python - NotImplementedError : Cannot convert a symbolic Tensor (lstm_2/strided_slice:0) to a numpy array. T

Question

tensorflow 版本 2.3.1
numpy 版本 1.20
代码下方

# define model
model = Sequential()
model.add(LSTM(50, activation='relu', input_shape=(n_steps, n_features)))
model.add(Dense(1))
model.compile(optimizer='adam', loss='mse')

我们有

NotImplementedError: Cannot convert a symbolic Tensor (lstm_2/strided_slice:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported

在我看来这是一个疯狂的错误!

Answer 1

我用 numpy 降级到 1.18.5 解决了

pip install -U numpy==1.18.5