[torch] Add support for aten.selu (#2640)

JianzheXiao · web-flow · commit 6ddeb1a6efe7 · 2023-12-13T20:28:08.000-08:00
Add `aten.selu` operation to `torch` dialect.
diff --git a/include/torch-mlir/Dialect/Torch/IR/GeneratedTorchOps.td b/include/torch-mlir/Dialect/Torch/IR/GeneratedTorchOps.td
@@ -346,6 +346,51 @@ def Torch_AtenLog_Op : Torch_Op<"aten.log_", [
   }];
 }
 
+def Torch_AtenSeluOp : Torch_Op<"aten.selu", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::selu : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self
+  );
+  let results = (outs
+    AnyTorchTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenSeluOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenSeluOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
+def Torch_AtenSelu_Op : Torch_Op<"aten.selu_", [
+    IsTrailingUnderscoreInplaceVariant,
+    AllowsTypeRefinement
+  ]> {
+  let summary = "Generated op for `aten::selu_ : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    Torch_NonValueTensorType:$self
+  );
+  let results = (outs
+    Torch_NonValueTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenSelu_Op::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenSelu_Op::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
 def Torch_AtenSigmoidOp : Torch_Op<"aten.sigmoid", [
     AllowsTypeRefinement,
     HasValueSemantics,
diff --git a/lib/Dialect/Torch/Transforms/AbstractInterpLibrary.cpp b/lib/Dialect/Torch/Transforms/AbstractInterpLibrary.cpp
@@ -6746,6 +6746,10 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg0) : (!torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
 "  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.selu\"(%arg0: !torch.list<int>) -> !torch.list<int> {\n"
+"    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg0) : (!torch.list<int>) -> !torch.list<int>\n"
+"    return %0 : !torch.list<int>\n"
+"  }\n"
 "  func.func @\"__torch_mlir_shape_fn.aten.gather\"(%arg0: !torch.list<int>, %arg1: !torch.int, %arg2: !torch.list<int>, %arg3: !torch.bool) -> !torch.list<int> {\n"
 "    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg2) : (!torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
@@ -10434,6 +10438,28 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    }\n"
 "    return %0#1 : !torch.int\n"
 "  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.selu\"(%arg0: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %none = torch.constant.none\n"
+"    %str = torch.constant.str \"AssertionError: \"\n"
+"    %int11 = torch.constant.int 11\n"
+"    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %1 = torch.aten.ne.int %0#1, %int11 : !torch.int, !torch.int -> !torch.bool\n"
+"    torch.prim.If %1 -> () {\n"
+"      torch.prim.If.yield\n"
+"    } else {\n"
+"      torch.prim.RaiseException %str, %none : !torch.str, !torch.none\n"
+"      torch.prim.If.yield\n"
+"    }\n"
+"    %2 = call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_integer_dtype(%0#1) : (!torch.int) -> !torch.bool\n"
+"    %3 = torch.aten.__not__ %2 : !torch.bool -> !torch.bool\n"
+"    torch.prim.If %3 -> () {\n"
+"      torch.prim.If.yield\n"
+"    } else {\n"
+"      torch.prim.RaiseException %str, %none : !torch.str, !torch.none\n"
+"      torch.prim.If.yield\n"
+"    }\n"
+"    return %0#1 : !torch.int\n"
+"  }\n"
 "  func.func @\"__torch_mlir_dtype_fn.aten.remainder.Scalar\"(%arg0: !torch.tuple<int, int>, %arg1: !torch.number) -> !torch.int {\n"
 "    %none = torch.constant.none\n"
 "    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
diff --git a/lib/Dialect/Torch/Transforms/DecomposeComplexOps.cpp b/lib/Dialect/Torch/Transforms/DecomposeComplexOps.cpp
@@ -1937,6 +1937,55 @@ class DecomposeAtenEluOp : public OpRewritePattern<AtenEluOp> {
 };
 } // namespace
 
+// Selu = scale * (max(0,x) + min(0,alpha * (exp(x) − 1)))
+namespace {
+class DecomposeAtenSeluOp : public OpRewritePattern<AtenSeluOp> {
+public:
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenSeluOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value input = op.getSelf();
+    auto resType = op.getType().cast<BaseTensorType>();
+    if (!resType.hasDtype()) {
+      return rewriter.notifyMatchFailure(op, "result should have dtype");
+    }
+
+    // Define λ and α
+    double scale = 1.0507009873554804934193349852946;
+    double alpha = 1.6732632423543772848170429916717;
+    
+    // Create constants for λ and α
+    Value scaleVal = rewriter.create<Torch::ConstantFloatOp>(loc, rewriter.getF64FloatAttr(scale));
+    Value alphaVal = rewriter.create<Torch::ConstantFloatOp>(loc, rewriter.getF64FloatAttr(alpha));
+
+    // Create zero tensor for comparison
+    Value constantZero =
+        rewriter.create<ConstantIntOp>(loc, rewriter.getI64IntegerAttr(0));
+    Value zeroTensor = createRank0Tensor(rewriter, loc, resType, constantZero);
+
+    // Calculate positive and negative parts
+    Value constantOne =
+        rewriter.create<ConstantFloatOp>(loc, rewriter.getF64FloatAttr(1.0));
+    Value positiveOutput = rewriter.create<AtenMaximumOp>(loc, resType, zeroTensor, input);
+    Value minZeroX =
+        rewriter.create<AtenMinimumOp>(loc, resType, zeroTensor, input);
+    Value expInput = rewriter.create<AtenExpOp>(loc, resType, minZeroX);
+    Value expInputMinusOne = rewriter.create<AtenSubScalarOp>(loc, resType, expInput, constantOne, constantOne);
+    Value negativeOutput = rewriter.create<AtenMulScalarOp>(loc, resType, expInputMinusOne,  alphaVal);
+
+    // Multiply the result by λ
+    Value seluOutput = rewriter.create<AtenAddTensorOp>(
+        loc, resType, positiveOutput, negativeOutput, constantOne);
+    seluOutput = rewriter.create<AtenMulScalarOp>(loc, resType, seluOutput, scaleVal);
+
+    // Replace the original operation
+    rewriter.replaceOp(op, seluOutput);
+    return success();
+  }
+};
+} // namespace
+
 namespace {
 class DecomposeAtenTOp : public OpRewritePattern<AtenTOp> {
 public:
@@ -6460,6 +6509,7 @@ class DecomposeComplexOpsPass
     addPatternIfTargetOpIsIllegal<DecomposeAtenRandnLikeOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenVarMeanOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenEluOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenSeluOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenLeakyReluOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenLeakyReluBackwardOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenNewEmptyStridedOp>(patterns);
diff --git a/lib/Dialect/Torch/Transforms/LowerToBackendContract.cpp b/lib/Dialect/Torch/Transforms/LowerToBackendContract.cpp
@@ -437,6 +437,7 @@ static void markDecomposedOpsAsIllegal(MLIRContext *context,
   target.addIllegalOp<AtenRelu6Op>();
   target.addIllegalOp<AtenEluOp>();
   target.addIllegalOp<AtenGluOp>();
+  target.addIllegalOp<AtenSeluOp>();
   target.addIllegalOp<AtenHardswishOp>();
   target.addIllegalOp<AtenSoftplusOp>();
   target.addIllegalOp<AtenSiluOp>();
diff --git a/projects/pt1/e2e_testing/xfail_sets.py b/projects/pt1/e2e_testing/xfail_sets.py
@@ -486,6 +486,7 @@
     "ElementwiseLeakyReluModule_basic",
     "ElementwiseEluModule_basic",
     "ElementwiseEluNonDefaultModule_basic",
+    "ElementwiseSeluModule_basic",
     "ElementwiseLogModule_basic",
     "ElementwiseNegModule_basic",
     "ElementwiseRsqrtModule_basic",
@@ -1115,6 +1116,7 @@
     "ElementwiseRemainderScalarModule_Int_basic",
     "ElementwiseRemainderScalarModule_Int_basic",
     "ElementwiseRsqrtModule_basic",
+    "ElementwiseSeluModule_basic",
     "ElementwiseSigmoidModule_basic",
     "ElementwiseSignModule_basic",
     "ElementwiseSqrtIntModule_basic",
diff --git a/projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/abstract_interp_lib_gen.py b/projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/abstract_interp_lib_gen.py
@@ -373,6 +373,9 @@ def aten〇elu〡shape(self: List[int], alpha: float = 1, scale: float = 1, inpu
 def aten〇prelu〡shape(self: List[int], weight: List[int]) -> List[int]:
     return upstream_shape_functions.unary(self)
 
+def aten〇selu〡shape(self: List[int]) -> List[int]:
+    return upstream_shape_functions.unary(self)
+
 def aten〇gather〡shape(self: List[int], dim: int, index: List[int], sparse_grad: bool = False) -> List[int]:
     return upstream_shape_functions.unary(index)
 
@@ -3066,6 +3069,14 @@ def aten〇elu〡dtype(self_rank_dtype: Tuple[int, int], alpha: Union[int, float
         assert not is_integer_dtype(self_dtype)
     return self_dtype
 
+@check_dtype_function(
+    _check_tensors_with_the_same_dtype(num_of_tensors=1, error_types={torch.bool, torch.int8, torch.uint8, torch.int16, torch.int32, torch.int64}))
+def aten〇selu〡dtype(self_rank_dtype: Tuple[int, int]) -> int:
+    self_rank, self_dtype = self_rank_dtype
+    assert self_dtype != torch.bool
+    assert not is_integer_dtype(self_dtype)
+    return self_dtype
+
 @check_dtype_function(
     _check_tensors_with_the_same_dtype(num_of_tensors=1, other=1) +
     _check_tensors_with_the_same_dtype(num_of_tensors=1, other=1.0))
diff --git a/projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/torch_ods_gen.py b/projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/torch_ods_gen.py
@@ -262,6 +262,7 @@ def emit_with_mutating_variants(key, **kwargs):
             "aten::relu6 : (Tensor) -> (Tensor)",
             "aten::leaky_relu : (Tensor, Scalar) -> (Tensor)",
             "aten::log : (Tensor) -> (Tensor)",
+            "aten::selu : (Tensor) -> (Tensor)",
             "aten::sigmoid : (Tensor) -> (Tensor)",
             "aten::sign : (Tensor) -> (Tensor)",
             "aten::sgn : (Tensor) -> (Tensor)",
diff --git a/projects/pt1/python/torch_mlir_e2e_test/test_suite/elementwise.py b/projects/pt1/python/torch_mlir_e2e_test/test_suite/elementwise.py
@@ -564,6 +564,27 @@ def ElementwiseGeluModule_basic(module, tu: TestUtils):
 # ==============================================================================
 
 
+class ElementwiseSeluModule(torch.nn.Module):
+
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args([
+        None,
+        ([-1, -1], torch.float32, True),
+    ])
+    def forward(self, x):
+        return torch.ops.aten.selu(x)
+
+@register_test_case(module_factory=lambda: ElementwiseSeluModule())
+def ElementwiseSeluModule_basic(module, tu: TestUtils):
+    module.forward(tu.rand(5, 3, low=-1, high=1))
+
+
+# ==============================================================================
+
+
 class ElementwiseSigmoidModule(torch.nn.Module):
 
     def __init__(self):
