[quantizer] add QHardsigmoid

docs/quantization_support.md

@@ -70,6 +70,7 @@ Quantized OPs that are natively not supported by PyTorch (and possibly TFLite).
 | `softmax` | For QATQuantizer/PostQuantizer, set `config={"set_quantizable_op_stats": True}`<br>For TFLiteConverter, set `rewrite_quantizable=True` |
 | `sum` | For TFLiteConverter, set `rewrite_quantizable=True` |
 | `torch.nn.GLU` | No action needed |
+| `torch.nn.Hardsigmoid` | No action needed |
 | `torch.nn.LogSoftmax` | For QATQuantizer/PostQuantizer, set `config={"set_quantizable_op_stats": True}`<br>For TFLiteConverter, set `rewrite_quantizable=True` |
 | `torch.nn.PReLU` | No action needed |
 | `torch.nn.SiLU` | No action needed |

tinynn/graph/quantization/modules.py

@@ -1,6 +1,7 @@
 import torch
 import torch.nn as nn
 import torch.nn.quantized as nnq
+import torch.quantization as torch_q
 
 
 class QPReLU(nn.Module):
@@ -70,3 +71,21 @@ def __init__(self, glu: nn.GLU) -> None:
     def forward(self, input: torch.Tensor) -> torch.Tensor:
         slices = torch.chunk(input, 2, self.dim)
         return self.f_mul.mul(slices[0], self.sigmoid(slices[1]))
+
+
+class QHardsigmoid(nn.Module):
+    def __init__(self, hardsigmoid: nn.Hardsigmoid) -> None:
+        super().__init__()
+
+        self.f_mul = nnq.FloatFunctional()
+        self.f_add = nnq.FloatFunctional()
+        self.q = torch_q.QuantStub()
+        self.dq = torch_q.DeQuantStub()
+        self.act_hs = nn.Hardsigmoid()
+        self.act_r = nn.ReLU6()
+
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        x1 = self.f_add.add_scalar(input, 3.0)
+        x2 = self.act_r(x1)
+        x3 = self.q(self.dq(x2))
+        return self.f_mul.mul_scalar(x3, 1 / 6)

tinynn/graph/quantization/quantizer.py

@@ -23,7 +23,7 @@
     FakeQuantizeBFloat16,
     FakeQuantizeTFLite,
 )
-from tinynn.graph.quantization.modules import QGLU, QPReLU, QSiLU
+from tinynn.graph.quantization.modules import QGLU, QHardsigmoid, QPReLU, QSiLU
 from tinynn.graph.quantization.observer import (
     HistogramObserverKL,
     MinMaxObserver,
@@ -223,6 +223,7 @@
 Q_MODULES_MAPPING = {
     nn.PReLU: QPReLU,
     nn.GLU: QGLU,
+    nn.Hardsigmoid: QHardsigmoid,
 }
 
 FUNCTIONAL_MODULE_MAPPING = {