import flexflow.torch.fx as fx
import torch.nn as nn
import torch
class Transpose(nn.Module):
def __init__(self):
super(Transpose, self).__init__()
def forward(self, x):
x = torch.transpose(x,-1,-2)
return x
model = nn.Sequential(Transpose(),nn.Flatten(229*229*3,10))
fx.torch_to_flexflow(model, "scalar_multiply.ff")
self.flat1 = nn.Flatten()
self.linear1 = nn.Linear(512, 512)
self.linear2 = nn.Linear(512, 10)
self.relu = nn.ReLU()
def forward(self, input1, input2):
y1 = self.conv1(input1)
y1 = self.relu(y1)
y2 = self.conv1(input2)
y2 = self.relu(y2)
y = torch.cat((y1, y2), 1)
(y1, y2) = torch.split(y, 1)
y = torch.cat((y1, y2), 1)
y = self.conv2(y)
y = self.relu(y)
y = self.pool1(y)
y = self.conv3(y)
y = self.relu(y)
y = self.conv4(y)
y = self.relu(y)
y = self.pool2(y)
y = self.flat1(y)
y = self.linear1(y)
y = self.relu(y)
yo = self.linear2(y)
return (yo, y)
model = CNN()
fx.torch_to_flexflow(model, "cnn.ff")
import flexflow.torch.fx as fx
import torch.nn as nn
class Addition(nn.Module):
def __init__(self, scalar):
super(Addition, self).__init__()
self.scalar = scalar
def forward(self, x):
x = x + self.scalar
return x
model = nn.Sequential(Addition(2.0), nn.Flatten(229 * 229 * 3, 10))
fx.torch_to_flexflow(model, "scalar_addition.ff")
progress (bool): If True, displays a progress bar of the download to stderr
"""
kwargs['width_per_group'] = 64 * 2
return _resnet('wide_resnet50_2', Bottleneck, [3, 4, 6, 3], pretrained,
progress, **kwargs)
def wide_resnet101_2(pretrained: bool = False,
progress: bool = True,
**kwargs: Any) -> ResNet:
r"""Wide ResNet-101-2 model from
`"Wide Residual Networks" <https://arxiv.org/pdf/1605.07146.pdf>`_.
The model is the same as ResNet except for the bottleneck number of channels
which is twice larger in every block. The number of channels in outer 1x1
convolutions is the same, e.g. last block in ResNet-50 has 2048-512-2048
channels, and in Wide ResNet-50-2 has 2048-1024-2048.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
progress (bool): If True, displays a progress bar of the download to stderr
"""
kwargs['width_per_group'] = 64 * 2
return _resnet('wide_resnet101_2', Bottleneck, [3, 4, 23, 3], pretrained,
progress, **kwargs)
input = torch.randn(64, 3, 224, 224)
model = resnet152()
fx.torch_to_flexflow(model, "resnet152.ff")
import flexflow.torch.fx as fx import torch.nn as nn model = nn.Sequential(nn.Identity(), nn.Flatten(229 * 229 * 3, 10)) fx.torch_to_flexflow(model, "identity.ff")
import flexflow.torch.fx as fx
import torch.nn as nn
import torch
class CustomParam(nn.Module):
def __init__(self):
super(CustomParam, self).__init__()
self.mat = nn.Parameter(torch.zeros(1, 1))
def forward(self, x):
x = torch.matmul(self.mat, x)
return x
model = nn.Sequential(CustomParam())
fx.torch_to_flexflow(model, "customParam.ff")
import flexflow.torch.fx as fx
import torch.nn as nn
import torch
class Division(nn.Module):
def __init__(self, scalar):
super(Division, self).__init__()
self.scalar = scalar
def forward(self, x):
x = x // self.scalar
return x
model = nn.Sequential(Division(2.0), nn.Flatten(229 * 229 * 3, 10))
fx.torch_to_flexflow(model, "scalar.ff")
import flexflow.torch.fx as fx import torch.nn as nn model = nn.Sequential(nn.GELU(),nn.Flatten(229*229*3,10)) fx.torch_to_flexflow(model, "gelu.ff")
import torch.nn as nn
import flexflow.torch.fx as fx
class MLP(nn.Module):
def __init__(self):
super().__init__()
self.linear1 = nn.Linear(784, 512)
self.linear2 = nn.Linear(512, 512)
self.linear3 = nn.Linear(512, 10)
self.relu = nn.ReLU()
def forward(self, x):
y = self.linear1(x)
y = self.relu(y)
y = self.linear2(y)
y = self.relu(y)
y = self.linear3(y)
return y
model = MLP()
fx.torch_to_flexflow(model, "mlp.ff")
import torch.nn as nn import torchvision.models as models import flexflow.torch.fx as fx # alexnet = models.alexnet() # fx.torch_to_flexflow(alexnet, "alexnet.ff") # # vgg16 = models.vgg16() # fx.torch_to_flexflow(vgg16, "vgg16.ff") # # squeezenet = models.squeezenet1_0() # fx.torch_to_flexflow(squeezenet, "squeezenet.ff") # densenet = models.densenet161() # fx.torch_to_flexflow(densenet, "densenet.ff") # inception = models.inception_v3() # fx.torch_to_flexflow(inception, "inception.ff") googlenet = models.googlenet() fx.torch_to_flexflow(googlenet, "googlenet.ff") # shufflenet = models.shufflenet_v2_x1_0() # fx.torch_to_flexflow(shufflenet, "shufflenet.ff") # mobilenet = models.mobilenet_v2() # fx.torch_to_flexflow(mobilenet, "mobilenet.ff")
nn.MaxPool2d(kernel_size=3, stride=2),
nn.Conv2d(64, 192, kernel_size=5, padding=2),
nn.ReLU(inplace=True),
nn.MaxPool2d(kernel_size=3, stride=2),
nn.Conv2d(192, 384, kernel_size=3, padding=1),
nn.ReLU(inplace=True),
nn.Conv2d(384, 256, kernel_size=3, padding=1),
nn.ReLU(inplace=True),
nn.Conv2d(256, 256, kernel_size=3, padding=1),
nn.ReLU(inplace=True),
nn.MaxPool2d(kernel_size=3, stride=2),
)
self.classifier = nn.Sequential(
nn.Linear(256 * 6 * 6, 4096),
nn.ReLU(inplace=True),
nn.Linear(4096, 4096),
nn.ReLU(inplace=True),
nn.Linear(4096, num_classes),
nn.Softmax(),
)
def forward(self, x: torch.Tensor) -> torch.Tensor:
x = self.features(x)
x = torch.flatten(x, 1)
x = self.classifier(x)
return x
model = AlexNet(num_classes=10)
fx.torch_to_flexflow(model, "alexnet.ff")
import timm.models.vision_transformer as vit import flexflow.torch.fx as fx import torch.nn as nn model = vit.vit_base_patch32_224_in21k() model = nn.Sequential(model,nn.Flatten(),nn.Linear(21843,1000)) fx.torch_to_flexflow(model, "vit_base_patch32_224_in21k.ff")
import classy_vision.models.regnet as rgn import flexflow.torch.fx as fx import torch.nn as nn model = rgn.RegNetX32gf() model = nn.Sequential(model,nn.Flatten(),nn.Linear(2520*7*7,1000)) fx.torch_to_flexflow(model, "regnetX32gf.ff")