Created by: shuoyin

测试代码如下，测试环境python2.7.13 + paddle 1.7，CPU

import paddle.fluid as fluid
def sample_generator():
    def __reader__():
        for i in range(10):
            yield (i,i)
    return __reader__

def sample_list_generator(batch_size=4):
    def __reader__():
        batch_data = []
        for i in range(10):
            batch_data.append((i,i))
            if len(batch_data) == batch_size:
                yield batch_data
                batch_data = []
        if len(batch_data) > 0:
            yield batch_data
    return __reader__

def net():
    x = fluid.data(name='x', dtype='int32', shape=[None])
    y = fluid.data(name='y', dtype='int32', shape=[None])
    loader = fluid.io.DataLoader.from_generator(
            feed_list=[x,y], capacity=2, use_double_buffer=True, iterable=True)
    fetch_dict = {'x': x.name, 'y': y.name} 
    return loader, fetch_dict

loader, fetch_dict = net() 
place = fluid.CPUPlace()
exe = fluid.Executor(place)
exe.run(fluid.default_startup_program())

print "use sample_generator"
gen = sample_generator()
loader.set_sample_generator(gen,batch_size=4, places=place)
for data in loader: 
    print exe.run(program=fluid.default_main_program(), feed=data,
            fetch_list=[fetch_dict['x'], fetch_dict['y']])

print "use sample_list_generator"
gen = sample_list_generator()
loader.set_sample_list_generator(gen,places=place)
for data in loader: 
    print exe.run(program=fluid.default_main_program(), feed=data,
            fetch_list=[fetch_dict['x'], fetch_dict['y']])

运行之后，输出如下:

use sample_generator
[array([0, 1, 2, 3], dtype=int32), array([0, 1, 2, 3], dtype=int32)]
[array([4, 5, 6, 7], dtype=int32), array([4, 5, 6, 7], dtype=int32)]
use sample_list_generator
[array([0, 1, 2, 3], dtype=int32), array([0, 1, 2, 3], dtype=int32)]
[array([4, 5, 6, 7], dtype=int32), array([4, 5, 6, 7], dtype=int32)]
[array([8, 9], dtype=int32), array([8, 9], dtype=int32)]

这个问题还请paddle的同学解答一下