KerasParallel.py

from keras.layers.merge import Concatenate
from keras.layers.core import Lambda
from keras.models import Model

import tensorflow as tf

#model_p=make_parallel(model,2)
#model.compile
#Kera多GPU并行，用model_p训练，model存模型及权重，否则会有bug
#多GPU时验证集样本数需要为batch_size整数倍,否则会有bug
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(Concatenate(axis=0)(outputs))
        
        return Model(model.inputs, merged)