[MXNET-392] Support for a bunch of unary functions for csr matrices #11559
Conversation
| @@ -70,7 +70,7 @@ static bool IdentityAttrLikeRhsStorageType(const nnvm::NodeAttrs& attrs, | |||
| } | |||
|
|
|||
| // relu | |||
| MXNET_OPERATOR_REGISTER_UNARY(relu) | |||
| MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP_CSR(relu, cpu, mshadow_op::relu) | |||
There was a problem hiding this comment.
We probably want to move MXNET_ADD_SPARSE_OP_ALIAS inside MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP_CSR. Do you mind changing that?
There was a problem hiding this comment.
Actually none of MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP_CSR/MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP/MXNET_OPERATOR_REGISTER_UNARY_WITH_SPARSE_DR has MXNET_ADD_SPARSE_OP_ALIAS in them, do you want to add that to all of them?
There was a problem hiding this comment.
Add for MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP_CSR and MXNET_OPERATOR_REGISTER_UNARY_WITH_RSP. Lets skip the ones with SPARSE_DR
| @@ -81,11 +81,9 @@ The storage type of ``relu`` output depends upon the input storage type: | |||
|
|
|||
| - relu(default) = default | |||
| - relu(row_sparse) = row_sparse | |||
| - relu(csr) = csr | |||
There was a problem hiding this comment.
Please add fluent methods to CSRNDArray documentation like what was did for RowSparseNDArray https://github.com/apache/incubator-mxnet/blob/master/docs/api/python/ndarray/sparse.md#trigonometric-functions
And add tests in https://github.com/apache/incubator-mxnet/blob/master/tests/python/unittest/test_sparse_ndarray.py#L847
haojin2
force-pushed
the
csr_unary
branch
2 times, most recently
from
73aefaf
to
ceb0fed
Compare
docs/api/python/ndarray/sparse.md
Outdated
| @@ -520,7 +583,7 @@ We summarize the interface for each class in the following sections. | |||
| ```eval_rst | |||
|
|
|||
| .. autoclass:: mxnet.ndarray.sparse.CSRNDArray | |||
| :members: shape, context, dtype, stype, data, indices, indptr, copy, copyto, as_in_context, asscipy, asnumpy, asscalar, astype, tostype, slice, wait_to_read, zeros_like, __neg__, sum, mean, norm, square, __getitem__, __setitem__, check_format | |||
| :members: shape, context, dtype, stype, data, indices, indptr, copy, copyto, as_in_context, asscipy, asnumpy, asscalar, astype, tostype, slice, wait_to_read, zeros_like, round, rint, fix, floor, ceil, trunc, sin, tan, arcsin, arctan, degrees, radians, sinh, tanh, arcsinh, arctanh, expm1, log1p, sqrt, square, __neg__, sum, mean, norm, square, __getitem__, __setitem__, check_format, clip, sign | |||
|
|
|||
| .. autoclass:: mxnet.ndarray.sparse.RowSparseNDArray | |||
| :members: shape, context, dtype, stype, data, indices, copy, copyto, as_in_context, asnumpy, asscalar, astype, tostype, wait_to_read, zeros_like, round, rint, fix, floor, ceil, trunc, sin, tan, arcsin, arctan, degrees, radians, sinh, tanh, arcsinh, arctanh, expm1, log1p, sqrt, square, __negative__, norm, __getitem__, __setitem__, check_format, retain, clip, sign | |||
There was a problem hiding this comment.
Also add abs for RowSparse: members?
There was a problem hiding this comment.
There was a problem hiding this comment.
Oh you mean this one, will do soon.
…pache#11559) * support for relu(csr) = csr * add support for sign(csr) = csr * add support for round * add support for ceil(csr)=csr * add support for floor(csr)=csr * add support for trunc(csr)=csr * add support for fix(csr)=csr * add support for sqrt(csr)=csr * add support for cbrt(csr)=csr * add support for log1p(csr)=csr * add support for expm1(csr)=csr * add support for sin(csr)=csr * add support for tan(csr)=csr * add support for arcsin(csr)=csr * add support for arctan(csr)=csr * add support for degrees(csr)=csr * add support for radians(csr)=csr * add support for sinh(csr)=csr * add support for tanh(csr)=csr * add support for arcsinh(csr)=csr * add support for arctanh(csr)=csr * restore original check_sparse_function func * add test for fluent and doc * address code review
Description
As title
Checklist
Essentials
Changes
Comments