d6c1b86594ef9a97e9f503547ab6567f89508486,tests/integration_tests/test_experiment.py,,test_experiment_image_inputs,#Any#,177

Before Change

def test_experiment_image_inputs(csv_filename):
    // Image Inputs
    image_dest_folder = os.path.join(os.getcwd(), "generated_images")
    input_features_template = Template(
        "[{type: text, name: random_text, vocab_size: 10,"
        " max_len: 10, encoder: stacked_cnn}, {type: numerical,"
        " name: random_number}, "
        "{type: image, name: random_image, encoder: ${encoder},"
        "preprocessing: {in_memory: ${in_memory}, height: 8, width: 8, "
        "num_channels: 3}, resnet_size: 8, destination_folder: ${folder}, "
        "fc_size: 32, num_filters: 8}]")

    // Resnet encoder
    input_features = input_features_template.substitute(
        encoder="resnet",
        folder=image_dest_folder,
        in_memory="true",
    )
    output_features = "[{type: category, name: intent, reduce_input: sum," \
                      " vocab_size: 2}," \
                      "{type: numerical, name: random_num_output}]"

After Change

    image_dest_folder = os.path.join(os.getcwd(), "generated_images")

    // Resnet encoder
    input_features = [
        image_feature(
            folder=image_dest_folder,
            encoder="resnet",
            preprocessing={
                "in_memory": True,
                "height": 8,
                "width": 8,
                "num_channels": 3
            },
            fc_size=16,
            num_filters=8
        ),
        text_feature(encoder="embed", min_len=1),
        numerical_feature()
    ]
    output_features = [
        categorical_feature(vocab_size=2, reduce_input="sum"),
        numerical_feature()
    ]

Instances