!pip list | grep timmtimm 1.0.3Timm - Loading Models
Timm - Loading Models
Using Timm
import timm
import torch
import torch.nn as nn
import torch.nn.functional as Ftimm??Type: module String form: <module 'timm' from '/home/ben/miniconda3/envs/pfast/lib/python3.12/site-packages/timm/__init__.py'> File: ~/miniconda3/envs/pfast/lib/python3.12/site-packages/timm/__init__.py Source: from .version import __version__ from .layers import is_scriptable, is_exportable, set_scriptable, set_exportable from .models import create_model, list_models, list_pretrained, is_model, list_modules, model_entrypoint, \ is_model_pretrained, get_pretrained_cfg, get_pretrained_cfg_value
List Models
print(timm.list_models("resnet*"))['resnet10t', 'resnet14t', 'resnet18', 'resnet18d', 'resnet26', 'resnet26d', 'resnet26t', 'resnet32ts', 'resnet33ts', 'resnet34', 'resnet34d', 'resnet50', 'resnet50_gn', 'resnet50c', 'resnet50d', 'resnet50s', 'resnet50t', 'resnet51q', 'resnet61q', 'resnet101', 'resnet101c', 'resnet101d', 'resnet101s', 'resnet152', 'resnet152c', 'resnet152d', 'resnet152s', 'resnet200', 'resnet200d', 'resnetaa34d', 'resnetaa50', 'resnetaa50d', 'resnetaa101d', 'resnetblur18', 'resnetblur50', 'resnetblur50d', 'resnetblur101d', 'resnetrs50', 'resnetrs101', 'resnetrs152', 'resnetrs200', 'resnetrs270', 'resnetrs350', 'resnetrs420', 'resnetv2_50', 'resnetv2_50d', 'resnetv2_50d_evos', 'resnetv2_50d_frn', 'resnetv2_50d_gn', 'resnetv2_50t', 'resnetv2_50x1_bit', 'resnetv2_50x3_bit', 'resnetv2_101', 'resnetv2_101d', 'resnetv2_101x1_bit', 'resnetv2_101x3_bit', 'resnetv2_152', 'resnetv2_152d', 'resnetv2_152x2_bit', 'resnetv2_152x4_bit']len(timm.list_models())1063Create Model
model = timm.create_model('resnet18', pretrained=True)List Pretrained
print(timm.list_pretrained("resnet*"))['resnet10t.c3_in1k', 'resnet14t.c3_in1k', 'resnet18.a1_in1k', 'resnet18.a2_in1k', 'resnet18.a3_in1k', 'resnet18.fb_ssl_yfcc100m_ft_in1k', 'resnet18.fb_swsl_ig1b_ft_in1k', 'resnet18.gluon_in1k', 'resnet18.tv_in1k', 'resnet18d.ra2_in1k', 'resnet26.bt_in1k', 'resnet26d.bt_in1k', 'resnet26t.ra2_in1k', 'resnet32ts.ra2_in1k', 'resnet33ts.ra2_in1k', 'resnet34.a1_in1k', 'resnet34.a2_in1k', 'resnet34.a3_in1k', 'resnet34.bt_in1k', 'resnet34.gluon_in1k', 'resnet34.tv_in1k', 'resnet34d.ra2_in1k', 'resnet50.a1_in1k', 'resnet50.a1h_in1k', 'resnet50.a2_in1k', 'resnet50.a3_in1k', 'resnet50.am_in1k', 'resnet50.b1k_in1k', 'resnet50.b2k_in1k', 'resnet50.bt_in1k', 'resnet50.c1_in1k', 'resnet50.c2_in1k', 'resnet50.d_in1k', 'resnet50.fb_ssl_yfcc100m_ft_in1k', 'resnet50.fb_swsl_ig1b_ft_in1k', 'resnet50.gluon_in1k', 'resnet50.ra_in1k', 'resnet50.ram_in1k', 'resnet50.tv2_in1k', 'resnet50.tv_in1k', 'resnet50_gn.a1h_in1k', 'resnet50c.gluon_in1k', 'resnet50d.a1_in1k', 'resnet50d.a2_in1k', 'resnet50d.a3_in1k', 'resnet50d.gluon_in1k', 'resnet50d.ra2_in1k', 'resnet50s.gluon_in1k', 'resnet51q.ra2_in1k', 'resnet61q.ra2_in1k', 'resnet101.a1_in1k', 'resnet101.a1h_in1k', 'resnet101.a2_in1k', 'resnet101.a3_in1k', 'resnet101.gluon_in1k', 'resnet101.tv2_in1k', 'resnet101.tv_in1k', 'resnet101c.gluon_in1k', 'resnet101d.gluon_in1k', 'resnet101d.ra2_in1k', 'resnet101s.gluon_in1k', 'resnet152.a1_in1k', 'resnet152.a1h_in1k', 'resnet152.a2_in1k', 'resnet152.a3_in1k', 'resnet152.gluon_in1k', 'resnet152.tv2_in1k', 'resnet152.tv_in1k', 'resnet152c.gluon_in1k', 'resnet152d.gluon_in1k', 'resnet152d.ra2_in1k', 'resnet152s.gluon_in1k', 'resnet200d.ra2_in1k', 'resnetaa50.a1h_in1k', 'resnetaa50d.d_in12k', 'resnetaa50d.sw_in12k', 'resnetaa50d.sw_in12k_ft_in1k', 'resnetaa101d.sw_in12k', 'resnetaa101d.sw_in12k_ft_in1k', 'resnetblur50.bt_in1k', 'resnetrs50.tf_in1k', 'resnetrs101.tf_in1k', 'resnetrs152.tf_in1k', 'resnetrs200.tf_in1k', 'resnetrs270.tf_in1k', 'resnetrs350.tf_in1k', 'resnetrs420.tf_in1k', 'resnetv2_50.a1h_in1k', 'resnetv2_50d_evos.ah_in1k', 'resnetv2_50d_gn.ah_in1k', 'resnetv2_50x1_bit.goog_distilled_in1k', 'resnetv2_50x1_bit.goog_in21k', 'resnetv2_50x1_bit.goog_in21k_ft_in1k', 'resnetv2_50x3_bit.goog_in21k', 'resnetv2_50x3_bit.goog_in21k_ft_in1k', 'resnetv2_101.a1h_in1k', 'resnetv2_101x1_bit.goog_in21k', 'resnetv2_101x1_bit.goog_in21k_ft_in1k', 'resnetv2_101x3_bit.goog_in21k', 'resnetv2_101x3_bit.goog_in21k_ft_in1k', 'resnetv2_152x2_bit.goog_in21k', 'resnetv2_152x2_bit.goog_in21k_ft_in1k', 'resnetv2_152x2_bit.goog_teacher_in21k_ft_in1k', 'resnetv2_152x2_bit.goog_teacher_in21k_ft_in1k_384', 'resnetv2_152x4_bit.goog_in21k', 'resnetv2_152x4_bit.goog_in21k_ft_in1k']Using torchvision models
import torchvision.modelsprint(torchvision.models.list_models())['alexnet', 'convnext_base', 'convnext_large', 'convnext_small', 'convnext_tiny', 'deeplabv3_mobilenet_v3_large', 'deeplabv3_resnet101', 'deeplabv3_resnet50', 'densenet121', 'densenet161', 'densenet169', 'densenet201', 'efficientnet_b0', 'efficientnet_b1', 'efficientnet_b2', 'efficientnet_b3', 'efficientnet_b4', 'efficientnet_b5', 'efficientnet_b6', 'efficientnet_b7', 'efficientnet_v2_l', 'efficientnet_v2_m', 'efficientnet_v2_s', 'fasterrcnn_mobilenet_v3_large_320_fpn', 'fasterrcnn_mobilenet_v3_large_fpn', 'fasterrcnn_resnet50_fpn', 'fasterrcnn_resnet50_fpn_v2', 'fcn_resnet101', 'fcn_resnet50', 'fcos_resnet50_fpn', 'googlenet', 'inception_v3', 'keypointrcnn_resnet50_fpn', 'lraspp_mobilenet_v3_large', 'maskrcnn_resnet50_fpn', 'maskrcnn_resnet50_fpn_v2', 'maxvit_t', 'mc3_18', 'mnasnet0_5', 'mnasnet0_75', 'mnasnet1_0', 'mnasnet1_3', 'mobilenet_v2', 'mobilenet_v3_large', 'mobilenet_v3_small', 'mvit_v1_b', 'mvit_v2_s', 'quantized_googlenet', 'quantized_inception_v3', 'quantized_mobilenet_v2', 'quantized_mobilenet_v3_large', 'quantized_resnet18', 'quantized_resnet50', 'quantized_resnext101_32x8d', 'quantized_resnext101_64x4d', 'quantized_shufflenet_v2_x0_5', 'quantized_shufflenet_v2_x1_0', 'quantized_shufflenet_v2_x1_5', 'quantized_shufflenet_v2_x2_0', 'r2plus1d_18', 'r3d_18', 'raft_large', 'raft_small', 'regnet_x_16gf', 'regnet_x_1_6gf', 'regnet_x_32gf', 'regnet_x_3_2gf', 'regnet_x_400mf', 'regnet_x_800mf', 'regnet_x_8gf', 'regnet_y_128gf', 'regnet_y_16gf', 'regnet_y_1_6gf', 'regnet_y_32gf', 'regnet_y_3_2gf', 'regnet_y_400mf', 'regnet_y_800mf', 'regnet_y_8gf', 'resnet101', 'resnet152', 'resnet18', 'resnet34', 'resnet50', 'resnext101_32x8d', 'resnext101_64x4d', 'resnext50_32x4d', 'retinanet_resnet50_fpn', 'retinanet_resnet50_fpn_v2', 's3d', 'shufflenet_v2_x0_5', 'shufflenet_v2_x1_0', 'shufflenet_v2_x1_5', 'shufflenet_v2_x2_0', 'squeezenet1_0', 'squeezenet1_1', 'ssd300_vgg16', 'ssdlite320_mobilenet_v3_large', 'swin3d_b', 'swin3d_s', 'swin3d_t', 'swin_b', 'swin_s', 'swin_t', 'swin_v2_b', 'swin_v2_s', 'swin_v2_t', 'vgg11', 'vgg11_bn', 'vgg13', 'vgg13_bn', 'vgg16', 'vgg16_bn', 'vgg19', 'vgg19_bn', 'vit_b_16', 'vit_b_32', 'vit_h_14', 'vit_l_16', 'vit_l_32', 'wide_resnet101_2', 'wide_resnet50_2']Model Summary
from torchinfo import summary
import numpy as nptimm.create_model?model = timm.create_model('resnet50', pretrained=True, num_classes=10)
random_input = np.random.rand(1, 3, 100, 100)modelResNet(
(conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
(layer1): Sequential(
(0): Bottleneck(
(conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
(downsample): Sequential(
(0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): Bottleneck(
(conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(2): Bottleneck(
(conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
)
(layer2): Sequential(
(0): Bottleneck(
(conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
(downsample): Sequential(
(0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): Bottleneck(
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(2): Bottleneck(
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(3): Bottleneck(
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
)
(layer3): Sequential(
(0): Bottleneck(
(conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
(downsample): Sequential(
(0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): Bottleneck(
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(2): Bottleneck(
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(3): Bottleneck(
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(4): Bottleneck(
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(5): Bottleneck(
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
)
(layer4): Sequential(
(0): Bottleneck(
(conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
(downsample): Sequential(
(0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): Bottleneck(
(conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
(2): Bottleneck(
(conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act1): ReLU(inplace=True)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(drop_block): Identity()
(act2): ReLU(inplace=True)
(aa): Identity()
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(act3): ReLU(inplace=True)
)
)
(global_pool): SelectAdaptivePool2d(pool_type=avg, flatten=Flatten(start_dim=1, end_dim=-1))
(fc): Linear(in_features=2048, out_features=10, bias=True)
)summary(model, input_size=random_input.shape, verbose = 0, depth = 5, col_names = (
"input_size",
"output_size",
"num_params",
"params_percent",
"kernel_size",
"mult_adds",
"trainable",
))=======================================================================================================================================================================================================================
Layer (type:depth-idx) Input Shape Output Shape Param # Param % Kernel Shape Mult-Adds Trainable
=======================================================================================================================================================================================================================
ResNet [1, 3, 100, 100] [1, 10] -- -- -- -- True
├─Conv2d: 1-1 [1, 3, 100, 100] [1, 64, 50, 50] 9,408 0.04% [7, 7] 23,520,000 True
├─BatchNorm2d: 1-2 [1, 64, 50, 50] [1, 64, 50, 50] 128 0.00% -- 128 True
├─ReLU: 1-3 [1, 64, 50, 50] [1, 64, 50, 50] -- -- -- -- --
├─MaxPool2d: 1-4 [1, 64, 50, 50] [1, 64, 25, 25] -- -- 3 -- --
├─Sequential: 1-5 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ └─Bottleneck: 2-1 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ └─Conv2d: 3-1 [1, 64, 25, 25] [1, 64, 25, 25] 4,096 0.02% [1, 1] 2,560,000 True
│ │ └─BatchNorm2d: 3-2 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─ReLU: 3-3 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-4 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ └─BatchNorm2d: 3-5 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─Identity: 3-6 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─ReLU: 3-7 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Identity: 3-8 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-9 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ └─BatchNorm2d: 3-10 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 512 True
│ │ └─Sequential: 3-11 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ │ └─Conv2d: 4-1 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ └─BatchNorm2d: 4-2 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 512 True
│ │ └─ReLU: 3-12 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
│ └─Bottleneck: 2-2 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ └─Conv2d: 3-13 [1, 256, 25, 25] [1, 64, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ └─BatchNorm2d: 3-14 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─ReLU: 3-15 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-16 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ └─BatchNorm2d: 3-17 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─Identity: 3-18 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─ReLU: 3-19 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Identity: 3-20 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-21 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ └─BatchNorm2d: 3-22 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 512 True
│ │ └─ReLU: 3-23 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
│ └─Bottleneck: 2-3 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ └─Conv2d: 3-24 [1, 256, 25, 25] [1, 64, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ └─BatchNorm2d: 3-25 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─ReLU: 3-26 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-27 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ └─BatchNorm2d: 3-28 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 128 True
│ │ └─Identity: 3-29 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─ReLU: 3-30 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Identity: 3-31 [1, 64, 25, 25] [1, 64, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-32 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ └─BatchNorm2d: 3-33 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 512 True
│ │ └─ReLU: 3-34 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
├─Sequential: 1-6 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ └─Bottleneck: 2-4 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ │ └─Conv2d: 3-35 [1, 256, 25, 25] [1, 128, 25, 25] 32,768 0.14% [1, 1] 20,480,000 True
│ │ └─BatchNorm2d: 3-36 [1, 128, 25, 25] [1, 128, 25, 25] 256 0.00% -- 256 True
│ │ └─ReLU: 3-37 [1, 128, 25, 25] [1, 128, 25, 25] -- -- -- -- --
│ │ └─Conv2d: 3-38 [1, 128, 25, 25] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ └─BatchNorm2d: 3-39 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─Identity: 3-40 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─ReLU: 3-41 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Identity: 3-42 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-43 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-44 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 1,024 True
│ │ └─Sequential: 3-45 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ │ │ └─Conv2d: 4-3 [1, 256, 25, 25] [1, 512, 13, 13] 131,072 0.56% [1, 1] 22,151,168 True
│ │ │ └─BatchNorm2d: 4-4 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-46 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ └─Bottleneck: 2-5 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ └─Conv2d: 3-47 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-48 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─ReLU: 3-49 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-50 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ └─BatchNorm2d: 3-51 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─Identity: 3-52 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─ReLU: 3-53 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Identity: 3-54 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-55 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-56 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-57 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ └─Bottleneck: 2-6 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ └─Conv2d: 3-58 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-59 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─ReLU: 3-60 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-61 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ └─BatchNorm2d: 3-62 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─Identity: 3-63 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─ReLU: 3-64 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Identity: 3-65 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-66 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-67 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-68 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ └─Bottleneck: 2-7 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ └─Conv2d: 3-69 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-70 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─ReLU: 3-71 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-72 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ └─BatchNorm2d: 3-73 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 256 True
│ │ └─Identity: 3-74 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─ReLU: 3-75 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Identity: 3-76 [1, 128, 13, 13] [1, 128, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-77 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ └─BatchNorm2d: 3-78 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-79 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
├─Sequential: 1-7 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ └─Bottleneck: 2-8 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-80 [1, 512, 13, 13] [1, 256, 13, 13] 131,072 0.56% [1, 1] 22,151,168 True
│ │ └─BatchNorm2d: 3-81 [1, 256, 13, 13] [1, 256, 13, 13] 512 0.00% -- 512 True
│ │ └─ReLU: 3-82 [1, 256, 13, 13] [1, 256, 13, 13] -- -- -- -- --
│ │ └─Conv2d: 3-83 [1, 256, 13, 13] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-84 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-85 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-86 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-87 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-88 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-89 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─Sequential: 3-90 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ └─Conv2d: 4-5 [1, 512, 13, 13] [1, 1024, 7, 7] 524,288 2.23% [1, 1] 25,690,112 True
│ │ │ └─BatchNorm2d: 4-6 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-91 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─Bottleneck: 2-9 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-92 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-93 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─ReLU: 3-94 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-95 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-96 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-97 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-98 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-99 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-100 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-101 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-102 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─Bottleneck: 2-10 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-103 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-104 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─ReLU: 3-105 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-106 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-107 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-108 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-109 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-110 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-111 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-112 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-113 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─Bottleneck: 2-11 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-114 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-115 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─ReLU: 3-116 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-117 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-118 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-119 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-120 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-121 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-122 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-123 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-124 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─Bottleneck: 2-12 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-125 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-126 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─ReLU: 3-127 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-128 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-129 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-130 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-131 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-132 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-133 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-134 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-135 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─Bottleneck: 2-13 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Conv2d: 3-136 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-137 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─ReLU: 3-138 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-139 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ └─BatchNorm2d: 3-140 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 512 True
│ │ └─Identity: 3-141 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─ReLU: 3-142 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Identity: 3-143 [1, 256, 7, 7] [1, 256, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-144 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ └─BatchNorm2d: 3-145 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 2,048 True
│ │ └─ReLU: 3-146 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
├─Sequential: 1-8 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ └─Bottleneck: 2-14 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ │ └─Conv2d: 3-147 [1, 1024, 7, 7] [1, 512, 7, 7] 524,288 2.23% [1, 1] 25,690,112 True
│ │ └─BatchNorm2d: 3-148 [1, 512, 7, 7] [1, 512, 7, 7] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-149 [1, 512, 7, 7] [1, 512, 7, 7] -- -- -- -- --
│ │ └─Conv2d: 3-150 [1, 512, 7, 7] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ └─BatchNorm2d: 3-151 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 1,024 True
│ │ └─Identity: 3-152 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─ReLU: 3-153 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Identity: 3-154 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Conv2d: 3-155 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ └─BatchNorm2d: 3-156 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 4,096 True
│ │ └─Sequential: 3-157 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ │ │ └─Conv2d: 4-7 [1, 1024, 7, 7] [1, 2048, 4, 4] 2,097,152 8.91% [1, 1] 33,554,432 True
│ │ │ └─BatchNorm2d: 4-8 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 4,096 True
│ │ └─ReLU: 3-158 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
│ └─Bottleneck: 2-15 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- True
│ │ └─Conv2d: 3-159 [1, 2048, 4, 4] [1, 512, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ └─BatchNorm2d: 3-160 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-161 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Conv2d: 3-162 [1, 512, 4, 4] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ └─BatchNorm2d: 3-163 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 1,024 True
│ │ └─Identity: 3-164 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─ReLU: 3-165 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Identity: 3-166 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Conv2d: 3-167 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ └─BatchNorm2d: 3-168 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 4,096 True
│ │ └─ReLU: 3-169 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
│ └─Bottleneck: 2-16 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- True
│ │ └─Conv2d: 3-170 [1, 2048, 4, 4] [1, 512, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ └─BatchNorm2d: 3-171 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 1,024 True
│ │ └─ReLU: 3-172 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Conv2d: 3-173 [1, 512, 4, 4] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ └─BatchNorm2d: 3-174 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 1,024 True
│ │ └─Identity: 3-175 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─ReLU: 3-176 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Identity: 3-177 [1, 512, 4, 4] [1, 512, 4, 4] -- -- -- -- --
│ │ └─Conv2d: 3-178 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ └─BatchNorm2d: 3-179 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 4,096 True
│ │ └─ReLU: 3-180 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
├─SelectAdaptivePool2d: 1-9 [1, 2048, 4, 4] [1, 2048] -- -- -- -- --
│ └─AdaptiveAvgPool2d: 2-17 [1, 2048, 4, 4] [1, 2048, 1, 1] -- -- -- -- --
│ └─Flatten: 2-18 [1, 2048, 1, 1] [1, 2048] -- -- -- -- --
├─Linear: 1-10 [1, 2048] [1, 10] 20,490 0.09% -- 20,490 True
=======================================================================================================================================================================================================================
Total params: 23,528,522
Trainable params: 23,528,522
Non-trainable params: 0
Total mult-adds (Units.MEGABYTES): 995.48
=======================================================================================================================================================================================================================
Input size (MB): 0.12
Forward/backward pass size (MB): 38.94
Params size (MB): 94.11
Estimated Total Size (MB): 133.17
=======================================================================================================================================================================================================================model = timm.create_model('resnetv2_50', pretrained=True, num_classes=10)
random_input = np.random.rand(1, 3, 100, 100)
summary(model, input_size=random_input.shape, verbose = 0, depth = 5, col_names = (
"input_size",
"output_size",
"num_params",
"params_percent",
"kernel_size",
"mult_adds",
"trainable",
))=================================================================================================================================================================================================================================
Layer (type:depth-idx) Input Shape Output Shape Param # Param % Kernel Shape Mult-Adds Trainable
=================================================================================================================================================================================================================================
ResNetV2 [1, 3, 100, 100] [1, 10] -- -- -- -- True
├─Sequential: 1-1 [1, 3, 100, 100] [1, 64, 25, 25] -- -- -- -- True
│ └─Conv2d: 2-1 [1, 3, 100, 100] [1, 64, 50, 50] 9,408 0.04% [7, 7] 23,520,000 True
│ └─MaxPool2d: 2-2 [1, 64, 50, 50] [1, 64, 25, 25] -- -- 3 -- --
├─Sequential: 1-2 [1, 64, 25, 25] [1, 2048, 4, 4] -- -- -- -- True
│ └─ResNetStage: 2-3 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ └─Sequential: 3-1 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ │ └─PreActBottleneck: 4-1 [1, 64, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-1 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─DownsampleConv: 5-2 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% -- 10,240,000 True
│ │ │ │ └─Conv2d: 5-3 [1, 64, 25, 25] [1, 64, 25, 25] 4,096 0.02% [1, 1] 2,560,000 True
│ │ │ │ └─BatchNormAct2d: 5-4 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-5 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ │ │ └─BatchNormAct2d: 5-6 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-7 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ │ └─Identity: 5-8 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-2 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-9 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-10 [1, 256, 25, 25] [1, 64, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ │ └─BatchNormAct2d: 5-11 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-12 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ │ │ └─BatchNormAct2d: 5-13 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-14 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ │ └─Identity: 5-15 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-3 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-16 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-17 [1, 256, 25, 25] [1, 64, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ │ └─BatchNormAct2d: 5-18 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-19 [1, 64, 25, 25] [1, 64, 25, 25] 36,864 0.16% [3, 3] 23,040,000 True
│ │ │ │ └─BatchNormAct2d: 5-20 [1, 64, 25, 25] [1, 64, 25, 25] 128 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-21 [1, 64, 25, 25] [1, 256, 25, 25] 16,384 0.07% [1, 1] 10,240,000 True
│ │ │ │ └─Identity: 5-22 [1, 256, 25, 25] [1, 256, 25, 25] -- -- -- -- --
│ └─ResNetStage: 2-4 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ │ └─Sequential: 3-2 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ │ │ └─PreActBottleneck: 4-4 [1, 256, 25, 25] [1, 512, 13, 13] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-23 [1, 256, 25, 25] [1, 256, 25, 25] 512 0.00% -- 0 True
│ │ │ │ └─DownsampleConv: 5-24 [1, 256, 25, 25] [1, 512, 13, 13] 131,072 0.56% -- 22,151,168 True
│ │ │ │ └─Conv2d: 5-25 [1, 256, 25, 25] [1, 128, 25, 25] 32,768 0.14% [1, 1] 20,480,000 True
│ │ │ │ └─BatchNormAct2d: 5-26 [1, 128, 25, 25] [1, 128, 25, 25] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-27 [1, 128, 25, 25] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ │ │ └─BatchNormAct2d: 5-28 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-29 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─Identity: 5-30 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-5 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-31 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-32 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─BatchNormAct2d: 5-33 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-34 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ │ │ └─BatchNormAct2d: 5-35 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-36 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─Identity: 5-37 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-6 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-38 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-39 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─BatchNormAct2d: 5-40 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-41 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ │ │ └─BatchNormAct2d: 5-42 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-43 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─Identity: 5-44 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-7 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-45 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-46 [1, 512, 13, 13] [1, 128, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─BatchNormAct2d: 5-47 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-48 [1, 128, 13, 13] [1, 128, 13, 13] 147,456 0.63% [3, 3] 24,920,064 True
│ │ │ │ └─BatchNormAct2d: 5-49 [1, 128, 13, 13] [1, 128, 13, 13] 256 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-50 [1, 128, 13, 13] [1, 512, 13, 13] 65,536 0.28% [1, 1] 11,075,584 True
│ │ │ │ └─Identity: 5-51 [1, 512, 13, 13] [1, 512, 13, 13] -- -- -- -- --
│ └─ResNetStage: 2-5 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ │ └─Sequential: 3-3 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ └─PreActBottleneck: 4-8 [1, 512, 13, 13] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-52 [1, 512, 13, 13] [1, 512, 13, 13] 1,024 0.00% -- 0 True
│ │ │ │ └─DownsampleConv: 5-53 [1, 512, 13, 13] [1, 1024, 7, 7] 524,288 2.23% -- 25,690,112 True
│ │ │ │ └─Conv2d: 5-54 [1, 512, 13, 13] [1, 256, 13, 13] 131,072 0.56% [1, 1] 22,151,168 True
│ │ │ │ └─BatchNormAct2d: 5-55 [1, 256, 13, 13] [1, 256, 13, 13] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-56 [1, 256, 13, 13] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-57 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-58 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-59 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-9 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-60 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─Conv2d: 5-61 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─BatchNormAct2d: 5-62 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-63 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-64 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-65 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-66 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-10 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-67 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─Conv2d: 5-68 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─BatchNormAct2d: 5-69 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-70 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-71 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-72 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-73 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-11 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-74 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─Conv2d: 5-75 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─BatchNormAct2d: 5-76 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-77 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-78 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-79 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-80 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-12 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-81 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─Conv2d: 5-82 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─BatchNormAct2d: 5-83 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-84 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-85 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-86 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-87 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-13 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-88 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─Conv2d: 5-89 [1, 1024, 7, 7] [1, 256, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─BatchNormAct2d: 5-90 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-91 [1, 256, 7, 7] [1, 256, 7, 7] 589,824 2.51% [3, 3] 28,901,376 True
│ │ │ │ └─BatchNormAct2d: 5-92 [1, 256, 7, 7] [1, 256, 7, 7] 512 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-93 [1, 256, 7, 7] [1, 1024, 7, 7] 262,144 1.11% [1, 1] 12,845,056 True
│ │ │ │ └─Identity: 5-94 [1, 1024, 7, 7] [1, 1024, 7, 7] -- -- -- -- --
│ └─ResNetStage: 2-6 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ │ └─Sequential: 3-4 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ │ │ └─PreActBottleneck: 4-14 [1, 1024, 7, 7] [1, 2048, 4, 4] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-95 [1, 1024, 7, 7] [1, 1024, 7, 7] 2,048 0.01% -- 0 True
│ │ │ │ └─DownsampleConv: 5-96 [1, 1024, 7, 7] [1, 2048, 4, 4] 2,097,152 8.92% -- 33,554,432 True
│ │ │ │ └─Conv2d: 5-97 [1, 1024, 7, 7] [1, 512, 7, 7] 524,288 2.23% [1, 1] 25,690,112 True
│ │ │ │ └─BatchNormAct2d: 5-98 [1, 512, 7, 7] [1, 512, 7, 7] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-99 [1, 512, 7, 7] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ │ │ └─BatchNormAct2d: 5-100 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-101 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ │ │ └─Identity: 5-102 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-15 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-103 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 0 True
│ │ │ │ └─Conv2d: 5-104 [1, 2048, 4, 4] [1, 512, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ │ │ └─BatchNormAct2d: 5-105 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-106 [1, 512, 4, 4] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ │ │ └─BatchNormAct2d: 5-107 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-108 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ │ │ └─Identity: 5-109 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
│ │ │ └─PreActBottleneck: 4-16 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- True
│ │ │ │ └─BatchNormAct2d: 5-110 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- 0 True
│ │ │ │ └─Conv2d: 5-111 [1, 2048, 4, 4] [1, 512, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ │ │ └─BatchNormAct2d: 5-112 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-113 [1, 512, 4, 4] [1, 512, 4, 4] 2,359,296 10.03% [3, 3] 37,748,736 True
│ │ │ │ └─BatchNormAct2d: 5-114 [1, 512, 4, 4] [1, 512, 4, 4] 1,024 0.00% -- 0 True
│ │ │ │ └─Conv2d: 5-115 [1, 512, 4, 4] [1, 2048, 4, 4] 1,048,576 4.46% [1, 1] 16,777,216 True
│ │ │ │ └─Identity: 5-116 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
├─BatchNormAct2d: 1-3 [1, 2048, 4, 4] [1, 2048, 4, 4] 4,096 0.02% -- -- True
│ └─Identity: 2-7 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
│ └─ReLU: 2-8 [1, 2048, 4, 4] [1, 2048, 4, 4] -- -- -- -- --
├─ClassifierHead: 1-4 [1, 2048, 4, 4] [1, 10] -- -- -- -- True
│ └─SelectAdaptivePool2d: 2-9 [1, 2048, 4, 4] [1, 2048, 1, 1] -- -- -- -- --
│ │ └─AdaptiveAvgPool2d: 3-5 [1, 2048, 4, 4] [1, 2048, 1, 1] -- -- -- -- --
│ │ └─Identity: 3-6 [1, 2048, 1, 1] [1, 2048, 1, 1] -- -- -- -- --
│ └─Dropout: 2-10 [1, 2048, 1, 1] [1, 2048, 1, 1] -- -- -- -- --
│ └─Conv2d: 2-11 [1, 2048, 1, 1] [1, 10, 1, 1] 20,490 0.09% [1, 1] 20,490 True
│ └─Flatten: 2-12 [1, 10, 1, 1] [1, 10] -- -- -- -- --
=================================================================================================================================================================================================================================
Total params: 23,520,842
Trainable params: 23,520,842
Non-trainable params: 0
Total mult-adds (Units.MEGABYTES): 995.42
=================================================================================================================================================================================================================================
Input size (MB): 0.12
Forward/backward pass size (MB): 19.47
Params size (MB): 93.90
Estimated Total Size (MB): 113.49
=================================================================================================================================================================================================================================Detection models
from torchvision.models import detection
import timm############## TENSORBOARD ########################
import sys
import torch.nn.functional as F
from torch.utils.tensorboard import SummaryWriter
# default `log_dir` is "runs" - we'll be more specific here
writer = SummaryWriter()
#################################################### Load ResNet model without the final classification layer
model = timm.create_model('resnetv2_50', pretrained=True, num_classes=10)
# weights = detection.FasterRCNN_ResNet50_FPN_V2_Weights.DEFAULT
# model = detection.fasterrcnn_resnet50_fpn_v2(weights=weights,
# box_score_thresh=0.9).train()import torch
# Generate random input tensor
input_tensor = torch.randn(1, 3, 224, 224)ONNX
torch.onnx.export(model, input_tensor, 'Data/resnet50_v2.onnx', input_names=["features"], output_names=["logits"])############## TENSORBOARD ########################
# writer.add_graph(model, example_data.reshape(-1, 28*28).to(device))
writer.add_graph(model, input_tensor)
writer.flush()
writer.close()
###################################################list(model.children())[:][Sequential(
(conv): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
(pool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
),
Sequential(
(0): ResNetStage(
(blocks): Sequential(
(0): PreActBottleneck(
(downsample): DownsampleConv(
(conv): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm): Identity()
)
(norm1): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(1): PreActBottleneck(
(norm1): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(2): PreActBottleneck(
(norm1): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
)
)
(1): ResNetStage(
(blocks): Sequential(
(0): PreActBottleneck(
(downsample): DownsampleConv(
(conv): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
(norm): Identity()
)
(norm1): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(1): PreActBottleneck(
(norm1): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(2): PreActBottleneck(
(norm1): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(3): PreActBottleneck(
(norm1): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
)
)
(2): ResNetStage(
(blocks): Sequential(
(0): PreActBottleneck(
(downsample): DownsampleConv(
(conv): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
(norm): Identity()
)
(norm1): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(1): PreActBottleneck(
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(2): PreActBottleneck(
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(3): PreActBottleneck(
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(4): PreActBottleneck(
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(5): PreActBottleneck(
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
)
)
(3): ResNetStage(
(blocks): Sequential(
(0): PreActBottleneck(
(downsample): DownsampleConv(
(conv): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
(norm): Identity()
)
(norm1): BatchNormAct2d(
1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(1): PreActBottleneck(
(norm1): BatchNormAct2d(
2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
(2): PreActBottleneck(
(norm1): BatchNormAct2d(
2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(norm3): BatchNormAct2d(
512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
)
(conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
(drop_path): Identity()
)
)
)
),
BatchNormAct2d(
2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True
(drop): Identity()
(act): ReLU(inplace=True)
),
ClassifierHead(
(global_pool): SelectAdaptivePool2d(pool_type=avg, flatten=Identity())
(drop): Dropout(p=0.0, inplace=False)
(fc): Conv2d(2048, 10, kernel_size=(1, 1), stride=(1, 1))
(flatten): Flatten(start_dim=1, end_dim=-1)
)]