Invalid arguments while using QNetwork

[bazylip]

I am implementing Wumpus World in Tensorflow using DQN with discrete action space and encounter following error while trying to collect data using greedy policy
tensorflow.python.framework.errors_impl.InvalidArgumentError: 'then' and 'else' must have the same size. but received: [1,1] vs. [1] [Op:Select]
raised by this line in EpsilonGreedyPolicy
action = tf.compat.v1.where(cond, greedy_action.action, random_action.action)
I reckon this is the same issue as in #255 and #253 but applying workarounds mentioned there to RandomTFPolicy does not work and the error still remains the same. However, when I change network which I am using from:

class QNetwork(q_network.QNetwork):

    def __init__(self, input_tensor_spec, action_spec, num_actions, name=None):
        super(QNetwork, self).__init__(
            input_tensor_spec=input_tensor_spec,
            action_spec=action_spec,
            name=name)
        self._sub_layers = [
            tf.keras.layers.Dense(num_actions),
        ]

    def call(self, inputs, step_type=None, network_state=(), **kwargs):
        del step_type
        inputs = tf.cast(inputs, tf.float32)
        for layer in self._sub_layers:
            inputs = layer(inputs)
        return inputs, network_state

to simply:

q_net = q_network.QNetwork(
            train_env.observation_spec(),
            train_env.action_spec(),
            fc_layer_params=(100,))

then the exception is not raised. Of course such network is not what I am looking for to base my agent's policy on, so I would be very thankful for any suggestions what might be wrong in my code.

[tfboyd]

Closing due to age and lack of recent activity. If the issue still exists, please open a new issue and reference this one.