resnet18 pytorch with bounded activation OK

src/acetone_nnet/importers/PYTORCH_importer/parse_pytorch.py

@@ -5,6 +5,7 @@
 from acetone_nnet.generator import (
     ActivationFunctions,
     Add,
+    Clip,
     Average,
     AveragePooling2D,
     BatchNormalization,
@@ -146,11 +147,14 @@ def visit_call_function(self, node: torch.fx.Node):
             self.layerdic[node.name] = (this_layer, node.meta['val'].shape)
             self.idx+=1
         elif "relu" in target_name:
-            self.layerdic[node.args[0].name][0].activation_function=ReLu()
-            self.layerdic[node.name] = self.layerdic[node.args[0].name]
+            self.layerdic[node.args[0].name][0].activation_function=ReLu() # set previous layer activation 
+            self.layerdic[node.name] = self.layerdic[node.args[0].name] # register 
         elif "clamp" in target_name:
-            self.layerdic[node.args[0].name][0].activation_function=ReLu()
-            self.layerdic[node.name] = self.layerdic[node.args[0].name]
+            self.layerdic[node.args[0].name][0].activation_function=Clip( # set previous layer activation 
+                max_value=node.args[2],
+                min_value=node.args[1]
+            )
+            self.layerdic[node.name] = self.layerdic[node.args[0].name] # register 
         elif "conv2d" in target_name:
             stride = node.args[3][0] if len(node.args) > 3 else 1
             padding = node.args[4] + node.args[4] if len(node.args) > 4 else [0,0,0,0]
@@ -169,13 +173,45 @@ def visit_call_function(self, node: torch.fx.Node):
                     input_shape=self.layerdic[node.args[0].name][1],
                     output_shape=node.meta['val'].shape,
                     weights=self.module.state_dict()[node.args[1].target].numpy(),
-                    biases=self.module.state_dict()[node.args[2].target].numpy(),
+                    biases=self.module.state_dict()[node.args[2].target].numpy() if node.args[2] is not None else np.zeros(node.meta['val'].shape[1],dtype=np.float32),
                     activation_function=Linear(),
                 )
             self.layerdic[node.args[0].name][0].next_layer.append(this_layer)
             this_layer.previous_layer.append(self.layerdic[node.args[0].name][0])
             self.layerdic[node.name] = (this_layer, node.meta['val'].shape)
             self.idx+=1
+        elif "batch_norm" in target_name:
+            # Fusing batch normalization to previous conv2d https://pytorch-cn.com/tutorials/intermediate/fx_conv_bn_fuser.html
+            self.layerdic[node.name] = self.layerdic[node.args[0].name]
+            bn_w = self.module.state_dict()[node.args[1].target]
+            bn_b = self.module.state_dict()[node.args[2].target]
+            bn_rm = self.module.state_dict()[node.args[3].target]
+            bn_rv = self.module.state_dict()[node.args[4].target]
+            bn_eps = node.args[7]
+            bn_var_rsqrt = torch.rsqrt(bn_rv + bn_eps)
+            conv_w = torch.from_numpy(self.layerdic[node.args[0].name][0].weights)
+            conv_b = torch.from_numpy(self.layerdic[node.args[0].name][0].biases)
+            conv_w = conv_w * (bn_w * bn_var_rsqrt).reshape([-1] + [1] * (len(conv_w.shape) - 1))
+            conv_b = (conv_b - bn_rm) * bn_var_rsqrt * bn_w + bn_b
+            self.layerdic[node.args[0].name][0].weights = conv_w.numpy().astype(np.float32)
+            self.layerdic[node.args[0].name][0].biases = conv_b.numpy().astype(np.float32)
+        elif "add" in target_name:
+            if node.args[0].name in self.layerdic.keys(): # bypass bn2_num_batches_tracked + 1
+                this_layer =  Add(
+                    original_name=node.name,
+                    idx=self.idx,
+                    size=node.meta['val'].numel(),
+                    input_shapes=[self.layerdic[node.args[i].name][1] for i in range(len(node.args))],
+                    output_shape=node.meta['val'].shape,
+                    activation_function=Linear(),
+                )
+                for i in range(len(node.args)):
+                    self.layerdic[node.args[i].name][0].next_layer.append(this_layer)
+                    this_layer.previous_layer.append(self.layerdic[node.args[i].name][0])
+                self.layerdic[node.name] = (this_layer, node.meta['val'].shape)
+                self.idx+=1
+            else:
+                logging.info(f"Add layer bypassed: {node.args[0].name}")
         elif "cat" in target_name:
             this_layer = Concatenate(
                     original_name=node.name,
@@ -193,6 +229,8 @@ def visit_call_function(self, node: torch.fx.Node):
             self.idx+=1
         elif "dropout" in target_name:
             self.layerdic[node.name] = self.layerdic[node.args[0].name]
+        elif "flatten" in target_name:
+            self.layerdic[node.name] = self.layerdic[node.args[0].name]
         elif "max_pool2d" in target_name:
             kernel = node.args[1][0] if len(node.args) > 1 else 1
             stride = node.args[2][0] if len(node.args) > 2 else 1

tests/models/resnet/resnet.py

@@ -0,0 +1,286 @@
+from functools import partial
+from typing import Any, Callable, Optional, Union
+
+import torch
+import torch.nn as nn
+from torch import Tensor
+
+class ReluN(nn.Module):
+    def __init__(self, n=1000) -> None:
+        super().__init__()
+        self.n = n
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return torch.clamp(x,0,self.n)
+
+def conv3x3(in_planes: int, out_planes: int, stride: int = 1, groups: int = 1, dilation: int = 1) -> nn.Conv2d:
+    """3x3 convolution with padding"""
+    return nn.Conv2d(
+        in_planes,
+        out_planes,
+        kernel_size=3,
+        stride=stride,
+        padding=dilation,
+        groups=groups,
+        bias=False,
+        dilation=dilation,
+    )
+
+
+def conv1x1(in_planes: int, out_planes: int, stride: int = 1) -> nn.Conv2d:
+    """1x1 convolution"""
+    return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)
+
+
+class BasicBlock(nn.Module):
+    expansion: int = 1
+
+    def __init__(
+        self,
+        inplanes: int,
+        planes: int,
+        stride: int = 1,
+        downsample: Optional[nn.Module] = None,
+        groups: int = 1,
+        base_width: int = 64,
+        dilation: int = 1,
+        norm_layer: Optional[Callable[..., nn.Module]] = None,
+    ) -> None:
+        super().__init__()
+        if norm_layer is None:
+            norm_layer = nn.BatchNorm2d
+        if groups != 1 or base_width != 64:
+            raise ValueError("BasicBlock only supports groups=1 and base_width=64")
+        if dilation > 1:
+            raise NotImplementedError("Dilation > 1 not supported in BasicBlock")
+        # Both self.conv1 and self.downsample layers downsample the input when stride != 1
+        self.conv1 = conv3x3(inplanes, planes, stride)
+        self.bn1 = norm_layer(planes)
+        self.relu1 = ReluN()
+        self.relu2 = ReluN()
+        self.conv2 = conv3x3(planes, planes)
+        self.bn2 = norm_layer(planes)
+        self.downsample = downsample
+        self.stride = stride
+
+    def forward(self, x: Tensor) -> Tensor:
+        identity = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu1(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+
+        if self.downsample is not None:
+            identity = self.downsample(x)
+
+        out += identity
+        out = self.relu2(out)
+
+        return out
+
+
+class Bottleneck(nn.Module):
+    # Bottleneck in torchvision places the stride for downsampling at 3x3 convolution(self.conv2)
+    # while original implementation places the stride at the first 1x1 convolution(self.conv1)
+    # according to "Deep residual learning for image recognition" https://arxiv.org/abs/1512.03385.
+    # This variant is also known as ResNet V1.5 and improves accuracy according to
+    # https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_pytorch.
+
+    expansion: int = 4
+
+    def __init__(
+        self,
+        inplanes: int,
+        planes: int,
+        stride: int = 1,
+        downsample: Optional[nn.Module] = None,
+        groups: int = 1,
+        base_width: int = 64,
+        dilation: int = 1,
+        norm_layer: Optional[Callable[..., nn.Module]] = None,
+    ) -> None:
+        super().__init__()
+        if norm_layer is None:
+            norm_layer = nn.BatchNorm2d
+        width = int(planes * (base_width / 64.0)) * groups
+        # Both self.conv2 and self.downsample layers downsample the input when stride != 1
+        self.conv1 = conv1x1(inplanes, width)
+        self.bn1 = norm_layer(width)
+        self.conv2 = conv3x3(width, width, stride, groups, dilation)
+        self.bn2 = norm_layer(width)
+        self.conv3 = conv1x1(width, planes * self.expansion)
+        self.bn3 = norm_layer(planes * self.expansion)
+        self.relu = nn.ReLU(inplace=True)
+        self.downsample = downsample
+        self.stride = stride
+
+    def forward(self, x: Tensor) -> Tensor:
+        identity = x
+
+        out = self.conv1(x)
+        out = self.bn1(out)
+        out = self.relu(out)
+
+        out = self.conv2(out)
+        out = self.bn2(out)
+        out = self.relu(out)
+
+        out = self.conv3(out)
+        out = self.bn3(out)
+
+        if self.downsample is not None:
+            identity = self.downsample(x)
+
+        out += identity
+        out = self.relu(out)
+
+        return out
+
+
+class ResNet(nn.Module):
+    def __init__(
+        self,
+        block: type[Union[BasicBlock, Bottleneck]],
+        layers: list[int],
+        num_classes: int = 1000,
+        zero_init_residual: bool = False,
+        groups: int = 1,
+        width_per_group: int = 64,
+        replace_stride_with_dilation: Optional[list[bool]] = None,
+        norm_layer: Optional[Callable[..., nn.Module]] = None,
+    ) -> None:
+        super().__init__()
+#        _log_api_usage_once(self)
+        if norm_layer is None:
+            norm_layer = nn.BatchNorm2d
+        self._norm_layer = norm_layer
+
+        self.inplanes = 64
+        self.dilation = 1
+        if replace_stride_with_dilation is None:
+            # each element in the tuple indicates if we should replace
+            # the 2x2 stride with a dilated convolution instead
+            replace_stride_with_dilation = [False, False, False]
+        if len(replace_stride_with_dilation) != 3:
+            raise ValueError(
+                "replace_stride_with_dilation should be None "
+                f"or a 3-element tuple, got {replace_stride_with_dilation}"
+            )
+        self.groups = groups
+        self.base_width = width_per_group
+        self.conv1 = nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False)
+        self.bn1 = norm_layer(self.inplanes)
+        self.relu = ReluN()
+        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
+        self.layer1 = self._make_layer(block, 64, layers[0])
+        self.layer2 = self._make_layer(block, 128, layers[1], stride=2, dilate=replace_stride_with_dilation[0])
+        self.layer3 = self._make_layer(block, 256, layers[2], stride=2, dilate=replace_stride_with_dilation[1])
+        self.layer4 = self._make_layer(block, 512, layers[3], stride=2, dilate=replace_stride_with_dilation[2])
+        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
+        self.fc = nn.Linear(512 * block.expansion, num_classes)
+
+        for m in self.modules():
+            if isinstance(m, nn.Conv2d):
+                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
+            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
+                nn.init.constant_(m.weight, 1)
+                nn.init.constant_(m.bias, 0)
+
+        # Zero-initialize the last BN in each residual branch,
+        # so that the residual branch starts with zeros, and each residual block behaves like an identity.
+        # This improves the model by 0.2~0.3% according to https://arxiv.org/abs/1706.02677
+        if zero_init_residual:
+            for m in self.modules():
+                if isinstance(m, Bottleneck) and m.bn3.weight is not None:
+                    nn.init.constant_(m.bn3.weight, 0)  # type: ignore[arg-type]
+                elif isinstance(m, BasicBlock) and m.bn2.weight is not None:
+                    nn.init.constant_(m.bn2.weight, 0)  # type: ignore[arg-type]
+
+    def _make_layer(
+        self,
+        block: type[Union[BasicBlock, Bottleneck]],
+        planes: int,
+        blocks: int,
+        stride: int = 1,
+        dilate: bool = False,
+    ) -> nn.Sequential:
+        norm_layer = self._norm_layer
+        downsample = None
+        previous_dilation = self.dilation
+        if dilate:
+            self.dilation *= stride
+            stride = 1
+        if stride != 1 or self.inplanes != planes * block.expansion:
+            downsample = nn.Sequential(
+                conv1x1(self.inplanes, planes * block.expansion, stride),
+                norm_layer(planes * block.expansion),
+            )
+
+        layers = []
+        layers.append(
+            block(
+                self.inplanes, planes, stride, downsample, self.groups, self.base_width, previous_dilation, norm_layer
+            )
+        )
+        self.inplanes = planes * block.expansion
+        for _ in range(1, blocks):
+            layers.append(
+                block(
+                    self.inplanes,
+                    planes,
+                    groups=self.groups,
+                    base_width=self.base_width,
+                    dilation=self.dilation,
+                    norm_layer=norm_layer,
+                )
+            )
+
+        return nn.Sequential(*layers)
+
+    def _forward_impl(self, x: Tensor) -> Tensor:
+        # See note [TorchScript super()]
+        x = self.conv1(x)
+        x = self.bn1(x)
+        x = self.relu(x)
+        x = self.maxpool(x)
+
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+
+        x = self.avgpool(x)
+        x = torch.flatten(x, 1)
+        x = self.fc(x)
+
+        return x
+
+    def forward(self, x: Tensor) -> Tensor:
+        return self._forward_impl(x)
+
+
+
+def resnet18(**kwargs:Any) -> ResNet:
+    """ResNet-18 from `Deep Residual Learning for Image Recognition <https://arxiv.org/abs/1512.03385>`__.
+
+    Args:
+        weights (:class:`~torchvision.models.ResNet18_Weights`, optional): The
+            pretrained weights to use. See
+            :class:`~torchvision.models.ResNet18_Weights` below for
+            more details, and possible values. By default, no pre-trained
+            weights are used.
+        progress (bool, optional): If True, displays a progress bar of the
+            download to stderr. Default is True.
+        **kwargs: parameters passed to the ``torchvision.models.resnet.ResNet``
+            base class. Please refer to the `source code
+            <https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py>`_
+            for more details about this class.
+
+    .. autoclass:: torchvision.models.ResNet18_Weights
+        :members:
+    """
+    return ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
+
+