c93334700c4c894cf9faedc1aea46ffe5dccb73a,shared/gan.py,,discriminator_wide_resnet,#Any#Any#,160

Before Change

    batch_size = int(x.get_shape()[0])
    layers = config["d_wide_resnet_depth"]
    result = x
    result = build_conv_tower(result, config["conv_d_layers"][:1], config["d_pre_res_filter"], config["batch_size"], config["d_batch_norm"], True, "d_", config["d_activation"], stride=config["d_pre_res_stride"])

    result = activation(result)
    result = conv2d(result, layers[0], name="d_expand", k_w=3, k_h=3, d_h=1, d_w=1)

After Change

    //result = build_conv_tower(result, config["conv_d_layers"][:1], config["d_pre_res_filter"], config["batch_size"], config["d_batch_norm"], True, "d_", config["d_activation"], stride=config["d_pre_res_stride"])

    //result = activation(result)
    result = conv2d(result, layers[0], name="d_expand1", k_w=3, k_h=3, d_h=2, d_w=2)
    result = batch_norm(config["batch_size"], name="d_expand_bn1")(result)
    result = activation(result)
    result = conv2d(result, layers[0], name="d_expand2", k_w=3, k_h=3, d_h=2, d_w=2)
    result = batch_norm(config["batch_size"], name="d_expand_bn2")(result)
    result = activation(result)
    result = conv2d(result, layers[0], name="d_expand", k_w=3, k_h=3, d_h=1, d_w=1)
    result = batch_norm(config["batch_size"], name="d_expand_bn")(result)
    result = activation(result)

Instances