成器，以及用于异步批处理生成的keras.utils.data_utils.OrderedEnqueuer )。但是mirrored_strategy.experimental_distribute_dataset方法只支持tf.data数据生成器。我该如何使用keras数据生成器呢？

谢谢!

Answer 1

我在同样的情况下使用tf.data.Dataset.from_generator和我的keras.utils.sequence，它解决了我的问题！

train_generator = SegmentationMultiGenerator(datasets, folder) # My keras.utils.sequence object

def generator():
    multi_enqueuer = OrderedEnqueuer(train_generator, use_multiprocessing=True)
    multi_enqueuer.start(workers=10, max_queue_size=10)
    while True:
        batch_xs, batch_ys, dset_index = next(multi_enqueuer.get()) # I have three outputs
        yield batch_xs, batch_ys, dset_index

dataset = tf.data.Dataset.from_generator(generator,
                                         output_types=(tf.float64, tf.float64, tf.int64),
                                         output_shapes=(tf.TensorShape([None, None, None, None]),
                                                        tf.TensorShape([None, None, None, None]),
                                                        tf.TensorShape([None, None])))

strategy = tf.distribute.MirroredStrategy()

train_dist_dataset = strategy.experimental_distribute_dataset(dataset)

请注意，这是我的第一个有效的解决方案--目前，我发现最方便的方法就是在实际的输出形状中放入“None”，我发现这是可行的。

Answer 2

假设你有一个生成器dg，它在调用时以(feature，label)的形式产生样本，而不使用Enqueuer，这是另一种方法：

import tensorflow as tf
import numpy as np


def get_tf_data_Dataset(data_generator_settings_dict):
    length_req = data_generator_settings_dict["length"]
    x_d1 = data_generator_settings_dict['x_d1']
    x_d2 = data_generator_settings_dict['x_d2']
    x_d3 = data_generator_settings_dict['x_d3']
    y_d1 = data_generator_settings_dict['y_d1']
    x_d2 = data_generator_settings_dict['x_d2']
    y_d3 = data_generator_settings_dict['y_d3']
    list_of_x_arrays = [np.zeros((x_d1, x_d2, x_d3)) for _ in range(length_req)]
    list_of_y_arrays = [np.zeros((y_d1, y_d2, y_d3)) for _ in range(length_req)]
    list_of_tuple_samples = [(x, y) for (x, y) in dg()]
    list_of_x_samples = [x for (x, y) in list_of_tuple_samples]
    list_of_y_samples = [y for (x, y) in list_of_tuple_samples]
    for sample_index in range(length_req):
        list_of_x[sample_index][:] = list_of_x_samples[sample_index]
        list_of_y[sample_index][:] = list_of_y_samples[sample_index]
    return tf.data.Dataset.from_tensor_slices((list_of_x, list_of_y))

它很复杂，但肯定会起作用。这也意味着dg的__call__方法是一个类似于for循环(当然是在__init__之后)：

def __call__(self):
    for _ in self.length:
        # generate x (single sample of feature)
        # generate y (single matching sample of label)
        yield x, y