Allow num parameter in tf.linspace to be int64. (#13755)
* Allow num parameter in tf.linspace to be int64. Currently tf.linspace is defined for num int32 or int64. However the kernel only allows int32, even though the op in core/ops/math_ops permits int64 too. I slightly changed the kernel to allow int64 too. I also added tests for RangeOp and LinSpaceOp. * Change variable names.
Loading
Please sign in to comment