[RELAY]reshape_like #1950
[RELAY]reshape_like #1950
Conversation
siju-samuel
force-pushed
the
relay_reshape_like
branch
from
a770faf to
dd0f642
Compare
src/relay/ir/type.cc
Outdated
| } | ||
|
|
||
| IndexExpr TensorTypeNode::Size() const { | ||
| IndexExpr size = make_const(Int(64), 1); |
There was a problem hiding this comment.
directly product from shape[0], when the shape.size() equals 0, return make_const(Int(32), 1);
|
@siju-samuel please request reviews from https://github.com/dmlc/tvm/blob/master/CONTRIBUTORS.md#reviewers |
|
@MarisaKirisame @slyubomirsky @nishi-t @junrushao1994 could you please help to review this PR. Thanks. |
src/relay/op/tensor/transform.cc
Outdated
| * \param num_inputs Number of input types in the args. | ||
| * \param attrs The additional attributes of the operator. | ||
| * \param reporter The reporter to report solution to. | ||
| * \return false if This relation cannot be resolved. true if this relation has been resolved. |
There was a problem hiding this comment.
cannot be -> has not been
it might be resolved later
src/relay/op/tensor/transform.cc
Outdated
|
|
||
|
|
||
| /*! | ||
| * \brief ReshapeLikeRel Output type and shape relation evaluation function. |
There was a problem hiding this comment.
I dont really get this line
siju-samuel
force-pushed
the
relay_reshape_like
branch
from
354586f to
52041cf
Compare
include/tvm/relay/type.h
Outdated
| v->Visit("span", &span); | ||
| } | ||
|
|
||
| /*! \brief Return product of elements in the shape */ |
There was a problem hiding this comment.
Could we add one more line to the docs such as (d_1, d_2, ..., d_n) -> d1 * d_2 ... * d_n.
Pytorch does a good job of this and I think its useful for users when reading the code.
siju-samuel
force-pushed
the
relay_reshape_like
branch
from
797b5ea to
57bfe43
Compare
|
cc @tqchen could you please check once. |
|
Thanks @siju-samuel @MarisaKirisame @jroesch this is now merged |
4 participants
#1799
Thanks for contributing to TVM! Please refer to guideline https://docs.tvm.ai/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from others in the community.