def test_AFN(afn_dnn_hidden_units, sparse_feature_num, dense_feature_num): model_name = 'AFN' sample_size = SAMPLE_SIZE x, y, feature_columns = get_test_data( sample_size, sparse_feature_num=sparse_feature_num, dense_feature_num=dense_feature_num) model = AFN(feature_columns, feature_columns, afn_dnn_hidden_units=afn_dnn_hidden_units, device=get_device()) check_model(model, model_name, x, y)
def test_CCPM_without_seq(sparse_feature_num, dense_feature_num): model_name = "CCPM" sample_size = SAMPLE_SIZE x, y, feature_dim_dict = get_test_data( sample_size, sparse_feature_num, dense_feature_num, sequence_feature=()) model = CCPM(feature_dim_dict, conv_kernel_width=(3, 2), conv_filters=(2, 1), hidden_size=[32, ], keep_prob=0.5, ) check_model(model, model_name, x, y)
def test_FGCNN_without_seq(sparse_feature_num, dense_feature_num): model_name = "FGCNN_noseq" sample_size = SAMPLE_SIZE x, y, feature_dim_dict = get_test_data( sample_size, sparse_feature_num, dense_feature_num, sequence_feature=()) model = FGCNN(feature_dim_dict, conv_kernel_width=(), conv_filters=( ), new_maps=(), pooling_width=(), dnn_hidden_units=(32,), dnn_dropout=0.5, ) # TODO: add model_io check check_model(model, model_name, x, y, check_model_io=False)
def test_FGCNN(sparse_feature_num, dense_feature_num): model_name = "FGCNN" sample_size = 32 x, y, feature_dim_dict = get_test_data( sample_size, sparse_feature_num, dense_feature_num) model = FGCNN(feature_dim_dict, conv_kernel_width=(3, 2), conv_filters=(2, 1), new_maps=( 2, 2), pooling_width=(2, 2), dnn_hidden_units=(32, ), dnn_dropout=0.5, ) # TODO: add model_io check check_model(model, model_name, x, y, check_model_io=False)
def test_CCPM_without_seq(sparse_feature_num, dense_feature_num): if tf.__version__ >= "2.0.0": return model_name = "CCPM" sample_size = SAMPLE_SIZE x, y, feature_columns = get_test_data( sample_size, sparse_feature_num, dense_feature_num, sequence_feature=()) model = CCPM(feature_columns, feature_columns,conv_kernel_width=(3, 2), conv_filters=( 2, 1), dnn_hidden_units=[32, ], dnn_dropout=0.5) check_model(model, model_name, x, y)
def test_CCPM(sparse_feature_num, dense_feature_num): model_name = "CCPM" sample_size = SAMPLE_SIZE x, y, feature_columns = get_test_data(sample_size, sparse_feature_num, dense_feature_num) model = CCPM(feature_columns, feature_columns, conv_kernel_width=(3, 2), conv_filters=(2, 1), dnn_hidden_units=[ 32, ], dnn_dropout=0.5) check_model(model, model_name, x, y)
def test_CCPM(sparse_feature_num, dense_feature_num): model_name = "CCPM" sample_size = 32 x, y, feature_dim_dict = get_test_data(sample_size, sparse_feature_num, dense_feature_num) model = CCPM( feature_dim_dict, conv_kernel_width=(3, 2), conv_filters=(2, 1), dnn_hidden_units=[ 32, ], dnn_dropout=0.5, ) check_model(model, model_name, x, y)
def test_FGCNN(sparse_feature_num, dense_feature_num): model_name = "FGCNN" sample_size = 32 x, y, feature_dim_dict = get_test_data(sample_size, sparse_feature_num, dense_feature_num) model = FGCNN( feature_dim_dict, conv_kernel_width=(3, 2), conv_filters=(2, 1), new_maps=(2, 2), pooling_width=(2, 2), hidden_size=[ 32, ], keep_prob=0.5, ) check_model(model, model_name, x, y, check_model_io=False)
def test_FGCNN_without_seq(sparse_feature_num, dense_feature_num): model_name = "FGCNN" sample_size = SAMPLE_SIZE x, y, feature_dim_dict = get_test_data(sample_size, sparse_feature_num, dense_feature_num, sequence_feature=()) model = FGCNN( feature_dim_dict, conv_kernel_width=(), conv_filters=(), new_maps=(), pooling_width=(), hidden_size=(32, ), keep_prob=0.5, ) check_model(model, model_name, x, y, check_model_io=False)
def test_FGCNN(sparse_feature_num, dense_feature_num): model_name = "FGCNN" sample_size = SAMPLE_SIZE x, y, feature_columns = get_test_data( sample_size, embedding_size=8, sparse_feature_num=sparse_feature_num, dense_feature_num=dense_feature_num) model = FGCNN( feature_columns, feature_columns, conv_kernel_width=(3, 2), conv_filters=(2, 1), new_maps=(2, 2), pooling_width=(2, 2), dnn_hidden_units=(32, ), dnn_dropout=0.5, ) # TODO: add model_io check check_model(model, model_name, x, y, check_model_io=False)