Neural network 当使用多个GPU进行训练时，加载预训练模型失败

我已经训练了一个网络模型，并通过

checkpoint=ModelCheckpoint（filepath='weights.hdf5'）

回调保存了它的权重和体系结构。在培训期间，我通过调用下面的函数来使用多个GPU：

def make_parallel(model, gpu_count):
    def get_slice(data, idx, parts):
        shape = tf.shape(data)
        size = tf.concat([ shape[:1] // parts, shape[1:] ],axis=0)
        stride = tf.concat([ shape[:1] // parts, shape[1:]*0 ],axis=0)
        start = stride * idx
        return tf.slice(data, start, size)

    outputs_all = []
    for i in range(len(model.outputs)):
        outputs_all.append([])

    #Place a copy of the model on each GPU, each getting a slice of the batch
    for i in range(gpu_count):
        with tf.device('/gpu:%d' % i):
            with tf.name_scope('tower_%d' % i) as scope:

                inputs = []
                #Slice each input into a piece for processing on this GPU
                for x in model.inputs:
                    input_shape = tuple(x.get_shape().as_list())[1:]
                    slice_n = Lambda(get_slice, output_shape=input_shape, arguments={'idx':i,'parts':gpu_count})(x)
                    inputs.append(slice_n)                

                outputs = model(inputs)

                if not isinstance(outputs, list):
                    outputs = [outputs]

                #Save all the outputs for merging back together later
                for l in range(len(outputs)):
                    outputs_all[l].append(outputs[l])

    # merge outputs on CPU
    with tf.device('/cpu:0'):
        merged = []
        for outputs in outputs_all:
            merged.append(merge(outputs, mode='concat', concat_axis=0))

        return Model(input=model.inputs, output=merged)

使用测试代码：

from keras.models import Model, load_model
import numpy as np
import tensorflow as tf

model = load_model('cpm_log/deneme.hdf5')

x_test = np.random.randint(0, 255, (1, 368, 368, 3))

output = model.predict(x = x_test, batch_size=1)

print output[4].shape

我得到的错误如下：

Traceback (most recent call last):
  File "cpm_test.py", line 5, in <module>
    model = load_model('cpm_log/Jun5_1000/deneme.hdf5')
  File "/usr/local/lib/python2.7/dist-packages/keras/models.py", line 240, in load_model
    model = model_from_config(model_config, custom_objects=custom_objects)
  File "/usr/local/lib/python2.7/dist-packages/keras/models.py", line 301, in model_from_config
    return layer_module.deserialize(config, custom_objects=custom_objects)
  File "/usr/local/lib/python2.7/dist-packages/keras/layers/__init__.py", line 46, in deserialize
    printable_module_name='layer')
  File "/usr/local/lib/python2.7/dist-packages/keras/utils/generic_utils.py", line 140, in deserialize_keras_object
    list(custom_objects.items())))
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 2378, in from_config
    process_layer(layer_data)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 2373, in process_layer
    layer(input_tensors[0], **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 578, in __call__
    output = self.call(inputs, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/keras/layers/core.py", line 659, in call
    return self.function(inputs, **arguments)
  File "/home/muhammed/DEV_LIBS/developments/mocap/pose_estimation/training/cpm/multi_gpu.py", line 12, in get_slice
    def get_slice(data, idx, parts):
NameError: global name 'tf' is not defined

回溯（最近一次呼叫最后一次）：
文件“cpm_test.py”，第5行，在
模型=负载模型（'cpm\u log/Jun5\u 1000/deneme.hdf5'）
文件“/usr/local/lib/python2.7/dist-packages/keras/models.py”，第240行，在load\u模型中
模型=来自配置的模型（模型配置，自定义对象=自定义对象）
文件“/usr/local/lib/python2.7/dist-packages/keras/models.py”，第301行，模型配置中的
返回层\模块。反序列化（配置，自定义\对象=自定义\对象）
文件“/usr/local/lib/python2.7/dist-packages/keras/layers/_-init__.py”，第46行，反序列化
可打印\u模块\u name='layer'）
文件“/usr/local/lib/python2.7/dist packages/keras/utils/generic_utils.py”，第140行，反序列化_keras_对象
列表（自定义对象.项（））
文件“/usr/local/lib/python2.7/dist packages/keras/engine/topology.py”，第2378行，from_config
处理层（层数据）
文件“/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py”，第2373行，进程层中
图层（输入_张量[0]，**kwargs）
文件“/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py”，第578行，在调用中__
输出=自调用（输入，**kwargs）
文件“/usr/local/lib/python2.7/dist packages/keras/layers/core.py”，第659行，在调用中
返回self.function（输入，**参数）
文件“/home/muhammed/DEV_LIBS/developments/mocap/pose_assessment/training/cpm/multi_gpu.py”，第12行，在get_切片中
def get_切片（数据、idx、零件）：
NameError:未定义全局名称“tf”

---

通过检查错误输出，我确定问题在于并行化代码。然而，我无法解决这个问题

您可能需要使用

自定义对象

来启用模型加载

import tensorflow as tf
model = load_model('model.h5', custom_objects={'tf': tf,})

---

如果在

get_slice

定义的开头添加

import tensorflow as tf

，会发生什么情况？它已被添加。一切都没有改变。