Skip to content

Support double in optimized log_softmax #23686

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
usmanmateen wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Support double in optimized log_softmax #23686

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 6 additions & 2 deletions kernels/optimized/cpu/op_log_softmax.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,6 +102,7 @@ void log_softmax_kernel(const Tensor& input, int64_t dim, Tensor& out) {
template <typename OUT_T>
bool log_softmax_wrapper(const Tensor& X, int64_t dim, Tensor& out) {
if constexpr (
std::is_same_v<OUT_T, double> ||
std::is_same_v<OUT_T, executorch::aten::BFloat16> ||
std::is_same_v<OUT_T, executorch::aten::Half>) {
// Input dtype equals output dtype (enforced by check_log_softmax_args).
Expand All @@ -112,7 +113,6 @@ bool log_softmax_wrapper(const Tensor& X, int64_t dim, Tensor& out) {
} else {
auto input_scalar_type = X.scalar_type();
switch (input_scalar_type) {
// TODO: support Double as well
case ScalarType::Float:
log_softmax_kernel<float, OUT_T>(X, dim, out);
return true;
Expand Down Expand Up @@ -149,12 +149,16 @@ Tensor& opt_log_softmax_out(

auto out_scalar_type = out.scalar_type();
switch (out_scalar_type) {
// TODO: support Double as well
case ScalarType::Float: {
bool success = log_softmax_wrapper<float>(self, dim, out);
ET_KERNEL_CHECK(context, success, InvalidArgument, out);
break;
}
case ScalarType::Double: {
bool success = log_softmax_wrapper<double>(self, dim, out);
ET_KERNEL_CHECK(context, success, InvalidArgument, out);
break;
}
case ScalarType::BFloat16: {
bool success =
log_softmax_wrapper<executorch::aten::BFloat16>(self, dim, out);
Expand Down
3 changes: 0 additions & 3 deletions kernels/optimized/test/supported_features_def.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,2 @@
- namespace: op_gelu
dtype_double: false

- namespace: op_log_softmax
dtype_double: false
Loading