projects / platform / upstream / tensorflow.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: c8f5b35) | patch
[XLA][BF16] Add bf16 rounding function.
authorYunxing Dai <yunxing@google.com>
Wed, 21 Mar 2018 21:12:07 +0000 (14:12 -0700)
committerTensorFlower Gardener <gardener@tensorflow.org>
Wed, 21 Mar 2018 21:14:40 +0000 (14:14 -0700)
commit4c56dcaade7cee9f75740e2f03e30bdf3a6f93be
treed3043ef31f16a57131200f1c240c8f67c5e1571ftree | snapshot
parentc8f5b35dd70d103fe8ae6a41ca907ccccaa3ac39commit | diff
[XLA][BF16] Add bf16 rounding function.

We now use truncation to convert a F32 to BF16 by default. This CL adds a rounding method (basically a roll forward of the rounding part in cl/175252067).

PiperOrigin-RevId: 189965138
tensorflow/core/framework/bfloat16_test.cc diff | blob | history
tensorflow/core/lib/bfloat16/bfloat16.h diff | blob | history
Domain: Machine Learning / ML Framework;