OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-06 12:20:52 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

[MPS] Fix smooth_l1_loss backward for fp16 (#166687)

Browse Source 
And enable fp16 implementation for CPU, which simplifies OpInfo definitions for the op

Pull Request resolved: https://github.com/pytorch/pytorch/pull/166687
Approved by: https://github.com/Skylion007
ghstack dependencies: #166214
This commit is contained in:
Nikita Shulga 2025-10-31 14:07:23 -07:00 committed by PyTorch MergeBot
parent 93a70c717a
commit 4e7232c5da
4 changed files with 7 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
aten/src/ATen/native/cpu/PointwiseOpsKernel.cpp
View File

@ -139,7 +139,7 @@ void smooth_l1_backward_cpu_kernel(TensorIterator& iter, const Scalar& norm, dou
} }
); );
} else { } else {
AT_DISPATCH_ALL_TYPES(dtype, "smooth_l1_backward_cpu_out", [&] { AT_DISPATCH_ALL_TYPES_AND(kHalf, dtype, "smooth_l1_backward_cpu_out", [&] {
auto norm_val = norm.to<scalar_t>(); auto norm_val = norm.to<scalar_t>();
scalar_t beta_val(beta); scalar_t beta_val(beta);
auto norm_val_vec = Vectorized<scalar_t>(norm_val); auto norm_val_vec = Vectorized<scalar_t>(norm_val);

8
aten/src/ATen/native/mps/operations/LossOps.mm
View File

@ -370,7 +370,7 @@ static void nllnd_loss_backward_impl(Tensor& grad_input_arg,
onValue:-1.0f onValue:-1.0f
offValue:0.0f offValue:0.0f
name:nil]; name:nil];
oneHotTensor = castMPSTensor(mpsGraph, oneHotTensor, inputTensor.dataType); oneHotTensor = castMPSTensor(mpsGraph, oneHotTensor, [inputTensor dataType]);
if (isWeightsArrayValid) { if (isWeightsArrayValid) {
oneHotTensor = [mpsGraph multiplicationWithPrimaryTensor:oneHotTensor oneHotTensor = [mpsGraph multiplicationWithPrimaryTensor:oneHotTensor
secondaryTensor:weightTensor secondaryTensor:weightTensor
@ -705,6 +705,7 @@ static void smooth_l1_loss_template(const Tensor& input,
TORCH_CHECK(beta >= 0, "smooth_l1_loss does not support negative values for beta."); TORCH_CHECK(beta >= 0, "smooth_l1_loss does not support negative values for beta.");
TORCH_CHECK(input.is_mps()); TORCH_CHECK(input.is_mps());
TORCH_CHECK(target.is_mps()); TORCH_CHECK(target.is_mps());
TORCH_CHECK_NOT_IMPLEMENTED(input.scalar_type() != kLong, "MPS doesn't know how to do square_i64");
if ((input.numel() == 0) || (target.numel() == 0)) { if ((input.numel() == 0) || (target.numel() == 0)) {
reduction == Reduction::Mean ? output.fill_(std::numeric_limits<float>::quiet_NaN()) : output.zero_(); reduction == Reduction::Mean ? output.fill_(std::numeric_limits<float>::quiet_NaN()) : output.zero_();
return; return;
@ -771,7 +772,7 @@ static void smooth_l1_loss_backward_impl(const Tensor& grad_output,
MPSGraphTensor* targetTensor = mpsGraphRankedPlaceHolder(mpsGraph, target); MPSGraphTensor* targetTensor = mpsGraphRankedPlaceHolder(mpsGraph, target);
MPSGraphTensor* gradOutputTensor = mpsGraphRankedPlaceHolder(mpsGraph, grad_output); MPSGraphTensor* gradOutputTensor = mpsGraphRankedPlaceHolder(mpsGraph, grad_output);
MPSGraphTensor* betaTensor = [mpsGraph constantWithScalar:beta dataType:MPSDataTypeFloat32]; MPSGraphTensor* betaTensor = [mpsGraph constantWithScalar:beta dataType:[inputTensor dataType]];
// xn - yn // xn - yn
MPSGraphTensor* diffTensor = [mpsGraph subtractionWithPrimaryTensor:inputTensor MPSGraphTensor* diffTensor = [mpsGraph subtractionWithPrimaryTensor:inputTensor
secondaryTensor:targetTensor secondaryTensor:targetTensor
@ -797,7 +798,8 @@ static void smooth_l1_loss_backward_impl(const Tensor& grad_output,
name:@"lossTensor"]; name:@"lossTensor"];
MPSGraphTensor* outputTensor = lossTensor; MPSGraphTensor* outputTensor = lossTensor;
if (reduction == Reduction::Mean) { if (reduction == Reduction::Mean) {
MPSGraphTensor* numelTensor = [mpsGraph constantWithScalar:(double)input.numel() dataType:MPSDataTypeFloat32]; MPSGraphTensor* numelTensor = [mpsGraph constantWithScalar:(double)input.numel()
dataType:[lossTensor dataType]];
outputTensor = [mpsGraph divisionWithPrimaryTensor:lossTensor secondaryTensor:numelTensor name:nil]; outputTensor = [mpsGraph divisionWithPrimaryTensor:lossTensor secondaryTensor:numelTensor name:nil];
} }
MPSGraphTensor* gradInputTensor = [mpsGraph multiplicationWithPrimaryTensor:outputTensor MPSGraphTensor* gradInputTensor = [mpsGraph multiplicationWithPrimaryTensor:outputTensor

4
torch/testing/_internal/common_methods_invocations.py
View File

@ -20340,9 +20340,7 @@ op_db: list[OpInfo] = [
ref=reference_smooth_l1_loss, ref=reference_smooth_l1_loss,
sample_inputs_func=sample_inputs_smooth_l1_loss, sample_inputs_func=sample_inputs_smooth_l1_loss,
dtypes=floating_types_and(torch.float16, torch.bfloat16), dtypes=floating_types_and(torch.float16, torch.bfloat16),
backward_dtypes=floating_types_and(torch.bfloat16), backward_dtypes=floating_types_and(torch.float16, torch.bfloat16),
dtypesIfCUDA=floating_types_and(torch.float16, torch.bfloat16),
backward_dtypesIfCUDA=floating_types_and(torch.float16, torch.bfloat16),
supports_out=False, supports_out=False,
supports_forward_ad=True, supports_forward_ad=True,
supports_fwgrad_bwgrad=True, supports_fwgrad_bwgrad=True,

2
torch/testing/_internal/common_mps.py
View File

@ -740,8 +740,6 @@ if torch.backends.mps.is_available():
"equal": [torch.float16, torch.float32], "equal": [torch.float16, torch.float32],
# 'float' object is not iterable # 'float' object is not iterable
"item": [torch.float16, torch.float32], "item": [torch.float16, torch.float32],
# "smooth_l1_backward_cpu_out" not implemented for 'Half'
"nn.functional.smooth_l1_loss": [torch.float16],
# cpu error: grad requires non-empty inputs # cpu error: grad requires non-empty inputs
"randn": [torch.float16, torch.float32], "randn": [torch.float16, torch.float32],
"signal.windows.bartlett": [torch.float32], "signal.windows.bartlett": [torch.float32],