我一直在使用下面的变分自编码器有一段时间了，但最近在环境重置后它就停止工作了。我已经查看了tensorflow的文档和最近的更改日志（但我似乎找不到任何超过一个月的良好发布历史），但没有发现这些函数基本定义的任何变化。此外，我得到了一个非常简洁的错误代码，它并不十分描述性或有帮助。
AssertionError: in user code:AssertionError: Could not compute output Tensor("sequential/dense_1/Sigmoid:0", shape=(None, 110), dtype=float32)错误似乎只是指向最后一行，因为它无法提供更多的描述。然而，在查看模型摘要时，模型似乎能够识别所有层及其构建。
vae.summary()Model: "functional_1"__________________________________________________________________________________________________Layer (type) Output Shape Param # Connected to ==================================================================================================input_1 (InputLayer) [(None, 110)] 0 __________________________________________________________________________________________________dense_2 (Dense) (None, 64) 7104 input_1[0][0] __________________________________________________________________________________________________dense_3 (Dense) (None, 2) 130 dense_2[0][0] __________________________________________________________________________________________________dense_4 (Dense) (None, 2) 130 dense_2[0][0] __________________________________________________________________________________________________kl_divergence_layer (KLDivergen [(None, 2), (None, 2 0 dense_3[0][0] dense_4[0][0] __________________________________________________________________________________________________lambda (Lambda) (None, 2) 0 kl_divergence_layer[0][1] __________________________________________________________________________________________________input_2 (InputLayer) [(None, 2)] 0 __________________________________________________________________________________________________multiply (Multiply) (None, 2) 0 lambda[0][0] input_2[0][0] __________________________________________________________________________________________________add (Add) (None, 2) 0 kl_divergence_layer[0][0] multiply[0][0] __________________________________________________________________________________________________sequential (Sequential) (None, 110) 7342 add[0][0] ==================================================================================================Total params: 14,706Trainable params: 14,706Non-trainable params: 0__________________________________________________________________________________________________因此，我不知道哪里出了问题。有没有想法？以下是使用变分自编码器的样本数据。
import pandas as pdfrom sklearn.datasets import make_blobs from sklearn.preprocessing import MinMaxScalerimport keras.backend as Kimport tensorflow as tffrom keras.layers import Input, Dense, Lambda, Layer, Add, Multiplyfrom keras.models import Model, Sequentialfrom keras.callbacks import EarlyStopping, LearningRateSchedulerfrom keras.objectives import binary_crossentropyx, labels = make_blobs(n_samples=150000, n_features=110, centers=16, cluster_std=4.0)scaler = MinMaxScaler()x = scaler.fit_transform(x)x = pd.DataFrame(x)train = x.sample(frac = 0.8)train_indexs = train.index.valuestest = x[~x.index.isin(train_indexs)]print(train.shape, test.shape)def nll(y_true, y_pred): """ Negative log likelihood (Bernoulli). """ return K.sum(K.binary_crossentropy(y_true, y_pred), axis = -1)class KLDivergenceLayer(Layer): """ Identity transform layer that adds KL divergence to the final model loss. """ def __init__(self, *args, **kwargs): self.is_placeholder = True super(KLDivergenceLayer, self).__init__(*args, **kwargs) def call(self, inputs): mu, log_var = inputs #changing form sum to mean kl_batch = - .5 * K.sum(1 + log_var - K.square(mu) - K.exp(log_var), axis=-1) self.add_loss(K.mean(kl_batch), inputs=inputs) return inputs#################### VAE ##############latent_dim = 2decoder = Sequential([ Dense(64, input_dim=latent_dim, activation='relu'), Dense(x.shape[1], activation='sigmoid')])data = Input(shape=(x.shape[1],))h = Dense(64, activation='relu')(data)z_mu = Dense(latent_dim)(h)z_log_var = Dense(latent_dim)(h)#applies KLDivergence here?z_mu, z_log_var = KLDivergenceLayer()([z_mu, z_log_var])#generates random samplesz_sigma = Lambda(lambda t: K.exp(.5*t))(z_log_var)eps = Input(tensor=K.random_normal(stddev=1.0, shape=(K.shape(data)[0], latent_dim)))z_eps = Multiply()([z_sigma, eps])z = Add()([z_mu, z_eps])#apply decoderx_pred = decoder(z)# defines final modelvae = Model(inputs=[data, eps], outputs=x_pred)vae.compile(optimizer='rmsprop', loss=nll)#runs modelvae.fit(train, train, shuffle = True, epochs = 1000, batch_size = 512, validation_data = (test, test), callbacks = [EarlyStopping(patience=50)])回答：
问题出在你的eps层上。它不是模型输入
你可以用一个简单的层来替换它，就像这样:
eps = Lambda(lambda t: K.random_normal(stddev=1.0, shape=(K.shape(t)[0], latent_dim)))(z_log_var)你可以在这里找到运行的笔记本: https://colab.research.google.com/drive/17ugeVin4yOlSD3fBvX4jQNJ2WT7nmlnz?usp=sharing
