Assign a name to a tensor?

Question

Is there any way to assign a name to a tensor after it has been created?

I'm building up a neural network bit by bit in a loop, like so:

    def build_logit_pipeline(data):
        # X --> *W1 --> +b1 --> relu --> *W2 --> +b2 ... --> softmax etc...
        pipeline = data

        for i in xrange(len(layer_sizes) - 1):
            with tf.name_scope("linear%d" % i):
                pipeline = tf.matmul(pipeline, weights[i])
                pipeline = tf.add(pipeline, biases[i])

            if i != len(layer_sizes) - 2:
                with tf.name_scope("relu%d" % i):
                    pipeline = tf.nn.relu(pipeline)

        return pipeline

I'd like to assign a name to the result of the entire operation, i.e. the last tf.add should be assigned a name. Is there a way to do this on the pipeline variable right before returning, instead of checking the end of loop condition on the last add, which would be less elegant?

Answer 1

You cannot modify the name of a Tensor.

However, one easy trick to do what you want is to use tf.identity with the name you want:

res = build_logit_pipeline(data)
res = tf.identity(res, name="your_name")