Abdelrauf
|
69d91e272a
|
- new implementations for Index Reductions (#421)
* - new implementations for Index Reductions
- small fix in the legacy reduction
- disabled index reduction bench tests inside Playground
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* Allow LIBND4J_TYPES
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* index reduction stuff split into bunch of units
* meh
* IMax switched to new impl
Signed-off-by: raver119@gmail.com <raver119@gmail.com>
* minor fix + test
* minor fix
* index range fix
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* noop on empty outputs
* minor fix
* minor fix
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* ArgMax replaces IMax
Signed-off-by: raver119@gmail.com <raver119@gmail.com>
* argmax/argmin/argamax/argamin shape functions updated
* ArgAmax/ArgAmin/ArgMin replaces IAMax/IAMin/IMin
Signed-off-by: raver119@gmail.com <raver119@gmail.com>
* argmax/argmin/argamax/argamin CUDA
* IMax replaced in dl4j
Signed-off-by: raver119@gmail.com <raver119@gmail.com>
* Codegen output
* imports fixed
Signed-off-by: raver119@gmail.com <raver119@gmail.com>
* fix compilation issue
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* Auto-generate compilation units
Signed-off-by: Abdelrauf <rauf@konduit.ai>
* Should fix NDArray refactored function calls in indexReductions.cu
Signed-off-by: Abdelrauf <rauf@konduit.ai>
Co-authored-by: raver119@gmail.com <raver119@gmail.com>
Co-authored-by: Alexander Stoyakin <alexander.stoyakin@gmail.com>
|
2020-05-14 13:41:55 +03:00 |