#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 {
at::Tensor& nll_loss_backward_out_nocheck(
at::Tensor& grad_input,
const at::Tensor& grad_output,
const at::Tensor& self,
const at::Tensor& target,
const at::Tensor& weight_tensor,
int64_t reduction,
int64_t ignore_index,
const at::Tensor& total_weight)
{
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_loss_backward"
+ OPS_ERROR(ErrCode::TYPE));
at::Tensor target_cast = (scalar_type == at::kLong) ? at_npu::native::custom_ops::_npu_dtype_cast(target, at::kInt) : target;
string reduction_str = op_plugin::utils::get_reduction_str(reduction);
at_npu::native::OpCommand cmd;
cmd.Name("NLLLossGrad")
.Input(self)
.Input(grad_output)
.Input(target_cast)
.Input(weight_tensor)
.Input(total_weight)
.Output(grad_input)
.Attr("reduction", reduction_str)
.Attr("ignore_index", ignore_index)
.Run();
if (self.dim() == 1) {
grad_input.squeeze_(0);
}
return grad_input;
}
}
at::Tensor& nll_loss_backward_out(
const at::Tensor& grad_output,
const at::Tensor& self,
const at::Tensor& target,
const c10::optional<at::Tensor>& weight,
int64_t reduction,
int64_t ignore_index,
const at::Tensor& total_weight,
at::Tensor& grad_input)
{
at::Tensor self_cp = self.dim() == 1 ? self.unsqueeze(0) : self;
const at::Tensor& weight_value_or = c10::value_or_else(weight, [] {return at::Tensor();});
at::Tensor weight_tensor = at::ones(self_cp.size(1), self_cp.options());
if (weight_value_or.defined()) {
weight_tensor = npu_utils::format_contiguous(weight_value_or);
}
if (ignore_index >= 0 && ignore_index < self_cp.size(-1)) {
at::Tensor zero = at::zeros(1, self_cp.options());
calcu_op_util::AclrtMemcpyAsync(
{weight_tensor, ignore_index},
weight_tensor.itemsize(),
{zero, 0},
weight_tensor.itemsize(),
ACL_MEMCPY_DEVICE_TO_DEVICE);
}
npu_preparation::CheckOut(
{self_cp},
grad_input,
self_cp);
if (!npu_utils::check_match(&grad_input)) {
at::Tensor contiguous_grad_input = npu_utils::format_contiguous(grad_input);
nll_loss_backward_out_nocheck(contiguous_grad_input, grad_output, self_cp, target, weight_tensor, reduction,
ignore_index, total_weight);
npu_utils::format_fresh_view(grad_input, contiguous_grad_input);
} else {
nll_loss_backward_out_nocheck(grad_input, grad_output, self_cp, target, weight_tensor, reduction, ignore_index,
total_weight);
}
return grad_input;
}
at::Tensor nll_loss_backward(
const at::Tensor& grad_output,
const at::Tensor& self,
const at::Tensor& target,
const c10::optional<at::Tensor>& weight,
int64_t reduction,
int64_t ignore_index,
const at::Tensor& total_weight)
{
at::Tensor grad_input = npu_preparation::apply_tensor(self);
acl_op::nll_loss_backward_out(grad_output, self, target, weight, reduction, ignore_index,
total_weight, grad_input);
return grad_input;
}
}