[Relax][PyTorch] Add RSub Op Support for Exported Program and FX graph

python/tvm/relax/frontend/torch/base_fx_graph_translator.py

@@ -401,6 +401,16 @@ def call_binary_op(op, lhs, rhs):
 
         return convert
 
+    def _rsub(self, node: fx.Node) -> relax.Var:
+        args = self.retrieve_args(node)
+        lhs = args[0]
+        rhs = args[1]
+
+        if isinstance(rhs, (int, float)):
+            rhs = relax.const(rhs)
+
+        return self.block_builder.emit(relax.op.subtract(rhs, lhs))
+
     ########## Linear Algebra ##########
 
     def _linalg_vector_norm(self, node: fx.Node) -> relax.Var:

python/tvm/relax/frontend/torch/exported_program_translator.py

@@ -304,6 +304,8 @@ def create_convert_map(
             "relu_.default": self._unary_op(relax.op.nn.relu),
             "round.default": self._round,
             "rsqrt.default": self._unary_op(relax.op.rsqrt),
+            "rsub.Tensor": self._rsub,
+            "rsub.Scalar": self._rsub,
             "selu.default": self._unary_op(relax.op.nn.selu),
             "sigmoid.default": self._unary_op(relax.op.sigmoid),
             "sign.default": self._unary_op(relax.op.sign),

python/tvm/relax/frontend/torch/fx_translator.py

@@ -699,6 +699,7 @@ def create_convert_map(
             "pow": self._binary_op(relax.op.power, operator.pow),
             "or_": self._binary_op(relax.op.bitwise_or, operator.or_),
             "rshift": self._binary_op(relax.op.right_shift, operator.rshift),
+            "rsub": self._rsub,
             "sub": self._binary_op(relax.op.subtract, operator.sub),
             "truediv": self._binary_op(relax.op.divide, operator.truediv),
             "xor": self._binary_op(relax.op.bitwise_xor, operator.xor),

tests/python/relax/test_frontend_from_exported_program.py

@@ -899,6 +899,7 @@ def test_binary3():
         torch.randn(10, 10, dtype=torch.float32),
         torch.randn(10, 10, dtype=torch.float32),
     )
+    example_args2 = (torch.randn(10, 10, dtype=torch.float32),)
 
     # Max
     class Max1(Module):
@@ -940,6 +941,42 @@ def main(
 
     verify_model(Min1(), example_args1, {}, expected_min1)
 
+    # RSub
+    class RSub1(Module):
+        def forward(self, x, y):
+            return torch.rsub(x, y)
+
+    class RSub2(Module):
+        def forward(self, x):
+            return torch.rsub(x, 5.0)
+
+    @tvm.script.ir_module
+    class expected_rsub1:
+        @R.function
+        def main(
+            x: R.Tensor((10, 10), dtype="float32"), y: R.Tensor((10, 10), dtype="float32")
+        ) -> R.Tuple(R.Tensor((10, 10), dtype="float32")):
+            with R.dataflow():
+                lv: R.Tensor((10, 10), dtype="float32") = R.subtract(y, x)
+                gv: R.Tuple(R.Tensor((10, 10), dtype="float32")) = (lv,)
+                R.output(gv)
+            return gv
+
+    @tvm.script.ir_module
+    class expected_rsub2:
+        @R.function
+        def main(
+            x: R.Tensor((10, 10), dtype="float32")
+        ) -> R.Tuple(R.Tensor((10, 10), dtype="float32")):
+            with R.dataflow():
+                lv: R.Tensor((10, 10), dtype="float32") = R.subtract(R.const(5.0, "float32"), x)
+                gv: R.Tuple(R.Tensor((10, 10), dtype="float32")) = (lv,)
+                R.output(gv)
+            return gv
+
+    verify_model(RSub1(), example_args1, {}, expected_rsub1)
+    verify_model(RSub2(), example_args2, {}, expected_rsub2)
+
 
 def test_batchnorm2d():
     class BatchNorm2d(Module):

tests/python/relax/test_frontend_from_fx.py

@@ -1702,6 +1702,45 @@ def main(
     verify_model(Binary2(op), input_info2, {}, expected_binary2)
 
 
+# RSub
+def test_rsub():
+    input_info1 = [([10, 10], "float32"), ([10, 10], "float32")]
+    input_info2 = [([10, 10], "float32")]
+
+    class RSub1(Module):
+        def forward(self, x, y):
+            return torch.rsub(x, y)
+
+    class RSub2(Module):
+        def forward(self, x):
+            return torch.rsub(x, 5.0)
+
+    @tvm.script.ir_module
+    class expected_rsub1:
+        @R.function
+        def main(
+            x: R.Tensor((10, 10), dtype="float32"), y: R.Tensor((10, 10), dtype="float32")
+        ) -> R.Tensor((10, 10), dtype="float32"):
+            with R.dataflow():
+                lv: R.Tensor((10, 10), dtype="float32") = R.subtract(y, x)
+                gv: R.Tensor((10, 10), dtype="float32") = lv
+                R.output(gv)
+            return gv
+
+    @tvm.script.ir_module
+    class expected_rsub2:
+        @R.function
+        def main(x: R.Tensor((10, 10), dtype="float32")) -> R.Tensor((10, 10), dtype="float32"):
+            with R.dataflow():
+                lv: R.Tensor((10, 10), dtype="float32") = R.subtract(R.const(5.0, "float32"), x)
+                gv: R.Tensor((10, 10), dtype="float32") = lv
+                R.output(gv)
+            return gv
+
+    verify_model(RSub1(), input_info1, {}, expected_rsub1)
+    verify_model(RSub2(), input_info2, {}, expected_rsub2)
+
+
 def test_size():
     input_info = [([1, 3, 10, 10], "float32")]