meanSquaredError Of Two int32 Tensors Fails

[dsanders11]

To get help from the community, check out our Google group.

TensorFlow.js version

0.13.3

Describe the problem or feature request

const tensor1 = tf.fromPixels(canvas, 1);
const tensor2 = tf.fromPixels(canvas, 1);

const mse = tf.losses.meanSquaredError(tensor1, tensor2);

Results in:

Error: The dtypes of the first(int32) and second(float32) input must match

This does work:

const mse = tf.losses.meanSquaredError(tensor1.toFloat(), tensor2.toFloat());

Code to reproduce the bug / link to feature request

http://jsfiddle.net/5c94d3tf/