Tensorflow basic_rnn_seq2seq TypeError: Expected int32, got -0.1 of type 'float' instead

Question

I am trying to run the encoder-decoder model on the dataset. Below is the sample code:

self._input_data = tf.placeholder(tf.int32, [batch_size, num_steps])
self._targets = tf.placeholder(tf.int32, [batch_size, num_steps])
enc_inputs.append(self._input_data) #one batch at once
dec_inputs.append(self._targets)
model = seq2seq.basic_rnn_seq2seq(enc_inputs, dec_inputs, tf.nn.rnn_cell.BasicLSTMCell(size, state_is_tuple=True))

I get an error of type mismatch (mentioned below). Does anyone know to solve the issue?

tensor_util.py, line 290, in _AssertCompatible
    (dtype.name, repr(mismatch), type(mismatch).__name__))
    TypeError: Expected int32, got -0.1 of type 'float' instead.

Solution

This is an issue of confusing error message. The actual cause is, when you call tf.get_variable() but do not set the default initializer, the error message will be confusing. You can use a tf.zero_initializer() or something like that to suppress this error.

Comments

[user3480922]

Seems, changing tf.int32 to tf.float32 solves the issue. I don't know why but the program runs error-free now.

[Paul Tucker]

The type error arises when you try to insert a float value into a tensor that has been typed as int32. tensor_util.py is used to construct tensors within python, prior to sending data or graph to the backend client for execution. It's not clear from your question exactly how to reproduce your error. How is _input_data being initialized? With values that include a float?

[user3480922]

encode_input and decode_input are simple the word ids (integers). Next all processes are handled inside the basic_rnn_seq2seq API.