paddlepaddle

1. paddlepaddle-v1.0.0编译

g++-5
cmake-3.5

2. 基本例子

#-*- coding:utf-8 -*-
import sys
reload(sys)
sys.setdefaultencoding("utf-8")
import os
import paddle.v2 as paddle
import paddle.v2.dataset.uci_housing as uci_housing
with_gpu = os.getenv('WITH_GPU', '0') != '0'

def main():
    # 初始化PaddlePaddle
    paddle.init(use_gpu=with_gpu, trainer_count=1)

    # 模型配置
    x = paddle.layer.data(name='x', type=paddle.data_type.dense_vector(13))#利用前13因数来预测房价
    y_predict = paddle.layer.fc(input=x, size=1, act=paddle.activation.Linear())#预测的房价值,线性激活函数
    y = paddle.layer.data(name='y', type=paddle.data_type.dense_vector(1))#实际的房价值
    cost = paddle.layer.square_error_cost(input=y_predict, label=y)#损失函数

    # 保存网络拓扑
    inference_topology = paddle.topology.Topology(layers=y_predict)
    with open("inference_topology.pkl", 'wb') as f:
        inference_topology.serialize_for_inference(f)

    # 创建参数
    parameters = paddle.parameters.create(cost)
    # 创建trainer
    optimizer = paddle.optimizer.Momentum(momentum=0)#learning_rate=0.0001 学习率
    trainer = paddle.trainer.SGD(
        cost=cost, parameters=parameters, update_equation=optimizer)#随机梯度下降算法
    feeding = {'x': 0, 'y': 1}

    # 读取数据且打印训练的中间信息
    def event_handler(event):
        if isinstance(event, paddle.event.EndIteration):
            if event.batch_id % 100 == 0:
                print "Pass %d, Batch %d, Cost %f" % (
                    event.pass_id, event.batch_id, event.cost)

        if isinstance(event, paddle.event.EndPass):
            if event.pass_id % 10 == 0:
                with open('params_pass_%d.tar' % event.pass_id, 'w') as f:
                    trainer.save_parameter_to_tar(f)
            result = trainer.test(
                reader=paddle.batch(uci_housing.test(), batch_size=2),
                feeding=feeding)#读取房价数据，将数据打乱，每次取出2条
            print "Test %d, Cost %f" % (event.pass_id, result.cost)

    # 开始训练
    trainer.train(
        reader=paddle.batch(
            paddle.reader.shuffle(uci_housing.train(), buf_size=500),
            batch_size=2),
        feeding=feeding,
        event_handler=event_handler,#提供一个 event_handler，来打印训练的进度：
        num_passes=3000)

    # 生成测试数据
    test_data_creator = paddle.dataset.uci_housing.test()
    test_data = []
    test_label = []

    #取出测试集中5条数据用于最后的预测
    for item in test_data_creator():
        test_data.append((item[0], ))
        test_label.append(item[1])
        if len(test_data) == 5:
            break

    #推测inference
    probs = paddle.infer(
        output_layer=y_predict, parameters=parameters, input=test_data)
    for i in xrange(len(probs)):
        print "label=" + str(test_label[i][0]) + ", predict=" + str(probs[i][0])
if __name__ == '__main__':
    main()

import sys
reload(sys)
sys.setdefaultencoding("utf-8")
import paddle.v2 as paddle
import multiprocessing
# Initialize PaddlePaddle.
paddle.init(use_gpu=False, trainer_count=1)

# Configure the neural network.
x = paddle.layer.data(name='x', type=paddle.data_type.dense_vector(13))
y_predict = paddle.layer.fc(input=x, size=1, act=paddle.activation.Linear())

# Infer using provided test data.
probs = paddle.infer(
    output_layer=y_predict,
    parameters=paddle.dataset.uci_housing.model(),
    #parameters="params_pass_20.tar",
    input=[item for item in paddle.dataset.uci_housing.test()()])

for i in xrange(len(probs)):
    print 'Predicted price: ${:,.2f}'.format(probs[i][0] * 1000)
print("-------------------------------")
if __name__ == '__main__':
    mp = multiprocessing.Pool(2)