[ONNX] Add silu operator support for onnx (#51193)

Support for yolov5 compound-scaled object detection models export.

[ghstack-poisoned]

test/onnx/test_pytorch_onnx_onnxruntime.py

@@ -4372,6 +4372,18 @@ def forward(self, x):
                       dynamic_axes={'x': [1, 2]},
                       test_with_inputs=[y])
 
+    def test_silu(self):
+        class SiLUModel(torch.nn.Module):
+            def __init__(self):
+                super(SiLUModel, self).__init__()
+                self.silu = torch.nn.SiLU()
+
+            def forward(self, x):
+                return self.silu(x)
+
+        x = torch.randn(2, 3, 4)
+        self.run_test(SiLUModel(), (x))
+
     def test_remainder(self):
         class RemainderModel(torch.nn.Module):
             def forward(self, input, other):

torch/onnx/symbolic_opset9.py

@@ -660,6 +660,10 @@ def prelu(g, self, weight):
     return g.op("PRelu", self, weight)
 
 
+def silu(g, input):
+    return g.op('Mul', input, g.op('Sigmoid', input))
+
+
 def relu(g, input):
     return g.op("Relu", input)