loss=nan,val loss=nan

021-07-23 09:43:04.902503: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN)to use the following CPU instructions in performance-critical operations:  AVX AVX2
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
59984/60000 [============================>.] - ETA: 0s - loss: nanWARNING:tensorflow:From D:\Anaconda\envs\tensorflow\lib\site-packages\tensorflow\python\keras\engine\training_v1.py:2048: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
This property should not be used in TensorFlow 2.0, as updates are applied automatically.
60000/60000 [==============================] - 156s 3ms/sample - loss: nan - val_loss: nan
Epoch 2/10
60000/60000 [==============================] - 161s 3ms/sample - loss: nan - val_loss: nan

View Code

原代码：

vae = Model(input_img, y)
vae.compile(optimizer='rmsprop', loss=None)
vae.summary()

修改代码：

training = True
if training:
    vae.compile(optimizer=tf.optimizers.RMSprop(lr=0.001, epsilon=1e-3),
                loss=None,
                experimental_run_tf_function = False)
    vae.summary()
else:
    vae.compile(optimizer=tf.optimizers.RMSprop(lr=0.001, epsilon=1e-3),
                loss=None)
    vae.summary()

修改后运行效果：

loss=nan,val loss=nan

推荐阅读