Commit message (Expand) | Author | Age | |
---|---|---|---|
* | Add int64 support for `tf.reduce_max` on GPU | 2017-09-26 | |
* | Switch the softmax to use the new deterministic reductions on the GPU, | 2017-09-13 | |
* | Split reduction template instantiations into multiple files for faster | 2017-09-07 |