#edit-mode: -*- python -*-
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from paddle.trainer_config_helpers import *

TrainData(SimpleData(
    files = "legacy/trainer/tests/sample_filelist.txt",
    feat_dim = 3,
    context_len = 0,
    buffer_capacity = 1000000,
    async_load_data = False))

settings(batch_size = 100)

data = data_layer(name='input', size=3)

wt = data_layer(name='weight', size=1)

fc1 = fc_layer(input=data, size=5,
               bias_attr=True,
               act=SigmoidActivation())

fc2 = fc_layer(input=data, size=12,
               bias_attr=True,
               param_attr=ParamAttr(name='sharew'),
               act=LinearActivation())

fc3 = fc_layer(input=data, size=3,
               bias_attr=True,
               act=TanhActivation())

fc4 = fc_layer(input=data, size=5,
               bias_attr=True,
               layer_attr=ExtraAttr(drop_rate=0.5),
               act=SquareActivation())

pool = img_pool_layer(input=fc2,
                      pool_size=2,
                      pool_size_y=3,
                      num_channels=1,
                      padding=1,
                      padding_y=2,
                      stride=2,
                      stride_y=3,
                      pool_type=CudnnAvgPooling())

concat = concat_layer(input=[fc3, fc4])

with mixed_layer(size=3, act=SoftmaxActivation()) as output:
    output += full_matrix_projection(input=fc1)
    output += trans_full_matrix_projection(input=fc2,
                                           param_attr=ParamAttr(name='sharew'))
    output += full_matrix_projection(input=concat)
    output += identity_projection(input=fc3)

lbl = data_layer(name='label', size=1)

cost = classification_cost(input=output, label=lbl, weight=wt,
                           layer_attr=ExtraAttr(device=-1))

nce = nce_layer(input=fc2, label=lbl, weight=wt,
                num_classes=3, 
                neg_distribution=[0.1, 0.3, 0.6])
                
outputs(cost, nce)