本文链接：https://blog.csdn.net/mr_muli/article/details/90929900

报错：tensorflow.python.framework.errors_impl.InvalidArgumentError: Input to reshape is a tensor with 76800 values, but the requested shape has 768

print("auxiliary_tensor:",auxiliary_tensor)    # Tensor("AuxLogits/SpatialSqueeze:0", shape=(1, 768), dtype=float32)
print("----------------------------------------------")
print("train_auxiliarys-shape:",np.shape(train_auxiliarys))      # (10, 768)
print("train_ground_truth-shape:",np.shape(train_ground_truth))  # (10, 5)
print("auxiliary_input:",auxiliary_input)   # Tensor("input/AuxiliaryInputPlaceholder:0", shape=(?, 768), dtype=float32)
print("ground_truth_input:",ground_truth_input) # Tensor("input/GroundTruthInput:0", shape=(?, 5), dtype=float32)
print("type train_auxiliarys:",type(train_auxiliarys))   # type train_auxiliarys: <class 'list'>
print("type train_ground_truth:",type(train_ground_truth))   # type train_ground_truth: <class 'list'>
print("----------------------------------------------")


train_summary, _ = sess.run([merged, train_step],
         feed_dict={auxiliary_input: train_auxiliarys,
                    ground_truth_input: train_ground_truth})

这个问题，我前后折腾了很久，还是没能解决…抱歉…