Create different data for each Literal when creating fake data.
authorMark Heffernan <meheff@google.com>
Sat, 27 Jan 2018 01:50:03 +0000 (17:50 -0800)
committerTensorFlower Gardener <gardener@tensorflow.org>
Sat, 27 Jan 2018 01:53:49 +0000 (17:53 -0800)
commit704361ad3650ebc891167adc41c459ca93392060
tree5152b71fb9ed92168cf439d60827671f2c77e910
parentca3ac2a464b92f4c0498dfde875f99102a0d410c
Create different data for each Literal when creating fake data.
Thread a generator through the functions for creating fake arguments so the same
generator can be reused which avoids repeating the same data patterns for each
argument generated.

Also tweak the position-dependent biasing heuristic to create both positive and
negative numbers for small literals.

PiperOrigin-RevId: 183473588
tensorflow/compiler/xla/tests/test_utils.cc