Skip to content

Fix empty indices for gather op #4477

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
BruceXinXin wants to merge 3 commits into
base: main
Choose a base branch
from
+74 −3
Open

Fix empty indices for gather op #4477

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7afe0e2
Fix empty indices for gather op
Feb 28, 2026
3a030bc
Merge remote-tracking branch 'origin/main' into bruce_fix_gather_zero…
Mar 13, 2026
33e3f84
Fix newline
Mar 13, 2026
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
29 changes: 26 additions & 3 deletions lib/Conversion/TorchToStablehlo/GatherScatter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,12 +61,37 @@ Value gatherTensorAlongSingleAxis(PatternRewriter &rewriter, Operation *op,
Value input, Value indices, int64_t axis,
size_t dimSizeIndexBits) {
auto loc = op->getLoc();

auto indicesRankTy = dyn_cast<RankedTensorType>(indices.getType());
auto indicesShape = indicesRankTy.getShape();
auto inputRankTy = dyn_cast<RankedTensorType>(input.getType());

// Check if indices tensor is empty (has any dimension with size 0)
bool isEmpty =
llvm::any_of(indicesShape, [](int64_t dim) { return dim == 0; });

if (isEmpty) {
// Special case: StableHLO doesn't support gather operations on empty
// tensors. Return an empty tensor with the correct output shape.
auto inputShape = inputRankTy.getShape();
SmallVector<int64_t, 4> outputShape(inputShape.begin(),
inputShape.begin() + axis);
outputShape.insert(outputShape.end(), indicesShape.begin(),
indicesShape.end());
outputShape.insert(outputShape.end(), inputShape.begin() + axis + 1,
inputShape.end());

auto outputTy =
RankedTensorType::get(outputShape, inputRankTy.getElementType());
auto emptyAttr = cast<DenseElementsAttr>(rewriter.getZeroAttr(outputTy));
return stablehlo::ConstantOp::create(rewriter, loc, emptyAttr);
}

Type intType = rewriter.getIntegerType(dimSizeIndexBits);
Value one = arith::ConstantOp::create(rewriter, loc,
rewriter.getIntegerAttr(intType, 1));

// sliceSizes
auto inputRankTy = dyn_cast<RankedTensorType>(input.getType());
auto inputRank = inputRankTy.getRank();
SmallVector<Value, 4> sliceSizes;
sliceSizes.reserve(inputRank);
Expand All @@ -88,7 +113,6 @@ Value gatherTensorAlongSingleAxis(PatternRewriter &rewriter, Operation *op,
for (int64_t r = 0; r < axis; ++r) {
offsetDims.push_back(r);
}
auto indicesRankTy = dyn_cast<RankedTensorType>(indices.getType());
auto indicesRank = indicesRankTy.getRank();
for (int64_t r = axis + 1; r < inputRank; ++r) {
offsetDims.push_back(r + indicesRank - 1);
Expand All @@ -112,7 +136,6 @@ Value gatherTensorAlongSingleAxis(PatternRewriter &rewriter, Operation *op,
// outputShape = input.shape[:axis] + indices.shape +
// input.shape[axis + 1:]
auto inputShape = inputRankTy.getShape();
auto indicesShape = indicesRankTy.getShape();
SmallVector<int64_t, 4> outputShape(inputShape.begin(),
inputShape.begin() + axis);
outputShape.insert(outputShape.end(), indicesShape.begin(),
Expand Down
48 changes: 48 additions & 0 deletions test/Conversion/TorchToStablehlo/gather.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,3 +63,51 @@ func.func @torch.aten.embedding$rank_two_indices(%weight: !torch.vtensor<[?,?],f
%ret = torch.aten.embedding %weight, %indices, %int-1, %false, %false : !torch.vtensor<[?,?],f32>, !torch.vtensor<[?,1], si64>, !torch.int, !torch.bool, !torch.bool -> !torch.vtensor<[?,1,?],f32>
return %ret: !torch.vtensor<[?,1,?],f32>
}


// -----

// CHECK-LABEL: func.func @torch.aten.index_select$empty_indices(
// CHECK-SAME: %[[ARG0:.*]]: !torch.vtensor<[4,8],f32>, %[[ARG1:.*]]: !torch.vtensor<[0],si64>) -> !torch.vtensor<[0,8],f32> {
// CHECK: %[[INT0:.*]] = torch.constant.int 0
// CHECK: %[[CST:.*]] = stablehlo.constant dense<0.000000e+00> : tensor<0x8xf32>
// CHECK: %[[CONVERT:.*]] = stablehlo.convert %[[CST]] : tensor<0x8xf32>
// CHECK: %[[RESULT:.*]] = torch_c.from_builtin_tensor %[[CONVERT]] : tensor<0x8xf32> -> !torch.vtensor<[0,8],f32>
// CHECK: return %[[RESULT]] : !torch.vtensor<[0,8],f32>
func.func @torch.aten.index_select$empty_indices(%arg0: !torch.vtensor<[4,8],f32>, %arg1: !torch.vtensor<[0],si64>) -> !torch.vtensor<[0,8],f32> {
%int0 = torch.constant.int 0
%0 = torch.aten.index_select %arg0, %int0, %arg1 : !torch.vtensor<[4,8],f32>, !torch.int, !torch.vtensor<[0],si64> -> !torch.vtensor<[0,8],f32>
return %0 : !torch.vtensor<[0,8],f32>
}

// -----

// CHECK-LABEL: func.func @torch.aten.embedding$empty_indices(
// CHECK-SAME: %[[ARG0:.*]]: !torch.vtensor<[10,8],f32>, %[[ARG1:.*]]: !torch.vtensor<[0],si64>) -> !torch.vtensor<[0,8],f32> {
// CHECK: %[[FALSE:.*]] = torch.constant.bool false
// CHECK: %[[INT_NEG1:.*]] = torch.constant.int -1
// CHECK: %[[CST:.*]] = stablehlo.constant dense<0.000000e+00> : tensor<0x8xf32>
// CHECK: %[[CONVERT:.*]] = stablehlo.convert %[[CST]] : tensor<0x8xf32>
// CHECK: %[[RESULT:.*]] = torch_c.from_builtin_tensor %[[CONVERT]] : tensor<0x8xf32> -> !torch.vtensor<[0,8],f32>
// CHECK: return %[[RESULT]] : !torch.vtensor<[0,8],f32>
func.func @torch.aten.embedding$empty_indices(%weight: !torch.vtensor<[10,8],f32>, %indices: !torch.vtensor<[0], si64>) -> !torch.vtensor<[0,8],f32> {
%false = torch.constant.bool false
%int-1 = torch.constant.int -1
%ret = torch.aten.embedding %weight, %indices, %int-1, %false, %false : !torch.vtensor<[10,8],f32>, !torch.vtensor<[0], si64>, !torch.int, !torch.bool, !torch.bool -> !torch.vtensor<[0,8],f32>
return %ret: !torch.vtensor<[0,8],f32>
}

// -----

// CHECK-LABEL: func.func @torch.aten.index_select$empty_indices_dim1(
// CHECK-SAME: %[[ARG0:.*]]: !torch.vtensor<[4,8],f32>, %[[ARG1:.*]]: !torch.vtensor<[0],si64>) -> !torch.vtensor<[4,0],f32> {
// CHECK: %[[INT1:.*]] = torch.constant.int 1
// CHECK: %[[CST:.*]] = stablehlo.constant dense<0.000000e+00> : tensor<4x0xf32>
// CHECK: %[[CONVERT:.*]] = stablehlo.convert %[[CST]] : tensor<4x0xf32>
// CHECK: %[[RESULT:.*]] = torch_c.from_builtin_tensor %[[CONVERT]] : tensor<4x0xf32> -> !torch.vtensor<[4,0],f32>
// CHECK: return %[[RESULT]] : !torch.vtensor<[4,0],f32>
func.func @torch.aten.index_select$empty_indices_dim1(%arg0: !torch.vtensor<[4,8],f32>, %arg1: !torch.vtensor<[0],si64>) -> !torch.vtensor<[4,0],f32> {
%int1 = torch.constant.int 1
%0 = torch.aten.index_select %arg0, %int1, %arg1 : !torch.vtensor<[4,8],f32>, !torch.int, !torch.vtensor<[0],si64> -> !torch.vtensor<[4,0],f32>
return %0 : !torch.vtensor<[4,0],f32>
}