Простое обучение модели RNN ⇐ Python

[Гость]

**I'm trying to run a simple RNN model using keras on Tensorflow and I'm getting an error, probably linked to the graphics card (I'm not sure). It should be noted that this code often works correctly and sometimes it gives me this error and I don't really understand. I've tried varying the number of units in the single RNN layer or the number of batches, but to no avail. I've also tried to install libraries by reading the error comments, without success either. Could this be due to the way Tensorflow-gpu was installed? I'd like to point out that there are times when the code works without a hitch and there are other times when I get this kind of error, so I don't really understand. I need some help, thanks a lot.

**

Код: Выделить всё

    learning_rate = 0.001     model=keras.models.Sequential()     model.add(keras.layers.InputLayer(input_shape=(120,6)))     model.add(keras.layers.SimpleRNN(50))     model.add(keras.layers.Dense(120))     model.compile(optimizer=keras.optimizers.Adam(learning_rate=learning_rate), loss="mse")     model.summary()     epochs = 5     model.fit(dataset_train,     epochs=epochs,     validation_data=dataset_val) 

Источник: https://stackoverflow.com/questions/781 ... l-training