#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/utils/OpAdapter.h"
namespace acl_op {
using npu_preparation = at_npu::native::OpPreparation;
using calcu_op_util = at_npu::native::CalcuOpUtil;
using npu_utils = at_npu::native::NpuUtils;
namespace {
std::tuple<c10::SmallVector<int64_t, SIZE>, c10::SmallVector<int64_t, SIZE>> nll_loss2d_npu_output_size(
const at::Tensor &self, int64_t reduction)
{
c10::SmallVector<int64_t, SIZE> output_size;
c10::SmallVector<int64_t, SIZE> total_weight_size;
if (reduction == at::Reduction::None) {
output_size = {self.size(0)};
}
return std::tuple<c10::SmallVector<int64_t, SIZE>, c10::SmallVector<int64_t, SIZE>>(output_size, total_weight_size);
}
at::Tensor check_weight_opt(const at::Tensor &self, const c10::optional<at::Tensor> &weight_opt, int64_t ignore_index)
{
at::Tensor weight = c10::value_or_else(weight_opt, [] { return at::Tensor(); });
at::Tensor weight_tensor = at::ones(self.size(1), self.options());
if (weight.defined()) {
weight_tensor = npu_utils::format_contiguous(weight);
}
if (ignore_index >= 0 && ignore_index < self.size(1)) {
at::Tensor zero = at::zeros(1, self.options());
calcu_op_util::AclrtMemcpyAsync({weight_tensor, ignore_index}, weight_tensor.itemsize(), {zero, 0},
weight_tensor.itemsize(), ACL_MEMCPY_DEVICE_TO_DEVICE);
}
return weight_tensor;
}
std::tuple<at::Tensor &, at::Tensor &> nll_loss2d_forward_out_nocheck(at::Tensor &result, at::Tensor &total_weight,
const at::Tensor &self, const at::Tensor &target,
const at::Tensor &weight_tensor,
int64_t reduction, int64_t ignore_index)
{
auto reduction_str = op_plugin::utils::get_reduction_str(reduction);
at_npu::native::OpCommand cmd;
cmd.Name("NLLLoss")
.Input(self)
.Input(target)
.Input(weight_tensor)
.Attr("reduction", reduction_str)
.Attr("ignore_index", ignore_index)
.Output(result)
.Output(total_weight)
.Run();
acl_op::npu_reshape_out(result, result.sizes(), true, result);
return std::tuple<at::Tensor &, at::Tensor &>(result, total_weight);
}
}
std::tuple<at::Tensor &, at::Tensor &> nll_loss2d_forward_out(const at::Tensor &self, const at::Tensor &target,
const c10::optional<at::Tensor> &weight,
int64_t reduction, int64_t ignore_index,
at::Tensor &output, at::Tensor &total_weight)
{
at::Tensor weight_tensor = check_weight_opt(self, weight, ignore_index);
auto output_sizes = nll_loss2d_npu_output_size(self, reduction);
npu_preparation::CheckOut({self, target, weight_tensor}, output, ACL_FORMAT_ND, self.scalar_type(),
std::get<0>(output_sizes));
npu_preparation::CheckOut({self, target, weight_tensor}, total_weight, ACL_FORMAT_ND, self.scalar_type(),
std::get<1>(output_sizes));
bool result_match = npu_utils::check_match(&output);
bool total_weight_match = npu_utils::check_match(&total_weight);
if (!(result_match && total_weight_match)) {
at::Tensor contiguous_result = result_match ? output : npu_utils::format_contiguous(output);
at::Tensor contiguous_total_weight =
total_weight_match ? total_weight : npu_utils::format_contiguous(total_weight);
nll_loss2d_forward_out_nocheck(contiguous_result, contiguous_total_weight, self, target, weight_tensor,
reduction, ignore_index);
if (!result_match) {
npu_utils::format_fresh_view(output, contiguous_result);
}
if (!total_weight_match) {
npu_utils::format_fresh_view(total_weight, contiguous_total_weight);
}
} else {
nll_loss2d_forward_out_nocheck(output, total_weight, self, target, weight_tensor, reduction, ignore_index);
}
return std::tuple<at::Tensor &, at::Tensor &>(output, total_weight);
}
std::tuple<at::Tensor, at::Tensor> nll_loss2d_forward(const at::Tensor &self, const at::Tensor &target,
const c10::optional<at::Tensor> &weight, int64_t reduction,
int64_t ignore_index)
{
TORCH_CHECK(self.dim() == 4, "Expected 4D input (got ", self.dim(), "D input)"
+ OPS_ERROR(ErrCode::PARAM));
auto scalar_type = target.scalar_type();
TORCH_CHECK(scalar_type == at::kLong || scalar_type == at::kInt, "Expected object of scalar type ", at::kLong,
" or ", at::kInt, " but got scalar type ", scalar_type,
" for argument 'target' in call to nll_loss2d_forward"
+ OPS_ERROR(ErrCode::TYPE));
at::Tensor target_cast =
(scalar_type == at::kLong) ? at_npu::native::custom_ops::_npu_dtype_cast(target, at::kInt) : target;
auto self_input = self.contiguous();
self_input = at_npu::native::custom_ops::npu_format_cast(self_input, ACL_FORMAT_ND);
self_input = self_input.permute({0, 2, 3, 1});
self_input = self_input.reshape({-1, self.size(1)});
auto target_input = target_cast.contiguous();
target_input = target_cast.reshape({-1});
auto output_sizes = nll_loss2d_npu_output_size(self_input, reduction);
at::Tensor result = npu_preparation::apply_tensor(self_input, std::get<0>(output_sizes));
at::Tensor total_weight = npu_preparation::apply_tensor(self_input, std::get<1>(output_sizes));
at::Tensor weight_tensor = check_weight_opt(self, weight, ignore_index);
nll_loss2d_forward_out_nocheck(result, total_weight, self_input, target_input, weight_tensor, reduction,
ignore_index);
if (reduction == at::Reduction::None) {
result.resize_({self.size(0), self.size(2), self.size(3)});
}
return std::tuple<at::Tensor, at::Tensor>(result, total_weight);
}
}