python - 无效参数错误 : input_1_1:0 is both fed and fetched

Question

我在 keras-vis 中使用 visualize_activation 函数:

from vis.visualization import visualize_activation, visualize_cam
from vis.utils import utils
from keras import activations

from matplotlib import pyplot as plt
%matplotlib inline
plt.rcParams['figure.figsize'] = (18, 6)

# Utility to search for layer index by name. 
# Alternatively we can specify this as -1 since it corresponds to the last 
# layer.
layer_idx = utils.find_layer_idx(my_model, 're_lu_3')

patches = np.expand_dims(patches,axis=3)

# This is the output node we want to maximize.
filter_idx = None
img = visualize_activation(my_model, layer_idx, filter_indices=filter_idx, 
seed_input=patches[0])
plt.imshow(img[..., 0])

但是，这会引发错误:InvalidArgumentError: input_1_1:0 is both fed and fetched。

如何解决？我尝试使用 tf.identity 创建 my_model 的副本，但这没有用。

Answer 1

keras-vis 在 pip 上似乎坏了，尝试直接在 GitHub master 分支上安装:

pip uninstall vis
pip install git+https://github.com/raghakot/keras-vis.git -U

使用 pip 上的版本，MNIST 和 ResNet 示例都输出错误:InvalidArgumentError: input_1_1:0 is both fed and fetched.。更新后，它们都运行良好。