├── README.md ├── style ├── in1.jpg ├── in2.jpg ├── in3.jpg └── in4.jpg ├── content ├── in1.jpg ├── in2.jpg ├── in3.jpg └── in4.jpg ├── output ├── 0.4 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.6 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.7 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.8 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg └── 1.0 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── output2 ├── 0.4 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.6 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.7 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.8 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg └── 1.0 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── output3 ├── 0.4 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.6 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.7 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.8 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg └── 1.0 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── output4 ├── 0.4 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.6 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.7 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 0.8 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg └── 1.0 │ ├── in1.jpg │ ├── in2.jpg │ ├── in3.jpg │ └── in4.jpg ├── 269 Final Report.pdf ├── coco.py ├── LICENSE ├── Loader.py ├── .gitignore ├── train1.py ├── train2.py ├── train3.py ├── train4.py ├── transfer.py ├── autoencoder4.py ├── autoencoder3.py ├── autoencoder2.py ├── autoencoder1.py ├── log3 └── nohup.out └── log2 └── nohup.out /README.md: -------------------------------------------------------------------------------- 1 | # resnet-autoencoder 2 | course project for ECS 269 3 | -------------------------------------------------------------------------------- /style/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/style/in1.jpg -------------------------------------------------------------------------------- /style/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/style/in2.jpg -------------------------------------------------------------------------------- /style/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/style/in3.jpg -------------------------------------------------------------------------------- /style/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/style/in4.jpg -------------------------------------------------------------------------------- /content/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/content/in1.jpg -------------------------------------------------------------------------------- /content/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/content/in2.jpg -------------------------------------------------------------------------------- /content/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/content/in3.jpg -------------------------------------------------------------------------------- /content/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/content/in4.jpg -------------------------------------------------------------------------------- /output/0.4/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.4/in1.jpg -------------------------------------------------------------------------------- /output/0.4/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.4/in2.jpg -------------------------------------------------------------------------------- /output/0.4/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.4/in3.jpg -------------------------------------------------------------------------------- /output/0.4/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.4/in4.jpg -------------------------------------------------------------------------------- /output/0.6/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.6/in1.jpg -------------------------------------------------------------------------------- /output/0.6/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.6/in2.jpg -------------------------------------------------------------------------------- /output/0.6/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.6/in3.jpg -------------------------------------------------------------------------------- /output/0.6/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.6/in4.jpg -------------------------------------------------------------------------------- /output/0.7/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.7/in1.jpg -------------------------------------------------------------------------------- /output/0.7/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.7/in2.jpg -------------------------------------------------------------------------------- /output/0.7/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.7/in3.jpg -------------------------------------------------------------------------------- /output/0.7/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.7/in4.jpg -------------------------------------------------------------------------------- /output/0.8/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.8/in1.jpg -------------------------------------------------------------------------------- /output/0.8/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.8/in2.jpg -------------------------------------------------------------------------------- /output/0.8/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.8/in3.jpg -------------------------------------------------------------------------------- /output/0.8/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/0.8/in4.jpg -------------------------------------------------------------------------------- /output/1.0/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/1.0/in1.jpg -------------------------------------------------------------------------------- /output/1.0/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/1.0/in2.jpg -------------------------------------------------------------------------------- /output/1.0/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/1.0/in3.jpg -------------------------------------------------------------------------------- /output/1.0/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output/1.0/in4.jpg -------------------------------------------------------------------------------- /output2/0.4/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.4/in1.jpg -------------------------------------------------------------------------------- /output2/0.4/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.4/in2.jpg -------------------------------------------------------------------------------- /output2/0.4/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.4/in3.jpg -------------------------------------------------------------------------------- /output2/0.4/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.4/in4.jpg -------------------------------------------------------------------------------- /output2/0.6/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.6/in1.jpg -------------------------------------------------------------------------------- /output2/0.6/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.6/in2.jpg -------------------------------------------------------------------------------- /output2/0.6/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.6/in3.jpg -------------------------------------------------------------------------------- /output2/0.6/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.6/in4.jpg -------------------------------------------------------------------------------- /output2/0.7/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.7/in1.jpg -------------------------------------------------------------------------------- /output2/0.7/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.7/in2.jpg -------------------------------------------------------------------------------- /output2/0.7/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.7/in3.jpg -------------------------------------------------------------------------------- /output2/0.7/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.7/in4.jpg -------------------------------------------------------------------------------- /output2/0.8/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.8/in1.jpg -------------------------------------------------------------------------------- /output2/0.8/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.8/in2.jpg -------------------------------------------------------------------------------- /output2/0.8/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.8/in3.jpg -------------------------------------------------------------------------------- /output2/0.8/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/0.8/in4.jpg -------------------------------------------------------------------------------- /output2/1.0/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/1.0/in1.jpg -------------------------------------------------------------------------------- /output2/1.0/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/1.0/in2.jpg -------------------------------------------------------------------------------- /output2/1.0/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/1.0/in3.jpg -------------------------------------------------------------------------------- /output2/1.0/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output2/1.0/in4.jpg -------------------------------------------------------------------------------- /output3/0.4/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.4/in1.jpg -------------------------------------------------------------------------------- /output3/0.4/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.4/in2.jpg -------------------------------------------------------------------------------- /output3/0.4/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.4/in3.jpg -------------------------------------------------------------------------------- /output3/0.4/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.4/in4.jpg -------------------------------------------------------------------------------- /output3/0.6/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.6/in1.jpg -------------------------------------------------------------------------------- /output3/0.6/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.6/in2.jpg -------------------------------------------------------------------------------- /output3/0.6/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.6/in3.jpg -------------------------------------------------------------------------------- /output3/0.6/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.6/in4.jpg -------------------------------------------------------------------------------- /output3/0.7/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.7/in1.jpg -------------------------------------------------------------------------------- /output3/0.7/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.7/in2.jpg -------------------------------------------------------------------------------- /output3/0.7/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.7/in3.jpg -------------------------------------------------------------------------------- /output3/0.7/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.7/in4.jpg -------------------------------------------------------------------------------- /output3/0.8/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.8/in1.jpg -------------------------------------------------------------------------------- /output3/0.8/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.8/in2.jpg -------------------------------------------------------------------------------- /output3/0.8/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.8/in3.jpg -------------------------------------------------------------------------------- /output3/0.8/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/0.8/in4.jpg -------------------------------------------------------------------------------- /output3/1.0/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/1.0/in1.jpg -------------------------------------------------------------------------------- /output3/1.0/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/1.0/in2.jpg -------------------------------------------------------------------------------- /output3/1.0/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/1.0/in3.jpg -------------------------------------------------------------------------------- /output3/1.0/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output3/1.0/in4.jpg -------------------------------------------------------------------------------- /output4/0.4/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.4/in1.jpg -------------------------------------------------------------------------------- /output4/0.4/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.4/in2.jpg -------------------------------------------------------------------------------- /output4/0.4/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.4/in3.jpg -------------------------------------------------------------------------------- /output4/0.4/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.4/in4.jpg -------------------------------------------------------------------------------- /output4/0.6/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.6/in1.jpg -------------------------------------------------------------------------------- /output4/0.6/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.6/in2.jpg -------------------------------------------------------------------------------- /output4/0.6/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.6/in3.jpg -------------------------------------------------------------------------------- /output4/0.6/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.6/in4.jpg -------------------------------------------------------------------------------- /output4/0.7/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.7/in1.jpg -------------------------------------------------------------------------------- /output4/0.7/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.7/in2.jpg -------------------------------------------------------------------------------- /output4/0.7/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.7/in3.jpg -------------------------------------------------------------------------------- /output4/0.7/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.7/in4.jpg -------------------------------------------------------------------------------- /output4/0.8/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.8/in1.jpg -------------------------------------------------------------------------------- /output4/0.8/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.8/in2.jpg -------------------------------------------------------------------------------- /output4/0.8/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.8/in3.jpg -------------------------------------------------------------------------------- /output4/0.8/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/0.8/in4.jpg -------------------------------------------------------------------------------- /output4/1.0/in1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/1.0/in1.jpg -------------------------------------------------------------------------------- /output4/1.0/in2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/1.0/in2.jpg -------------------------------------------------------------------------------- /output4/1.0/in3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/1.0/in3.jpg -------------------------------------------------------------------------------- /output4/1.0/in4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/output4/1.0/in4.jpg -------------------------------------------------------------------------------- /269 Final Report.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Alvinhech/resnet-autoencoder/HEAD/269 Final Report.pdf -------------------------------------------------------------------------------- /coco.py: -------------------------------------------------------------------------------- 1 | import torchvision.datasets 2 | import torchvision.transforms 3 | import torch 4 | 5 | 6 | def load_dataset(path): 7 | data_path = path 8 | transform = torchvision.transforms.Compose([ 9 | torchvision.transforms.Resize((224, 224)), 10 | torchvision.transforms.ToTensor() 11 | ]) 12 | train_dataset = torchvision.datasets.ImageFolder( 13 | root=data_path, 14 | transform=transform 15 | ) 16 | train_loader = torch.utils.data.DataLoader( 17 | train_dataset, 18 | batch_size=16, 19 | num_workers=0, 20 | shuffle=False 21 | ) 22 | return train_loader 23 | 24 | 25 | ''' 26 | need to resize 27 | ''' 28 | 29 | if __name__ == "__main__": 30 | dataloader = load_dataset("/home/achhe_ucdavis_edu/resnet-autoencoder") 31 | for batch_idx, (image, target) in enumerate(dataloader): 32 | image = image.cuda() 33 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2018 Alvin 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /Loader.py: -------------------------------------------------------------------------------- 1 | from PIL import Image 2 | import torchvision.transforms as transforms 3 | import torchvision.utils as vutils 4 | import torch.utils.data as data 5 | from os import listdir 6 | from os.path import join 7 | import numpy as np 8 | import torch 9 | import os 10 | import torch.nn as nn 11 | from torch.autograd import Variable 12 | import numpy as np 13 | 14 | 15 | def is_image_file(filename): 16 | return any(filename.endswith(extension) for extension in [".png", ".jpg", ".jpeg"]) 17 | 18 | 19 | def default_loader(path): 20 | return Image.open(path).convert('RGB') 21 | 22 | 23 | class Dataset(data.Dataset): 24 | def __init__(self, contentPath, stylePath): 25 | super(Dataset, self).__init__() 26 | self.contentPath = contentPath 27 | self.image_list = [x for x in listdir(contentPath) if is_image_file(x)] 28 | self.stylePath = stylePath 29 | 30 | def __getitem__(self, index): 31 | contentImgPath = os.path.join(self.contentPath, self.image_list[index]) 32 | styleImgPath = os.path.join(self.stylePath, self.image_list[index]) 33 | contentImg = default_loader(contentImgPath) 34 | styleImg = default_loader(styleImgPath) 35 | 36 | # resize 37 | contentImg = contentImg.resize((224, 224)) 38 | styleImg = styleImg.resize((224, 224)) 39 | 40 | # Preprocess Images 41 | contentImg = transforms.ToTensor()(contentImg) 42 | styleImg = transforms.ToTensor()(styleImg) 43 | return contentImg.squeeze(0), styleImg.squeeze(0), self.image_list[index] 44 | 45 | def __len__(self): 46 | # You should change 0 to the total size of your dataset. 47 | return len(self.image_list) 48 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | *.egg-info/ 24 | .installed.cfg 25 | *.egg 26 | MANIFEST 27 | 28 | # PyInstaller 29 | # Usually these files are written by a python script from a template 30 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 31 | *.manifest 32 | *.spec 33 | 34 | # Installer logs 35 | pip-log.txt 36 | pip-delete-this-directory.txt 37 | 38 | # Unit test / coverage reports 39 | htmlcov/ 40 | .tox/ 41 | .coverage 42 | .coverage.* 43 | .cache 44 | nosetests.xml 45 | coverage.xml 46 | *.cover 47 | .hypothesis/ 48 | .pytest_cache/ 49 | 50 | # Translations 51 | *.mo 52 | *.pot 53 | 54 | # Django stuff: 55 | *.log 56 | local_settings.py 57 | db.sqlite3 58 | 59 | # Flask stuff: 60 | instance/ 61 | .webassets-cache 62 | 63 | # Scrapy stuff: 64 | .scrapy 65 | 66 | # Sphinx documentation 67 | docs/_build/ 68 | 69 | # PyBuilder 70 | target/ 71 | 72 | # Jupyter Notebook 73 | .ipynb_checkpoints 74 | 75 | # pyenv 76 | .python-version 77 | 78 | # celery beat schedule file 79 | celerybeat-schedule 80 | 81 | # SageMath parsed files 82 | *.sage.py 83 | 84 | # Environments 85 | .env 86 | .venv 87 | env/ 88 | venv/ 89 | ENV/ 90 | env.bak/ 91 | venv.bak/ 92 | 93 | # Spyder project settings 94 | .spyderproject 95 | .spyproject 96 | 97 | # Rope project settings 98 | .ropeproject 99 | 100 | # mkdocs documentation 101 | /site 102 | 103 | # mypy 104 | .mypy_cache/ 105 | -------------------------------------------------------------------------------- /train1.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | from autoencoder1 import ResNet_autoencoder, Bottleneck, DeconvBottleneck 4 | from coco import load_dataset 5 | import matplotlib.pyplot as plt 6 | from torch.autograd import Variable 7 | 8 | EPOCH = 10 9 | 10 | if __name__ == "__main__": 11 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 12 | 3, 4, 6, 3], 3).cuda() 13 | 14 | # load data 15 | print("start loading.") 16 | dataloader = load_dataset('/home/achhe_ucdavis_edu/resnet-autoencoder/data') 17 | print("load data success.") 18 | ''' 19 | load pre_trained_model 20 | ''' 21 | pretrained_dict = torch.load('./resnet50-19c8e357.pth') 22 | print("load pretrained model success") 23 | 24 | model_dict = model.state_dict() 25 | # 1. filter out unnecessary keys 26 | pretrained_dict = {k: v for k, 27 | v in pretrained_dict.items() if k in model_dict} 28 | # 2. overwrite entries in the existing state dict 29 | model_dict.update(pretrained_dict) 30 | model.load_state_dict(model_dict) 31 | 32 | # fix encoder 33 | fix_length = len(pretrained_dict.keys()) 34 | all_length = len(model.state_dict().keys()) 35 | for idx, k in enumerate(model_dict.keys()): 36 | if idx < fix_length: 37 | model.state_dict()[k].requires_grad = False 38 | 39 | params = filter(lambda p: p.requires_grad, model.parameters()) 40 | 41 | # Loss and Optimizer 42 | criterion = nn.MSELoss() 43 | 44 | optimizer = torch.optim.Adam(params, lr=1e-4) 45 | 46 | model.train() 47 | 48 | loss_list=[] 49 | 50 | print("start training.") 51 | 52 | for epoch in range(EPOCH): 53 | for batch_idx, (image, target) in enumerate(dataloader): 54 | image = Variable(image.cuda()) 55 | 56 | # Forward + Backward + Optimize 57 | 58 | optimizer.zero_grad() 59 | 60 | tmp1, tmp2= model(image) 61 | 62 | loss = criterion(tmp2,image.detach()) 63 | 64 | loss.backward() 65 | 66 | optimizer.step() 67 | 68 | if (batch_idx+1) % 10 == 0: 69 | print ("Epoch [%d/%d], Iter [%d] Loss: %.4f" % (epoch+1, EPOCH, batch_idx+1, loss.data[0])) 70 | ''' 71 | loss_list.append(loss) 72 | 73 | plt.plot(loss_list) 74 | plt.ylable('loss') 75 | plt.show() 76 | ''' 77 | if((batch_idx+1)%1000==0): 78 | torch.save(model.state_dict(), './save5/resnet'+str(epoch+1)+'_'+str(batch_idx+1)+'.pkl') 79 | -------------------------------------------------------------------------------- /train2.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | from autoencoder2 import ResNet_autoencoder, Bottleneck, DeconvBottleneck 4 | from coco import load_dataset 5 | import matplotlib.pyplot as plt 6 | from torch.autograd import Variable 7 | 8 | EPOCH = 10 9 | 10 | if __name__ == "__main__": 11 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 12 | 3, 4, 6, 3], 3).cuda() 13 | 14 | # load data 15 | print("start loading.") 16 | dataloader = load_dataset('/home/achhe_ucdavis_edu/resnet-autoencoder/data') 17 | print("load data success.") 18 | ''' 19 | load pre_trained_model 20 | ''' 21 | pretrained_dict = torch.load('./resnet50-19c8e357.pth') 22 | print("load pretrained model success") 23 | 24 | model_dict = model.state_dict() 25 | # 1. filter out unnecessary keys 26 | pretrained_dict = {k: v for k, 27 | v in pretrained_dict.items() if k in model_dict} 28 | # 2. overwrite entries in the existing state dict 29 | model_dict.update(pretrained_dict) 30 | model.load_state_dict(model_dict) 31 | 32 | # fix encoder 33 | fix_length = len(pretrained_dict.keys()) 34 | all_length = len(model.state_dict().keys()) 35 | for idx, k in enumerate(model_dict.keys()): 36 | if idx < fix_length: 37 | model.state_dict()[k].requires_grad = False 38 | 39 | params = filter(lambda p: p.requires_grad, model.parameters()) 40 | 41 | # Loss and Optimizer 42 | criterion = nn.MSELoss() 43 | 44 | optimizer = torch.optim.Adam(params, lr=1e-4) 45 | 46 | model.train() 47 | 48 | loss_list=[] 49 | 50 | print("start training.") 51 | 52 | for epoch in range(EPOCH): 53 | for batch_idx, (image, target) in enumerate(dataloader): 54 | image = Variable(image.cuda()) 55 | 56 | # Forward + Backward + Optimize 57 | 58 | optimizer.zero_grad() 59 | 60 | tmp1, tmp2, tmp3 = model(image) 61 | 62 | loss1 = criterion(tmp2,image.detach()) 63 | loss2 = criterion(tmp3,tmp1.detach()) 64 | loss = loss1 + loss2 65 | 66 | loss.backward() 67 | 68 | optimizer.step() 69 | 70 | if (batch_idx+1) % 10 == 0: 71 | print ("Epoch [%d/%d], Iter [%d] Loss: %.4f" % (epoch+1, EPOCH, batch_idx+1, loss.data[0])) 72 | ''' 73 | loss_list.append(loss) 74 | 75 | plt.plot(loss_list) 76 | plt.ylable('loss') 77 | plt.show() 78 | ''' 79 | if((batch_idx+1)%1000==0): 80 | torch.save(model.state_dict(), './save2/resnet'+str(epoch+1)+'_'+str(batch_idx+1)+'.pkl') 81 | -------------------------------------------------------------------------------- /train3.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | from autoencoder3 import ResNet_autoencoder, Bottleneck, DeconvBottleneck 4 | from coco import load_dataset 5 | import matplotlib.pyplot as plt 6 | from torch.autograd import Variable 7 | 8 | EPOCH = 10 9 | 10 | if __name__ == "__main__": 11 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 12 | 3, 4, 6, 3], 3).cuda() 13 | 14 | # load data 15 | print("start loading.") 16 | dataloader = load_dataset('/home/achhe_ucdavis_edu/resnet-autoencoder/data') 17 | print("load data success.") 18 | ''' 19 | load pre_trained_model 20 | ''' 21 | pretrained_dict = torch.load('./resnet50-19c8e357.pth') 22 | print("load pretrained model success") 23 | 24 | model_dict = model.state_dict() 25 | # 1. filter out unnecessary keys 26 | pretrained_dict = {k: v for k, 27 | v in pretrained_dict.items() if k in model_dict} 28 | # 2. overwrite entries in the existing state dict 29 | model_dict.update(pretrained_dict) 30 | model.load_state_dict(model_dict) 31 | 32 | # fix encoder 33 | fix_length = len(pretrained_dict.keys()) 34 | all_length = len(model.state_dict().keys()) 35 | for idx, k in enumerate(model_dict.keys()): 36 | if idx < fix_length: 37 | model.state_dict()[k].requires_grad = False 38 | 39 | params = filter(lambda p: p.requires_grad, model.parameters()) 40 | 41 | # Loss and Optimizer 42 | criterion = nn.MSELoss() 43 | 44 | optimizer = torch.optim.Adam(params, lr=1e-4) 45 | 46 | model.train() 47 | 48 | loss_list=[] 49 | 50 | print("start training.") 51 | 52 | for epoch in range(EPOCH): 53 | for batch_idx, (image, target) in enumerate(dataloader): 54 | image = Variable(image.cuda()) 55 | 56 | # Forward + Backward + Optimize 57 | 58 | optimizer.zero_grad() 59 | 60 | tmp1, tmp2, tmp3 = model(image) 61 | 62 | loss1 = criterion(tmp2,image.detach()) 63 | loss2 = criterion(tmp3,tmp1.detach()) 64 | loss = loss1 + loss2 65 | 66 | loss.backward() 67 | 68 | optimizer.step() 69 | 70 | if (batch_idx+1) % 10 == 0: 71 | print ("Epoch [%d/%d], Iter [%d] Loss: %.4f" % (epoch+1, EPOCH, batch_idx+1, loss.data[0])) 72 | ''' 73 | loss_list.append(loss) 74 | 75 | plt.plot(loss_list) 76 | plt.ylable('loss') 77 | plt.show() 78 | ''' 79 | if((batch_idx+1)%1000==0): 80 | torch.save(model.state_dict(), './save3/resnet'+str(epoch+1)+'_'+str(batch_idx+1)+'.pkl') 81 | -------------------------------------------------------------------------------- /train4.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | from autoencoder4 import ResNet_autoencoder, Bottleneck, DeconvBottleneck 4 | from coco import load_dataset 5 | import matplotlib.pyplot as plt 6 | from torch.autograd import Variable 7 | 8 | EPOCH = 10 9 | 10 | if __name__ == "__main__": 11 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 12 | 3, 4, 6, 3], 3).cuda() 13 | 14 | # load data 15 | print("start loading.") 16 | dataloader = load_dataset('/home/achhe_ucdavis_edu/resnet-autoencoder/data') 17 | print("load data success.") 18 | ''' 19 | load pre_trained_model 20 | ''' 21 | pretrained_dict = torch.load('./resnet50-19c8e357.pth') 22 | print("load pretrained model success") 23 | 24 | model_dict = model.state_dict() 25 | # 1. filter out unnecessary keys 26 | pretrained_dict = {k: v for k, 27 | v in pretrained_dict.items() if k in model_dict} 28 | # 2. overwrite entries in the existing state dict 29 | model_dict.update(pretrained_dict) 30 | model.load_state_dict(model_dict) 31 | 32 | # fix encoder 33 | fix_length = len(pretrained_dict.keys()) 34 | all_length = len(model.state_dict().keys()) 35 | for idx, k in enumerate(model_dict.keys()): 36 | if idx < fix_length: 37 | model.state_dict()[k].requires_grad = False 38 | 39 | params = filter(lambda p: p.requires_grad, model.parameters()) 40 | 41 | # Loss and Optimizer 42 | criterion = nn.MSELoss() 43 | 44 | optimizer = torch.optim.Adam(params, lr=1e-4) 45 | 46 | model.train() 47 | 48 | loss_list=[] 49 | 50 | print("start training.") 51 | 52 | for epoch in range(EPOCH): 53 | for batch_idx, (image, target) in enumerate(dataloader): 54 | image = Variable(image.cuda()) 55 | 56 | # Forward + Backward + Optimize 57 | 58 | optimizer.zero_grad() 59 | 60 | tmp1, tmp2, tmp3 = model(image) 61 | 62 | loss1 = criterion(tmp2,image.detach()) 63 | loss2 = criterion(tmp3,tmp1.detach()) 64 | loss = loss1 + loss2 65 | 66 | loss.backward() 67 | 68 | optimizer.step() 69 | 70 | if (batch_idx+1) % 10 == 0: 71 | print ("Epoch [%d/%d], Iter [%d] Loss: %.4f" % (epoch+1, EPOCH, batch_idx+1, loss.data[0])) 72 | ''' 73 | loss_list.append(loss) 74 | 75 | plt.plot(loss_list) 76 | plt.ylable('loss') 77 | plt.show() 78 | ''' 79 | if((batch_idx+1)%1000==0): 80 | torch.save(model.state_dict(), './save4/resnet'+str(epoch+1)+'_'+str(batch_idx+1)+'.pkl') 81 | -------------------------------------------------------------------------------- /transfer.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torchvision.transforms as transforms 4 | from autoencoder4 import ResNet_autoencoder, Bottleneck, DeconvBottleneck 5 | from coco import load_dataset 6 | import matplotlib.pyplot as plt 7 | from torch.autograd import Variable 8 | from Loader import Dataset 9 | import torchvision.transforms.functional as F 10 | from PIL import Image 11 | 12 | 13 | def wct(c, s, alpha): 14 | ''' 15 | Apply whitening and coloring transforms on content and style images. 16 | 17 | :param c: matrix of content image 18 | :param s: matrix of style image 19 | :param alpha: coefficient to control the strength of transformation 20 | :return: transformed matrix 21 | ''' 22 | 23 | # content image whitening 24 | c = c.double() 25 | c_channels, c_width, c_height = c.size(0), c.size(1), c.size(2) 26 | c_new = c.view(c_channels, -1) # c * (h * w) 27 | c_mean = torch.mean(c_new, 1) # c 28 | c_mean = c_mean.unsqueeze(1).expand_as(c_new) # c * 1 -> c * (h * w) 29 | c_new = c_new - c_mean # subtract mean 30 | 31 | c_cov = torch.mm(c_new, c_new.t()).div( 32 | c_width * c_height - 1) # covariance matrix of c*cT 33 | # singular value decomposition, c_e is a diagonal matrix 34 | c_u, c_e, c_v = torch.svd(c_cov, some=False) 35 | # find all positive eigenvalues of c*cT at the beginning 36 | c_pos_eigens_idx = c_channels 37 | for i in range(c_channels): 38 | if c_e[i] < 0.00001: 39 | c_pos_eigens_idx = i 40 | break 41 | # list of all positive eigenvalues of c*cT at the beginning 42 | c_pos_eigens = c_e[0: c_pos_eigens_idx] 43 | c_o = c_v[:, 0:c_pos_eigens_idx] # orthogonal matrix of eigenvalues, Ec 44 | 45 | whitened = torch.mm(c_o, torch.diag( 46 | c_pos_eigens.pow(-0.5))) # Ec * Dc^(-1/2) 47 | whitened = torch.mm(whitened, c_o.t()) # Ec * Dc^(-1/2) * EcT 48 | whitened = torch.mm(whitened, c_new) # Ec * Dc^(-1/2) * EcT * c 49 | 50 | # style image coloring 51 | s = s.double() 52 | _, s_width, s_height = s.size(0), s.size(1), s.size(2) 53 | s_new = s.view(c_channels, -1) # c * (h * w) 54 | s_mean = torch.mean(s_new, 1) # c 55 | s_mean = s_mean.unsqueeze(1).expand_as(s_new) # c * 1 -> c * (h * w) 56 | s_new = s_new - s_mean 57 | 58 | s_cov = torch.mm(s_new, s_new.t()).div( 59 | s_width * s_height - 1) # covariance matrix of s*sT 60 | # singular value decomposition, s_e is a diagonal matrix 61 | s_u, s_e, s_v = torch.svd(s_cov, some=False) 62 | # find all positive eigenvalues of s*sT at the beginning 63 | s_pos_eigens_idx = c_channels 64 | for i in range(c_channels): 65 | if s_e[i] < 0.00001: 66 | s_pos_eigens_idx = i 67 | break 68 | # list of all positive eigenvalues of s*sT at the beginning 69 | s_pos_eigens = s_e[0: s_pos_eigens_idx] 70 | s_o = s_v[:, 0:s_pos_eigens_idx] # orthogonal matrix of eigenvalues, Es 71 | 72 | colored = torch.mm(s_o, torch.diag(s_pos_eigens.pow(0.5))) # Es * Ds^(1/2) 73 | colored = torch.mm(colored, s_o.t()) # Es * Ds^(1/2) * EsT 74 | colored = torch.mm(colored, whitened) # Es * Ds^(1/2) * EsT * whitened 75 | # re-center the colored matrix with the mean 76 | colored = colored + s_mean.resize_(colored.size()) 77 | colored = colored.view_as(c) 78 | 79 | # use alpha to control the strength of transformation 80 | transformed = alpha * colored + (1.0 - alpha) * c 81 | return transformed.float().unsqueeze(0) 82 | 83 | 84 | PATH = "./save4/resnet3_2000.pkl" 85 | 86 | 87 | def default_loader(path): 88 | return Image.open(path).convert('RGB') 89 | 90 | 91 | if __name__ == "__main__": 92 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 93 | 3, 4, 6, 3], 3).cuda() 94 | 95 | model.load_state_dict(torch.load(PATH)) 96 | model.eval() 97 | 98 | # Data loading code 99 | dataset = Dataset("./content", "./style", ) 100 | loader = torch.utils.data.DataLoader(dataset=dataset, 101 | batch_size=1, 102 | shuffle=False) 103 | alpha_list = [0.4, 0.6, 0.7, 0.8, 1.0] 104 | for alpha in alpha_list: 105 | for i, (contentImg, styleImg, imname) in enumerate(loader): 106 | imname = imname[0] 107 | print('Transferring ' + imname) 108 | contentImg = contentImg.cuda() 109 | styleImg = styleImg.cuda() 110 | cImg = Variable(contentImg, volatile=True) 111 | sImg = Variable(styleImg, volatile=True) 112 | 113 | c, s = model.encoder(cImg), model.encoder(sImg) 114 | c = c.data.cpu().squeeze(0) 115 | s = s.data.cpu().squeeze(0) 116 | o = wct(c, s, alpha) 117 | o = Variable(o.cuda(), volatile=True) 118 | out = model.decoder(o, cImg.size()) 119 | out = out.data.cpu().squeeze(0) 120 | output = F.to_pil_image(out) 121 | output.save("./output4/" + str(alpha) + "/" + imname) 122 | -------------------------------------------------------------------------------- /autoencoder4.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torch.nn.functional as F 4 | 5 | ''' 6 | batch_size 8 7 | epoch every 100 checkpoint 8 | learning_rate 1e-4? 9 | ''' 10 | 11 | 12 | class Bottleneck(nn.Module): 13 | expansion = 4 14 | 15 | def __init__(self, in_channels, out_channels, stride=1, downsample=None): 16 | super(Bottleneck, self).__init__() 17 | self.conv1 = nn.Conv2d(in_channels, out_channels, 18 | kernel_size=1, bias=False) 19 | self.bn1 = nn.BatchNorm2d(out_channels) 20 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 21 | stride=stride, bias=False, padding=1) 22 | self.bn2 = nn.BatchNorm2d(out_channels) 23 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 24 | kernel_size=1, bias=False) 25 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 26 | self.relu = nn.ReLU() 27 | self.downsample = downsample 28 | 29 | def forward(self, x): 30 | shortcut = x 31 | 32 | out = self.conv1(x) 33 | out = self.bn1(out) 34 | out = self.relu(out) 35 | 36 | out = self.conv2(out) 37 | out = self.bn2(out) 38 | out = self.relu(out) 39 | 40 | out = self.conv3(out) 41 | out = self.bn3(out) 42 | out = self.relu(out) 43 | 44 | if self.downsample is not None: 45 | shortcut = self.downsample(x) 46 | 47 | out += shortcut 48 | out = self.relu(out) 49 | 50 | return out 51 | 52 | 53 | class DeconvBottleneck(nn.Module): 54 | def __init__(self, in_channels, out_channels, expansion=2, stride=1, upsample=None): 55 | super(DeconvBottleneck, self).__init__() 56 | self.expansion = expansion 57 | self.conv1 = nn.Conv2d(in_channels, out_channels, 58 | kernel_size=1, bias=False) 59 | self.bn1 = nn.BatchNorm2d(out_channels) 60 | if stride == 1: 61 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 62 | stride=stride, bias=False, padding=1) 63 | else: 64 | self.conv2 = nn.ConvTranspose2d(out_channels, out_channels, 65 | kernel_size=3, 66 | stride=stride, bias=False, 67 | padding=1, 68 | output_padding=1) 69 | self.bn2 = nn.BatchNorm2d(out_channels) 70 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 71 | kernel_size=1, bias=False) 72 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 73 | self.relu = nn.ReLU() 74 | self.upsample = upsample 75 | 76 | def forward(self, x): 77 | shortcut = x 78 | 79 | out = self.conv1(x) 80 | out = self.bn1(out) 81 | out = self.relu(out) 82 | 83 | out = self.conv2(out) 84 | out = self.bn2(out) 85 | out = self.relu(out) 86 | 87 | out = self.conv3(out) 88 | out = self.bn3(out) 89 | out = self.relu(out) 90 | 91 | if self.upsample is not None: 92 | shortcut = self.upsample(x) 93 | 94 | out += shortcut 95 | out = self.relu(out) 96 | 97 | return out 98 | 99 | 100 | class ResNet_autoencoder(nn.Module): 101 | def __init__(self, downblock, upblock, num_layers, n_classes): 102 | super(ResNet_autoencoder, self).__init__() 103 | 104 | self.in_channels = 64 105 | 106 | self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, 107 | bias=False) 108 | self.bn1 = nn.BatchNorm2d(64) 109 | self.relu = nn.ReLU() 110 | self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) 111 | 112 | self.layer1 = self._make_downlayer(downblock, 64, num_layers[0]) 113 | self.layer2 = self._make_downlayer(downblock, 128, num_layers[1], 114 | stride=2) 115 | self.layer3 = self._make_downlayer(downblock, 256, num_layers[2], 116 | stride=2) 117 | self.layer4 = self._make_downlayer(downblock, 512, num_layers[3], 118 | stride=2) 119 | 120 | self.uplayer1 = self._make_up_block( 121 | upblock, 512, num_layers[3], stride=2) 122 | self.uplayer2 = self._make_up_block( 123 | upblock, 256, num_layers[2], stride=2) 124 | self.uplayer3 = self._make_up_block( 125 | upblock, 128, num_layers[1], stride=2) 126 | self.uplayer4 = self._make_up_block( 127 | upblock, 64, num_layers[0], stride=2) 128 | 129 | upsample = nn.Sequential( 130 | nn.ConvTranspose2d(self.in_channels, # 256 131 | 64, 132 | kernel_size=1, stride=2, 133 | bias=False, output_padding=1), 134 | nn.BatchNorm2d(64), 135 | ) 136 | self.uplayer_top = DeconvBottleneck( 137 | self.in_channels, 64, 1, 2, upsample) 138 | 139 | self.conv1_1 = nn.ConvTranspose2d(64, n_classes, kernel_size=1, stride=1, 140 | bias=False) 141 | 142 | def _make_downlayer(self, block, init_channels, num_layer, stride=1): 143 | downsample = None 144 | if stride != 1 or self.in_channels != init_channels * block.expansion: 145 | downsample = nn.Sequential( 146 | nn.Conv2d(self.in_channels, init_channels * block.expansion, 147 | kernel_size=1, stride=stride, bias=False), 148 | nn.BatchNorm2d(init_channels * block.expansion), 149 | ) 150 | layers = [] 151 | layers.append( 152 | block(self.in_channels, init_channels, stride, downsample)) 153 | self.in_channels = init_channels * block.expansion 154 | for i in range(1, num_layer): 155 | layers.append(block(self.in_channels, init_channels)) 156 | 157 | return nn.Sequential(*layers) 158 | 159 | def _make_up_block(self, block, init_channels, num_layer, stride=1): 160 | upsample = None 161 | # expansion = block.expansion 162 | if stride != 1 or self.in_channels != init_channels * 2: 163 | upsample = nn.Sequential( 164 | nn.ConvTranspose2d(self.in_channels, init_channels * 2, 165 | kernel_size=1, stride=stride, 166 | bias=False, output_padding=1), 167 | nn.BatchNorm2d(init_channels * 2), 168 | ) 169 | layers = [] 170 | for i in range(1, num_layer): 171 | layers.append(block(self.in_channels, init_channels, 4)) 172 | layers.append( 173 | block(self.in_channels, init_channels, 2, stride, upsample)) 174 | self.in_channels = init_channels * 2 175 | return nn.Sequential(*layers) 176 | 177 | def encoder(self, x): 178 | x = self.conv1(x) 179 | x = self.bn1(x) 180 | x = self.relu(x) 181 | x = self.maxpool(x) 182 | 183 | x = self.layer1(x) 184 | return x 185 | 186 | def decoder(self, x, image_size): 187 | x = self.uplayer4(x) 188 | x = self.uplayer_top(x) 189 | 190 | x = self.conv1_1(x, output_size=image_size) 191 | return x 192 | 193 | def forward(self, x): 194 | img = x 195 | tmp1 = self.encoder(x) 196 | tmp2 = self.decoder(tmp1, img.size()) 197 | tmp3 = self.encoder(tmp2) 198 | 199 | return tmp1, tmp2, tmp3 200 | 201 | 202 | def ResNet50(**kwargs): 203 | return ResNet_autoencoder(Bottleneck, DeconvBottleneck, [3, 4, 6, 3], 3, **kwargs) 204 | 205 | 206 | def ResNet101(**kwargs): 207 | return ResNet_autoencoder(Bottleneck, [3, 4, 23, 2], 3, **kwargs) 208 | 209 | 210 | if __name__ == "__main__": 211 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 212 | 3, 4, 6, 3], 3).cuda() 213 | ''' 214 | load pre_trained_model 215 | ''' 216 | pretrained_dict = torch.load("./resnet50-19c8e357.pth") 217 | model_dict = model.state_dict() 218 | # 1. filter out unnecessary keys 219 | pretrained_dict = {k: v for k, 220 | v in pretrained_dict.items() if k in model_dict} 221 | # 2. overwrite entries in the existing state dict 222 | model_dict.update(pretrained_dict) 223 | model.load_state_dict(model_dict) 224 | 225 | input = torch.autograd.Variable(torch.randn(2, 3, 224, 224)).cuda() 226 | o = model(input) 227 | print(o) 228 | -------------------------------------------------------------------------------- /autoencoder3.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torch.nn.functional as F 4 | 5 | ''' 6 | batch_size 8 7 | epoch every 100 checkpoint 8 | learning_rate 1e-4? 9 | ''' 10 | 11 | 12 | class Bottleneck(nn.Module): 13 | expansion = 4 14 | 15 | def __init__(self, in_channels, out_channels, stride=1, downsample=None): 16 | super(Bottleneck, self).__init__() 17 | self.conv1 = nn.Conv2d(in_channels, out_channels, 18 | kernel_size=1, bias=False) 19 | self.bn1 = nn.BatchNorm2d(out_channels) 20 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 21 | stride=stride, bias=False, padding=1) 22 | self.bn2 = nn.BatchNorm2d(out_channels) 23 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 24 | kernel_size=1, bias=False) 25 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 26 | self.relu = nn.ReLU() 27 | self.downsample = downsample 28 | 29 | def forward(self, x): 30 | shortcut = x 31 | 32 | out = self.conv1(x) 33 | out = self.bn1(out) 34 | out = self.relu(out) 35 | 36 | out = self.conv2(out) 37 | out = self.bn2(out) 38 | out = self.relu(out) 39 | 40 | out = self.conv3(out) 41 | out = self.bn3(out) 42 | out = self.relu(out) 43 | 44 | if self.downsample is not None: 45 | shortcut = self.downsample(x) 46 | 47 | out += shortcut 48 | out = self.relu(out) 49 | 50 | return out 51 | 52 | 53 | class DeconvBottleneck(nn.Module): 54 | def __init__(self, in_channels, out_channels, expansion=2, stride=1, upsample=None): 55 | super(DeconvBottleneck, self).__init__() 56 | self.expansion = expansion 57 | self.conv1 = nn.Conv2d(in_channels, out_channels, 58 | kernel_size=1, bias=False) 59 | self.bn1 = nn.BatchNorm2d(out_channels) 60 | if stride == 1: 61 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 62 | stride=stride, bias=False, padding=1) 63 | else: 64 | self.conv2 = nn.ConvTranspose2d(out_channels, out_channels, 65 | kernel_size=3, 66 | stride=stride, bias=False, 67 | padding=1, 68 | output_padding=1) 69 | self.bn2 = nn.BatchNorm2d(out_channels) 70 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 71 | kernel_size=1, bias=False) 72 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 73 | self.relu = nn.ReLU() 74 | self.upsample = upsample 75 | 76 | def forward(self, x): 77 | shortcut = x 78 | 79 | out = self.conv1(x) 80 | out = self.bn1(out) 81 | out = self.relu(out) 82 | 83 | out = self.conv2(out) 84 | out = self.bn2(out) 85 | out = self.relu(out) 86 | 87 | out = self.conv3(out) 88 | out = self.bn3(out) 89 | out = self.relu(out) 90 | 91 | if self.upsample is not None: 92 | shortcut = self.upsample(x) 93 | 94 | out += shortcut 95 | out = self.relu(out) 96 | 97 | return out 98 | 99 | 100 | class ResNet_autoencoder(nn.Module): 101 | def __init__(self, downblock, upblock, num_layers, n_classes): 102 | super(ResNet_autoencoder, self).__init__() 103 | 104 | self.in_channels = 64 105 | 106 | self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, 107 | bias=False) 108 | self.bn1 = nn.BatchNorm2d(64) 109 | self.relu = nn.ReLU() 110 | self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) 111 | 112 | self.layer1 = self._make_downlayer(downblock, 64, num_layers[0]) 113 | self.layer2 = self._make_downlayer(downblock, 128, num_layers[1], 114 | stride=2) 115 | self.layer3 = self._make_downlayer(downblock, 256, num_layers[2], 116 | stride=2) 117 | self.layer4 = self._make_downlayer(downblock, 512, num_layers[3], 118 | stride=2) 119 | 120 | self.uplayer1 = self._make_up_block( 121 | upblock, 512, num_layers[3], stride=2) 122 | self.uplayer2 = self._make_up_block( 123 | upblock, 256, num_layers[2], stride=2) 124 | self.uplayer3 = self._make_up_block( 125 | upblock, 128, num_layers[1], stride=2) 126 | self.uplayer4 = self._make_up_block( 127 | upblock, 64, num_layers[0], stride=2) 128 | 129 | upsample = nn.Sequential( 130 | nn.ConvTranspose2d(self.in_channels, # 256 131 | 64, 132 | kernel_size=1, stride=2, 133 | bias=False, output_padding=1), 134 | nn.BatchNorm2d(64), 135 | ) 136 | self.uplayer_top = DeconvBottleneck( 137 | self.in_channels, 64, 1, 2, upsample) 138 | 139 | self.conv1_1 = nn.ConvTranspose2d(64, n_classes, kernel_size=1, stride=1, 140 | bias=False) 141 | 142 | def _make_downlayer(self, block, init_channels, num_layer, stride=1): 143 | downsample = None 144 | if stride != 1 or self.in_channels != init_channels * block.expansion: 145 | downsample = nn.Sequential( 146 | nn.Conv2d(self.in_channels, init_channels * block.expansion, 147 | kernel_size=1, stride=stride, bias=False), 148 | nn.BatchNorm2d(init_channels * block.expansion), 149 | ) 150 | layers = [] 151 | layers.append( 152 | block(self.in_channels, init_channels, stride, downsample)) 153 | self.in_channels = init_channels * block.expansion 154 | for i in range(1, num_layer): 155 | layers.append(block(self.in_channels, init_channels)) 156 | 157 | return nn.Sequential(*layers) 158 | 159 | def _make_up_block(self, block, init_channels, num_layer, stride=1): 160 | upsample = None 161 | # expansion = block.expansion 162 | if stride != 1 or self.in_channels != init_channels * 2: 163 | upsample = nn.Sequential( 164 | nn.ConvTranspose2d(self.in_channels, init_channels * 2, 165 | kernel_size=1, stride=stride, 166 | bias=False, output_padding=1), 167 | nn.BatchNorm2d(init_channels * 2), 168 | ) 169 | layers = [] 170 | for i in range(1, num_layer): 171 | layers.append(block(self.in_channels, init_channels, 4)) 172 | layers.append( 173 | block(self.in_channels, init_channels, 2, stride, upsample)) 174 | self.in_channels = init_channels * 2 175 | return nn.Sequential(*layers) 176 | 177 | def encoder(self, x): 178 | x = self.conv1(x) 179 | x = self.bn1(x) 180 | x = self.relu(x) 181 | x = self.maxpool(x) 182 | 183 | x = self.layer1(x) 184 | x = self.layer2(x) 185 | return x 186 | 187 | def decoder(self, x, image_size): 188 | x = self.uplayer3(x) 189 | x = self.uplayer4(x) 190 | x = self.uplayer_top(x) 191 | 192 | x = self.conv1_1(x, output_size=image_size) 193 | return x 194 | 195 | def forward(self, x): 196 | img = x 197 | tmp1 = self.encoder(x) 198 | tmp2 = self.decoder(tmp1, img.size()) 199 | tmp3 = self.encoder(tmp2) 200 | 201 | return tmp1, tmp2, tmp3 202 | 203 | 204 | def ResNet50(**kwargs): 205 | return ResNet_autoencoder(Bottleneck, DeconvBottleneck, [3, 4, 6, 3], 3, **kwargs) 206 | 207 | 208 | def ResNet101(**kwargs): 209 | return ResNet_autoencoder(Bottleneck, [3, 4, 23, 2], 3, **kwargs) 210 | 211 | 212 | if __name__ == "__main__": 213 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 214 | 3, 4, 6, 3], 3).cuda() 215 | ''' 216 | load pre_trained_model 217 | ''' 218 | pretrained_dict = torch.load("./resnet50-19c8e357.pth") 219 | model_dict = model.state_dict() 220 | # 1. filter out unnecessary keys 221 | pretrained_dict = {k: v for k, 222 | v in pretrained_dict.items() if k in model_dict} 223 | # 2. overwrite entries in the existing state dict 224 | model_dict.update(pretrained_dict) 225 | model.load_state_dict(model_dict) 226 | 227 | input = torch.autograd.Variable(torch.randn(2, 3, 224, 224)).cuda() 228 | o = model(input) 229 | print(o) 230 | -------------------------------------------------------------------------------- /autoencoder2.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torch.nn.functional as F 4 | 5 | ''' 6 | batch_size 8 7 | epoch every 100 checkpoint 8 | learning_rate 1e-4? 9 | ''' 10 | 11 | 12 | class Bottleneck(nn.Module): 13 | expansion = 4 14 | 15 | def __init__(self, in_channels, out_channels, stride=1, downsample=None): 16 | super(Bottleneck, self).__init__() 17 | self.conv1 = nn.Conv2d(in_channels, out_channels, 18 | kernel_size=1, bias=False) 19 | self.bn1 = nn.BatchNorm2d(out_channels) 20 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 21 | stride=stride, bias=False, padding=1) 22 | self.bn2 = nn.BatchNorm2d(out_channels) 23 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 24 | kernel_size=1, bias=False) 25 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 26 | self.relu = nn.ReLU() 27 | self.downsample = downsample 28 | 29 | def forward(self, x): 30 | shortcut = x 31 | 32 | out = self.conv1(x) 33 | out = self.bn1(out) 34 | out = self.relu(out) 35 | 36 | out = self.conv2(out) 37 | out = self.bn2(out) 38 | out = self.relu(out) 39 | 40 | out = self.conv3(out) 41 | out = self.bn3(out) 42 | out = self.relu(out) 43 | 44 | if self.downsample is not None: 45 | shortcut = self.downsample(x) 46 | 47 | out += shortcut 48 | out = self.relu(out) 49 | 50 | return out 51 | 52 | 53 | class DeconvBottleneck(nn.Module): 54 | def __init__(self, in_channels, out_channels, expansion=2, stride=1, upsample=None): 55 | super(DeconvBottleneck, self).__init__() 56 | self.expansion = expansion 57 | self.conv1 = nn.Conv2d(in_channels, out_channels, 58 | kernel_size=1, bias=False) 59 | self.bn1 = nn.BatchNorm2d(out_channels) 60 | if stride == 1: 61 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 62 | stride=stride, bias=False, padding=1) 63 | else: 64 | self.conv2 = nn.ConvTranspose2d(out_channels, out_channels, 65 | kernel_size=3, 66 | stride=stride, bias=False, 67 | padding=1, 68 | output_padding=1) 69 | self.bn2 = nn.BatchNorm2d(out_channels) 70 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 71 | kernel_size=1, bias=False) 72 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 73 | self.relu = nn.ReLU() 74 | self.upsample = upsample 75 | 76 | def forward(self, x): 77 | shortcut = x 78 | 79 | out = self.conv1(x) 80 | out = self.bn1(out) 81 | out = self.relu(out) 82 | 83 | out = self.conv2(out) 84 | out = self.bn2(out) 85 | out = self.relu(out) 86 | 87 | out = self.conv3(out) 88 | out = self.bn3(out) 89 | out = self.relu(out) 90 | 91 | if self.upsample is not None: 92 | shortcut = self.upsample(x) 93 | 94 | out += shortcut 95 | out = self.relu(out) 96 | 97 | return out 98 | 99 | 100 | class ResNet_autoencoder(nn.Module): 101 | def __init__(self, downblock, upblock, num_layers, n_classes): 102 | super(ResNet_autoencoder, self).__init__() 103 | 104 | self.in_channels = 64 105 | 106 | self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, 107 | bias=False) 108 | self.bn1 = nn.BatchNorm2d(64) 109 | self.relu = nn.ReLU() 110 | self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) 111 | 112 | self.layer1 = self._make_downlayer(downblock, 64, num_layers[0]) 113 | self.layer2 = self._make_downlayer(downblock, 128, num_layers[1], 114 | stride=2) 115 | self.layer3 = self._make_downlayer(downblock, 256, num_layers[2], 116 | stride=2) 117 | self.layer4 = self._make_downlayer(downblock, 512, num_layers[3], 118 | stride=2) 119 | 120 | self.uplayer1 = self._make_up_block( 121 | upblock, 512, num_layers[3], stride=2) 122 | self.uplayer2 = self._make_up_block( 123 | upblock, 256, num_layers[2], stride=2) 124 | self.uplayer3 = self._make_up_block( 125 | upblock, 128, num_layers[1], stride=2) 126 | self.uplayer4 = self._make_up_block( 127 | upblock, 64, num_layers[0], stride=2) 128 | 129 | upsample = nn.Sequential( 130 | nn.ConvTranspose2d(self.in_channels, # 256 131 | 64, 132 | kernel_size=1, stride=2, 133 | bias=False, output_padding=1), 134 | nn.BatchNorm2d(64), 135 | ) 136 | self.uplayer_top = DeconvBottleneck( 137 | self.in_channels, 64, 1, 2, upsample) 138 | 139 | self.conv1_1 = nn.ConvTranspose2d(64, n_classes, kernel_size=1, stride=1, 140 | bias=False) 141 | 142 | def _make_downlayer(self, block, init_channels, num_layer, stride=1): 143 | downsample = None 144 | if stride != 1 or self.in_channels != init_channels * block.expansion: 145 | downsample = nn.Sequential( 146 | nn.Conv2d(self.in_channels, init_channels * block.expansion, 147 | kernel_size=1, stride=stride, bias=False), 148 | nn.BatchNorm2d(init_channels * block.expansion), 149 | ) 150 | layers = [] 151 | layers.append( 152 | block(self.in_channels, init_channels, stride, downsample)) 153 | self.in_channels = init_channels * block.expansion 154 | for i in range(1, num_layer): 155 | layers.append(block(self.in_channels, init_channels)) 156 | 157 | return nn.Sequential(*layers) 158 | 159 | def _make_up_block(self, block, init_channels, num_layer, stride=1): 160 | upsample = None 161 | # expansion = block.expansion 162 | if stride != 1 or self.in_channels != init_channels * 2: 163 | upsample = nn.Sequential( 164 | nn.ConvTranspose2d(self.in_channels, init_channels * 2, 165 | kernel_size=1, stride=stride, 166 | bias=False, output_padding=1), 167 | nn.BatchNorm2d(init_channels * 2), 168 | ) 169 | layers = [] 170 | for i in range(1, num_layer): 171 | layers.append(block(self.in_channels, init_channels, 4)) 172 | layers.append( 173 | block(self.in_channels, init_channels, 2, stride, upsample)) 174 | self.in_channels = init_channels * 2 175 | return nn.Sequential(*layers) 176 | 177 | def encoder(self, x): 178 | x = self.conv1(x) 179 | x = self.bn1(x) 180 | x = self.relu(x) 181 | x = self.maxpool(x) 182 | 183 | x = self.layer1(x) 184 | x = self.layer2(x) 185 | x = self.layer3(x) 186 | return x 187 | 188 | def decoder(self, x, image_size): 189 | x = self.uplayer2(x) 190 | x = self.uplayer3(x) 191 | x = self.uplayer4(x) 192 | x = self.uplayer_top(x) 193 | 194 | x = self.conv1_1(x, output_size=image_size) 195 | return x 196 | 197 | def forward(self, x): 198 | img = x 199 | tmp1 = self.encoder(x) 200 | tmp2 = self.decoder(tmp1, img.size()) 201 | tmp3 = self.encoder(tmp2) 202 | 203 | return tmp1, tmp2, tmp3 204 | 205 | 206 | def ResNet50(**kwargs): 207 | return ResNet_autoencoder(Bottleneck, DeconvBottleneck, [3, 4, 6, 3], 3, **kwargs) 208 | 209 | 210 | def ResNet101(**kwargs): 211 | return ResNet_autoencoder(Bottleneck, [3, 4, 23, 2], 3, **kwargs) 212 | 213 | 214 | if __name__ == "__main__": 215 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 216 | 3, 4, 6, 3], 3).cuda() 217 | ''' 218 | load pre_trained_model 219 | ''' 220 | pretrained_dict = torch.load("./resnet50-19c8e357.pth") 221 | model_dict = model.state_dict() 222 | # 1. filter out unnecessary keys 223 | pretrained_dict = {k: v for k, 224 | v in pretrained_dict.items() if k in model_dict} 225 | # 2. overwrite entries in the existing state dict 226 | model_dict.update(pretrained_dict) 227 | model.load_state_dict(model_dict) 228 | 229 | input = torch.autograd.Variable(torch.randn(2, 3, 224, 224)).cuda() 230 | o = model(input) 231 | print(o) 232 | -------------------------------------------------------------------------------- /autoencoder1.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch import nn 3 | import torch.nn.functional as F 4 | 5 | ''' 6 | batch_size 8 7 | epoch every 100 checkpoint 8 | learning_rate 1e-4? 9 | ''' 10 | 11 | 12 | class Bottleneck(nn.Module): 13 | expansion = 4 14 | 15 | def __init__(self, in_channels, out_channels, stride=1, downsample=None): 16 | super(Bottleneck, self).__init__() 17 | self.conv1 = nn.Conv2d(in_channels, out_channels, 18 | kernel_size=1, bias=False) 19 | self.bn1 = nn.BatchNorm2d(out_channels) 20 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 21 | stride=stride, bias=False, padding=1) 22 | self.bn2 = nn.BatchNorm2d(out_channels) 23 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 24 | kernel_size=1, bias=False) 25 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 26 | self.relu = nn.ReLU() 27 | self.downsample = downsample 28 | 29 | def forward(self, x): 30 | shortcut = x 31 | 32 | out = self.conv1(x) 33 | out = self.bn1(out) 34 | out = self.relu(out) 35 | 36 | out = self.conv2(out) 37 | out = self.bn2(out) 38 | out = self.relu(out) 39 | 40 | out = self.conv3(out) 41 | out = self.bn3(out) 42 | out = self.relu(out) 43 | 44 | if self.downsample is not None: 45 | shortcut = self.downsample(x) 46 | 47 | out += shortcut 48 | out = self.relu(out) 49 | 50 | return out 51 | 52 | 53 | class DeconvBottleneck(nn.Module): 54 | def __init__(self, in_channels, out_channels, expansion=2, stride=1, upsample=None): 55 | super(DeconvBottleneck, self).__init__() 56 | self.expansion = expansion 57 | self.conv1 = nn.Conv2d(in_channels, out_channels, 58 | kernel_size=1, bias=False) 59 | self.bn1 = nn.BatchNorm2d(out_channels) 60 | if stride == 1: 61 | self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, 62 | stride=stride, bias=False, padding=1) 63 | else: 64 | self.conv2 = nn.ConvTranspose2d(out_channels, out_channels, 65 | kernel_size=3, 66 | stride=stride, bias=False, 67 | padding=1, 68 | output_padding=1) 69 | self.bn2 = nn.BatchNorm2d(out_channels) 70 | self.conv3 = nn.Conv2d(out_channels, out_channels * self.expansion, 71 | kernel_size=1, bias=False) 72 | self.bn3 = nn.BatchNorm2d(out_channels * self.expansion) 73 | self.relu = nn.ReLU() 74 | self.upsample = upsample 75 | 76 | def forward(self, x): 77 | shortcut = x 78 | 79 | out = self.conv1(x) 80 | out = self.bn1(out) 81 | out = self.relu(out) 82 | 83 | out = self.conv2(out) 84 | out = self.bn2(out) 85 | out = self.relu(out) 86 | 87 | out = self.conv3(out) 88 | out = self.bn3(out) 89 | out = self.relu(out) 90 | 91 | if self.upsample is not None: 92 | shortcut = self.upsample(x) 93 | 94 | out += shortcut 95 | out = self.relu(out) 96 | 97 | return out 98 | 99 | 100 | class ResNet_autoencoder(nn.Module): 101 | def __init__(self, downblock, upblock, num_layers, n_classes): 102 | super(ResNet_autoencoder, self).__init__() 103 | 104 | self.in_channels = 64 105 | 106 | self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, 107 | bias=False) 108 | self.bn1 = nn.BatchNorm2d(64) 109 | self.relu = nn.ReLU() 110 | self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) 111 | 112 | self.layer1 = self._make_downlayer(downblock, 64, num_layers[0]) 113 | self.layer2 = self._make_downlayer(downblock, 128, num_layers[1], 114 | stride=2) 115 | self.layer3 = self._make_downlayer(downblock, 256, num_layers[2], 116 | stride=2) 117 | self.layer4 = self._make_downlayer(downblock, 512, num_layers[3], 118 | stride=2) 119 | 120 | self.uplayer1 = self._make_up_block( 121 | upblock, 512, num_layers[3], stride=2) 122 | self.uplayer2 = self._make_up_block( 123 | upblock, 256, num_layers[2], stride=2) 124 | self.uplayer3 = self._make_up_block( 125 | upblock, 128, num_layers[1], stride=2) 126 | self.uplayer4 = self._make_up_block( 127 | upblock, 64, num_layers[0], stride=2) 128 | 129 | upsample = nn.Sequential( 130 | nn.ConvTranspose2d(self.in_channels, # 256 131 | 64, 132 | kernel_size=1, stride=2, 133 | bias=False, output_padding=1), 134 | nn.BatchNorm2d(64), 135 | ) 136 | self.uplayer_top = DeconvBottleneck( 137 | self.in_channels, 64, 1, 2, upsample) 138 | 139 | self.conv1_1 = nn.ConvTranspose2d(64, n_classes, kernel_size=1, stride=1, 140 | bias=False) 141 | 142 | def _make_downlayer(self, block, init_channels, num_layer, stride=1): 143 | downsample = None 144 | if stride != 1 or self.in_channels != init_channels * block.expansion: 145 | downsample = nn.Sequential( 146 | nn.Conv2d(self.in_channels, init_channels * block.expansion, 147 | kernel_size=1, stride=stride, bias=False), 148 | nn.BatchNorm2d(init_channels * block.expansion), 149 | ) 150 | layers = [] 151 | layers.append( 152 | block(self.in_channels, init_channels, stride, downsample)) 153 | self.in_channels = init_channels * block.expansion 154 | for i in range(1, num_layer): 155 | layers.append(block(self.in_channels, init_channels)) 156 | 157 | return nn.Sequential(*layers) 158 | 159 | def _make_up_block(self, block, init_channels, num_layer, stride=1): 160 | upsample = None 161 | # expansion = block.expansion 162 | if stride != 1 or self.in_channels != init_channels * 2: 163 | upsample = nn.Sequential( 164 | nn.ConvTranspose2d(self.in_channels, init_channels * 2, 165 | kernel_size=1, stride=stride, 166 | bias=False, output_padding=1), 167 | nn.BatchNorm2d(init_channels * 2), 168 | ) 169 | layers = [] 170 | for i in range(1, num_layer): 171 | layers.append(block(self.in_channels, init_channels, 4)) 172 | layers.append( 173 | block(self.in_channels, init_channels, 2, stride, upsample)) 174 | self.in_channels = init_channels * 2 175 | return nn.Sequential(*layers) 176 | 177 | def encoder(self, x): 178 | x = self.conv1(x) 179 | x = self.bn1(x) 180 | x = self.relu(x) 181 | x = self.maxpool(x) 182 | 183 | x = self.layer1(x) 184 | x = self.layer2(x) 185 | x = self.layer3(x) 186 | x = self.layer4(x) 187 | return x 188 | 189 | def decoder(self, x, image_size): 190 | x = self.uplayer1(x) 191 | x = self.uplayer2(x) 192 | x = self.uplayer3(x) 193 | x = self.uplayer4(x) 194 | x = self.uplayer_top(x) 195 | 196 | x = self.conv1_1(x, output_size=image_size) 197 | return x 198 | 199 | def forward(self, x): 200 | img = x 201 | tmp1 = self.encoder(x) 202 | tmp2 = self.decoder(tmp1, img.size()) 203 | 204 | return tmp1, tmp2 205 | 206 | 207 | def ResNet50(**kwargs): 208 | return ResNet_autoencoder(Bottleneck, DeconvBottleneck, [3, 4, 6, 3], 3, **kwargs) 209 | 210 | 211 | def ResNet101(**kwargs): 212 | return ResNet_autoencoder(Bottleneck, [3, 4, 23, 2], 3, **kwargs) 213 | 214 | 215 | if __name__ == "__main__": 216 | model = ResNet_autoencoder(Bottleneck, DeconvBottleneck, [ 217 | 3, 4, 6, 3], 3).cuda() 218 | ''' 219 | load pre_trained_model 220 | ''' 221 | pretrained_dict = torch.load("./resnet50-19c8e357.pth") 222 | model_dict = model.state_dict() 223 | # 1. filter out unnecessary keys 224 | pretrained_dict = {k: v for k, 225 | v in pretrained_dict.items() if k in model_dict} 226 | # 2. overwrite entries in the existing state dict 227 | model_dict.update(pretrained_dict) 228 | model.load_state_dict(model_dict) 229 | 230 | input = torch.autograd.Variable(torch.randn(2, 3, 224, 224)).cuda() 231 | o = model(input) 232 | print(o) 233 | -------------------------------------------------------------------------------- /log3/nohup.out: -------------------------------------------------------------------------------- 1 | start loading. 2 | load data success. 3 | load pretrained model success 4 | start training. 5 | start loading. 6 | load data success. 7 | load pretrained model success 8 | start training. 9 | Epoch [1/10], Iter [10] Loss: 0.3295 10 | Epoch [1/10], Iter [20] Loss: 0.2387 11 | Epoch [1/10], Iter [30] Loss: 0.2184 12 | Epoch [1/10], Iter [40] Loss: 0.1648 13 | Epoch [1/10], Iter [50] Loss: 0.1026 14 | Epoch [1/10], Iter [60] Loss: 0.0950 15 | Epoch [1/10], Iter [70] Loss: 0.0889 16 | Epoch [1/10], Iter [80] Loss: 0.0710 17 | Epoch [1/10], Iter [90] Loss: 0.0546 18 | Epoch [1/10], Iter [100] Loss: 0.0598 19 | Epoch [1/10], Iter [110] Loss: 0.0469 20 | Epoch [1/10], Iter [120] Loss: 0.0548 21 | Epoch [1/10], Iter [130] Loss: 0.0339 22 | Epoch [1/10], Iter [140] Loss: 0.0300 23 | Epoch [1/10], Iter [150] Loss: 0.0318 24 | Epoch [1/10], Iter [160] Loss: 0.0339 25 | Epoch [1/10], Iter [170] Loss: 0.0201 26 | Epoch [1/10], Iter [180] Loss: 0.0230 27 | Epoch [1/10], Iter [190] Loss: 0.0191 28 | Epoch [1/10], Iter [200] Loss: 0.0216 29 | Epoch [1/10], Iter [210] Loss: 0.0147 30 | Epoch [1/10], Iter [220] Loss: 0.0137 31 | Epoch [1/10], Iter [230] Loss: 0.0215 32 | Epoch [1/10], Iter [240] Loss: 0.0165 33 | Epoch [1/10], Iter [250] Loss: 0.0141 34 | Epoch [1/10], Iter [260] Loss: 0.0117 35 | Epoch [1/10], Iter [270] Loss: 0.0139 36 | Epoch [1/10], Iter [280] Loss: 0.0236 37 | Epoch [1/10], Iter [290] Loss: 0.0187 38 | Epoch [1/10], Iter [300] Loss: 0.0122 39 | Epoch [1/10], Iter [310] Loss: 0.0150 40 | Epoch [1/10], Iter [320] Loss: 0.0131 41 | Epoch [1/10], Iter [330] Loss: 0.0135 42 | Epoch [1/10], Iter [340] Loss: 0.0107 43 | Epoch [1/10], Iter [350] Loss: 0.0106 44 | Epoch [1/10], Iter [360] Loss: 0.0110 45 | Epoch [1/10], Iter [370] Loss: 0.0101 46 | Epoch [1/10], Iter [380] Loss: 0.0098 47 | Epoch [1/10], Iter [390] Loss: 0.0174 48 | Epoch [1/10], Iter [400] Loss: 0.0080 49 | Epoch [1/10], Iter [410] Loss: 0.0122 50 | Epoch [1/10], Iter [420] Loss: 0.0105 51 | Epoch [1/10], Iter [430] Loss: 0.0104 52 | Epoch [1/10], Iter [440] Loss: 0.0099 53 | Epoch [1/10], Iter [450] Loss: 0.0109 54 | Epoch [1/10], Iter [460] Loss: 0.0102 55 | Epoch [1/10], Iter [470] Loss: 0.0110 56 | Epoch [1/10], Iter [480] Loss: 0.0102 57 | Epoch [1/10], Iter [490] Loss: 0.0088 58 | Epoch [1/10], Iter [500] Loss: 0.0123 59 | Epoch [1/10], Iter [510] Loss: 0.0094 60 | Epoch [1/10], Iter [520] Loss: 0.0090 61 | Epoch [1/10], Iter [530] Loss: 0.0076 62 | Epoch [1/10], Iter [540] Loss: 0.0115 63 | Epoch [1/10], Iter [550] Loss: 0.0117 64 | Epoch [1/10], Iter [560] Loss: 0.0071 65 | Epoch [1/10], Iter [570] Loss: 0.0092 66 | Epoch [1/10], Iter [580] Loss: 0.0117 67 | Epoch [1/10], Iter [590] Loss: 0.0096 68 | Epoch [1/10], Iter [600] Loss: 0.0103 69 | Epoch [1/10], Iter [610] Loss: 0.0069 70 | Epoch [1/10], Iter [620] Loss: 0.0069 71 | Epoch [1/10], Iter [630] Loss: 0.0093 72 | Epoch [1/10], Iter [640] Loss: 0.0073 73 | Epoch [1/10], Iter [650] Loss: 0.0091 74 | Epoch [1/10], Iter [660] Loss: 0.0091 75 | Epoch [1/10], Iter [670] Loss: 0.0078 76 | Epoch [1/10], Iter [680] Loss: 0.0090 77 | Epoch [1/10], Iter [690] Loss: 0.0084 78 | Epoch [1/10], Iter [700] Loss: 0.0090 79 | Epoch [1/10], Iter [710] Loss: 0.0090 80 | Epoch [1/10], Iter [720] Loss: 0.0068 81 | Epoch [1/10], Iter [730] Loss: 0.0070 82 | Epoch [1/10], Iter [740] Loss: 0.0086 83 | Epoch [1/10], Iter [750] Loss: 0.0063 84 | Epoch [1/10], Iter [760] Loss: 0.0076 85 | Epoch [1/10], Iter [770] Loss: 0.0078 86 | Epoch [1/10], Iter [780] Loss: 0.0075 87 | Epoch [1/10], Iter [790] Loss: 0.0064 88 | Epoch [1/10], Iter [800] Loss: 0.0100 89 | Epoch [1/10], Iter [810] Loss: 0.0071 90 | Epoch [1/10], Iter [820] Loss: 0.0076 91 | Epoch [1/10], Iter [830] Loss: 0.0067 92 | Epoch [1/10], Iter [840] Loss: 0.0061 93 | Epoch [1/10], Iter [850] Loss: 0.0073 94 | Epoch [1/10], Iter [860] Loss: 0.0069 95 | Epoch [1/10], Iter [870] Loss: 0.0081 96 | Epoch [1/10], Iter [880] Loss: 0.0096 97 | Epoch [1/10], Iter [890] Loss: 0.0059 98 | Epoch [1/10], Iter [900] Loss: 0.0087 99 | Epoch [1/10], Iter [910] Loss: 0.0062 100 | Epoch [1/10], Iter [920] Loss: 0.0060 101 | Epoch [1/10], Iter [930] Loss: 0.0082 102 | Epoch [1/10], Iter [940] Loss: 0.0068 103 | Epoch [1/10], Iter [950] Loss: 0.0063 104 | Epoch [1/10], Iter [960] Loss: 0.0072 105 | Epoch [1/10], Iter [970] Loss: 0.0055 106 | Epoch [1/10], Iter [980] Loss: 0.0078 107 | Epoch [1/10], Iter [990] Loss: 0.0066 108 | Epoch [1/10], Iter [1000] Loss: 0.0070 109 | Epoch [1/10], Iter [1010] Loss: 0.0097 110 | Epoch [1/10], Iter [1020] Loss: 0.0068 111 | Epoch [1/10], Iter [1030] Loss: 0.0058 112 | Epoch [1/10], Iter [1040] Loss: 0.0092 113 | Epoch [1/10], Iter [1050] Loss: 0.0054 114 | Epoch [1/10], Iter [1060] Loss: 0.0062 115 | Epoch [1/10], Iter [1070] Loss: 0.0073 116 | Epoch [1/10], Iter [1080] Loss: 0.0057 117 | Epoch [1/10], Iter [1090] Loss: 0.0066 118 | Epoch [1/10], Iter [1100] Loss: 0.0072 119 | Epoch [1/10], Iter [1110] Loss: 0.0067 120 | Epoch [1/10], Iter [1120] Loss: 0.0061 121 | Epoch [1/10], Iter [1130] Loss: 0.0081 122 | Epoch [1/10], Iter [1140] Loss: 0.0075 123 | Epoch [1/10], Iter [1150] Loss: 0.0078 124 | Epoch [1/10], Iter [1160] Loss: 0.0062 125 | Epoch [1/10], Iter [1170] Loss: 0.0059 126 | Epoch [1/10], Iter [1180] Loss: 0.0066 127 | Epoch [1/10], Iter [1190] Loss: 0.0073 128 | Epoch [1/10], Iter [1200] Loss: 0.0062 129 | Epoch [1/10], Iter [1210] Loss: 0.0054 130 | Epoch [1/10], Iter [1220] Loss: 0.0086 131 | Epoch [1/10], Iter [1230] Loss: 0.0056 132 | Epoch [1/10], Iter [1240] Loss: 0.0059 133 | Epoch [1/10], Iter [1250] Loss: 0.0064 134 | Epoch [1/10], Iter [1260] Loss: 0.0071 135 | Epoch [1/10], Iter [1270] Loss: 0.0082 136 | Epoch [1/10], Iter [1280] Loss: 0.0087 137 | Epoch [1/10], Iter [1290] Loss: 0.0060 138 | Epoch [1/10], Iter [1300] Loss: 0.0066 139 | Epoch [1/10], Iter [1310] Loss: 0.0059 140 | Epoch [1/10], Iter [1320] Loss: 0.0053 141 | Epoch [1/10], Iter [1330] Loss: 0.0058 142 | Epoch [1/10], Iter [1340] Loss: 0.0078 143 | Epoch [1/10], Iter [1350] Loss: 0.0056 144 | Epoch [1/10], Iter [1360] Loss: 0.0058 145 | Epoch [1/10], Iter [1370] Loss: 0.0057 146 | Epoch [1/10], Iter [1380] Loss: 0.0041 147 | Epoch [1/10], Iter [1390] Loss: 0.0052 148 | Epoch [1/10], Iter [1400] Loss: 0.0064 149 | Epoch [1/10], Iter [1410] Loss: 0.0050 150 | Epoch [1/10], Iter [1420] Loss: 0.0050 151 | Epoch [1/10], Iter [1430] Loss: 0.0068 152 | Epoch [1/10], Iter [1440] Loss: 0.0057 153 | Epoch [1/10], Iter [1450] Loss: 0.0067 154 | Epoch [1/10], Iter [1460] Loss: 0.0057 155 | Epoch [1/10], Iter [1470] Loss: 0.0080 156 | Epoch [1/10], Iter [1480] Loss: 0.0064 157 | Epoch [1/10], Iter [1490] Loss: 0.0070 158 | Epoch [1/10], Iter [1500] Loss: 0.0049 159 | Epoch [1/10], Iter [1510] Loss: 0.0060 160 | Epoch [1/10], Iter [1520] Loss: 0.0055 161 | Epoch [1/10], Iter [1530] Loss: 0.0057 162 | Epoch [1/10], Iter [1540] Loss: 0.0047 163 | Epoch [1/10], Iter [1550] Loss: 0.0074 164 | Epoch [1/10], Iter [1560] Loss: 0.0059 165 | Epoch [1/10], Iter [1570] Loss: 0.0061 166 | Epoch [1/10], Iter [1580] Loss: 0.0050 167 | Epoch [1/10], Iter [1590] Loss: 0.0058 168 | Epoch [1/10], Iter [1600] Loss: 0.0058 169 | Epoch [1/10], Iter [1610] Loss: 0.0055 170 | Epoch [1/10], Iter [1620] Loss: 0.0046 171 | Epoch [1/10], Iter [1630] Loss: 0.0060 172 | Epoch [1/10], Iter [1640] Loss: 0.0087 173 | Epoch [1/10], Iter [1650] Loss: 0.0060 174 | Epoch [1/10], Iter [1660] Loss: 0.0045 175 | Epoch [1/10], Iter [1670] Loss: 0.0058 176 | Epoch [1/10], Iter [1680] Loss: 0.0058 177 | Epoch [1/10], Iter [1690] Loss: 0.0080 178 | Epoch [1/10], Iter [1700] Loss: 0.0052 179 | Epoch [1/10], Iter [1710] Loss: 0.0041 180 | Epoch [1/10], Iter [1720] Loss: 0.0056 181 | Epoch [1/10], Iter [1730] Loss: 0.0055 182 | Epoch [1/10], Iter [1740] Loss: 0.0109 183 | Epoch [1/10], Iter [1750] Loss: 0.0058 184 | Epoch [1/10], Iter [1760] Loss: 0.0048 185 | Epoch [1/10], Iter [1770] Loss: 0.0081 186 | Epoch [1/10], Iter [1780] Loss: 0.0045 187 | Epoch [1/10], Iter [1790] Loss: 0.0065 188 | Epoch [1/10], Iter [1800] Loss: 0.0065 189 | Epoch [1/10], Iter [1810] Loss: 0.0058 190 | Epoch [1/10], Iter [1820] Loss: 0.0050 191 | Epoch [1/10], Iter [1830] Loss: 0.0071 192 | Epoch [1/10], Iter [1840] Loss: 0.0043 193 | Epoch [1/10], Iter [1850] Loss: 0.0060 194 | Epoch [1/10], Iter [1860] Loss: 0.0103 195 | Epoch [1/10], Iter [1870] Loss: 0.0055 196 | Epoch [1/10], Iter [1880] Loss: 0.0047 197 | Epoch [1/10], Iter [1890] Loss: 0.0060 198 | Epoch [1/10], Iter [1900] Loss: 0.0051 199 | Epoch [1/10], Iter [1910] Loss: 0.0055 200 | Epoch [1/10], Iter [1920] Loss: 0.0059 201 | Epoch [1/10], Iter [1930] Loss: 0.0055 202 | Epoch [1/10], Iter [1940] Loss: 0.0052 203 | Epoch [1/10], Iter [1950] Loss: 0.0043 204 | Epoch [1/10], Iter [1960] Loss: 0.0075 205 | Epoch [1/10], Iter [1970] Loss: 0.0045 206 | Epoch [1/10], Iter [1980] Loss: 0.0085 207 | Epoch [1/10], Iter [1990] Loss: 0.0049 208 | Epoch [1/10], Iter [2000] Loss: 0.0039 209 | Epoch [1/10], Iter [2010] Loss: 0.0066 210 | Epoch [1/10], Iter [2020] Loss: 0.0054 211 | Epoch [1/10], Iter [2030] Loss: 0.0043 212 | Epoch [1/10], Iter [2040] Loss: 0.0051 213 | Epoch [1/10], Iter [2050] Loss: 0.0074 214 | Epoch [1/10], Iter [2060] Loss: 0.0073 215 | Epoch [1/10], Iter [2070] Loss: 0.0053 216 | Epoch [1/10], Iter [2080] Loss: 0.0070 217 | Epoch [1/10], Iter [2090] Loss: 0.0098 218 | Epoch [1/10], Iter [2100] Loss: 0.0051 219 | Epoch [1/10], Iter [2110] Loss: 0.0059 220 | Epoch [1/10], Iter [2120] Loss: 0.0043 221 | Epoch [1/10], Iter [2130] Loss: 0.0036 222 | Epoch [1/10], Iter [2140] Loss: 0.0044 223 | Epoch [1/10], Iter [2150] Loss: 0.0071 224 | Epoch [1/10], Iter [2160] Loss: 0.0044 225 | Epoch [1/10], Iter [2170] Loss: 0.0051 226 | Epoch [1/10], Iter [2180] Loss: 0.0052 227 | Epoch [1/10], Iter [2190] Loss: 0.0062 228 | Epoch [1/10], Iter [2200] Loss: 0.0040 229 | Epoch [1/10], Iter [2210] Loss: 0.0049 230 | Epoch [1/10], Iter [2220] Loss: 0.0039 231 | Epoch [1/10], Iter [2230] Loss: 0.0049 232 | Epoch [1/10], Iter [2240] Loss: 0.0052 233 | Epoch [1/10], Iter [2250] Loss: 0.0073 234 | Epoch [1/10], Iter [2260] Loss: 0.0055 235 | Epoch [1/10], Iter [2270] Loss: 0.0047 236 | Epoch [1/10], Iter [2280] Loss: 0.0041 237 | Epoch [1/10], Iter [2290] Loss: 0.0061 238 | Epoch [1/10], Iter [2300] Loss: 0.0041 239 | Epoch [1/10], Iter [2310] Loss: 0.0035 240 | Epoch [1/10], Iter [2320] Loss: 0.0038 241 | Epoch [1/10], Iter [2330] Loss: 0.0056 242 | Epoch [1/10], Iter [2340] Loss: 0.0050 243 | Epoch [1/10], Iter [2350] Loss: 0.0037 244 | Epoch [1/10], Iter [2360] Loss: 0.0057 245 | Epoch [1/10], Iter [2370] Loss: 0.0047 246 | Epoch [1/10], Iter [2380] Loss: 0.0053 247 | Epoch [1/10], Iter [2390] Loss: 0.0040 248 | Epoch [1/10], Iter [2400] Loss: 0.0043 249 | Epoch [1/10], Iter [2410] Loss: 0.0068 250 | Epoch [1/10], Iter [2420] Loss: 0.0045 251 | Epoch [1/10], Iter [2430] Loss: 0.0044 252 | Epoch [1/10], Iter [2440] Loss: 0.0041 253 | Epoch [1/10], Iter [2450] Loss: 0.0033 254 | Epoch [1/10], Iter [2460] Loss: 0.0033 255 | Epoch [1/10], Iter [2470] Loss: 0.0064 256 | Epoch [1/10], Iter [2480] Loss: 0.0043 257 | Epoch [1/10], Iter [2490] Loss: 0.0051 258 | Epoch [1/10], Iter [2500] Loss: 0.0042 259 | Epoch [1/10], Iter [2510] Loss: 0.0042 260 | Epoch [1/10], Iter [2520] Loss: 0.0049 261 | Epoch [1/10], Iter [2530] Loss: 0.0067 262 | Epoch [1/10], Iter [2540] Loss: 0.0044 263 | Epoch [1/10], Iter [2550] Loss: 0.0049 264 | Epoch [1/10], Iter [2560] Loss: 0.0041 265 | Epoch [1/10], Iter [2570] Loss: 0.0041 266 | Epoch [1/10], Iter [2580] Loss: 0.0079 267 | Epoch [1/10], Iter [2590] Loss: 0.0045 268 | Epoch [1/10], Iter [2600] Loss: 0.0048 269 | Epoch [1/10], Iter [2610] Loss: 0.0050 270 | Epoch [1/10], Iter [2620] Loss: 0.0051 271 | Epoch [1/10], Iter [2630] Loss: 0.0043 272 | Epoch [1/10], Iter [2640] Loss: 0.0051 273 | Epoch [1/10], Iter [2650] Loss: 0.0048 274 | Epoch [1/10], Iter [2660] Loss: 0.0048 275 | Epoch [1/10], Iter [2670] Loss: 0.0043 276 | Epoch [1/10], Iter [2680] Loss: 0.0032 277 | Epoch [1/10], Iter [2690] Loss: 0.0043 278 | Epoch [1/10], Iter [2700] Loss: 0.0039 279 | Epoch [1/10], Iter [2710] Loss: 0.0038 280 | Epoch [1/10], Iter [2720] Loss: 0.0044 281 | Epoch [1/10], Iter [2730] Loss: 0.0044 282 | Epoch [1/10], Iter [2740] Loss: 0.0071 283 | Epoch [1/10], Iter [2750] Loss: 0.0039 284 | Epoch [1/10], Iter [2760] Loss: 0.0047 285 | Epoch [1/10], Iter [2770] Loss: 0.0043 286 | Epoch [1/10], Iter [2780] Loss: 0.0056 287 | Epoch [1/10], Iter [2790] Loss: 0.0034 288 | Epoch [1/10], Iter [2800] Loss: 0.0050 289 | Epoch [1/10], Iter [2810] Loss: 0.0047 290 | Epoch [1/10], Iter [2820] Loss: 0.0059 291 | Epoch [1/10], Iter [2830] Loss: 0.0050 292 | Epoch [1/10], Iter [2840] Loss: 0.0050 293 | Epoch [1/10], Iter [2850] Loss: 0.0093 294 | Epoch [1/10], Iter [2860] Loss: 0.0037 295 | Epoch [1/10], Iter [2870] Loss: 0.0042 296 | Epoch [1/10], Iter [2880] Loss: 0.0054 297 | Epoch [1/10], Iter [2890] Loss: 0.0077 298 | Epoch [1/10], Iter [2900] Loss: 0.0046 299 | Epoch [1/10], Iter [2910] Loss: 0.0039 300 | Epoch [1/10], Iter [2920] Loss: 0.0035 301 | Epoch [1/10], Iter [2930] Loss: 0.0039 302 | Epoch [1/10], Iter [2940] Loss: 0.0035 303 | Epoch [1/10], Iter [2950] Loss: 0.0044 304 | Epoch [1/10], Iter [2960] Loss: 0.0047 305 | Epoch [1/10], Iter [2970] Loss: 0.0042 306 | Epoch [1/10], Iter [2980] Loss: 0.0045 307 | Epoch [1/10], Iter [2990] Loss: 0.0055 308 | Epoch [1/10], Iter [3000] Loss: 0.0046 309 | Epoch [1/10], Iter [3010] Loss: 0.0060 310 | Epoch [1/10], Iter [3020] Loss: 0.0040 311 | Epoch [1/10], Iter [3030] Loss: 0.0041 312 | Epoch [1/10], Iter [3040] Loss: 0.0047 313 | Epoch [1/10], Iter [3050] Loss: 0.0050 314 | Epoch [1/10], Iter [3060] Loss: 0.0046 315 | Epoch [1/10], Iter [3070] Loss: 0.0062 316 | Epoch [1/10], Iter [3080] Loss: 0.0034 317 | Epoch [1/10], Iter [3090] Loss: 0.0055 318 | Epoch [1/10], Iter [3100] Loss: 0.0042 319 | Epoch [1/10], Iter [3110] Loss: 0.0045 320 | Epoch [1/10], Iter [3120] Loss: 0.0038 321 | Epoch [1/10], Iter [3130] Loss: 0.0042 322 | Epoch [1/10], Iter [3140] Loss: 0.0032 323 | Epoch [1/10], Iter [3150] Loss: 0.0044 324 | Epoch [1/10], Iter [3160] Loss: 0.0035 325 | Epoch [1/10], Iter [3170] Loss: 0.0041 326 | Epoch [1/10], Iter [3180] Loss: 0.0040 327 | Epoch [1/10], Iter [3190] Loss: 0.0044 328 | Epoch [1/10], Iter [3200] Loss: 0.0036 329 | Epoch [1/10], Iter [3210] Loss: 0.0036 330 | Epoch [1/10], Iter [3220] Loss: 0.0044 331 | Epoch [1/10], Iter [3230] Loss: 0.0066 332 | Epoch [1/10], Iter [3240] Loss: 0.0032 333 | Epoch [1/10], Iter [3250] Loss: 0.0032 334 | Epoch [1/10], Iter [3260] Loss: 0.0047 335 | Epoch [1/10], Iter [3270] Loss: 0.0034 336 | Epoch [1/10], Iter [3280] Loss: 0.0039 337 | Epoch [1/10], Iter [3290] Loss: 0.0031 338 | Epoch [1/10], Iter [3300] Loss: 0.0040 339 | Epoch [1/10], Iter [3310] Loss: 0.0036 340 | Epoch [1/10], Iter [3320] Loss: 0.0043 341 | Epoch [1/10], Iter [3330] Loss: 0.0040 342 | Epoch [1/10], Iter [3340] Loss: 0.0042 343 | Epoch [1/10], Iter [3350] Loss: 0.0036 344 | Epoch [1/10], Iter [3360] Loss: 0.0034 345 | Epoch [1/10], Iter [3370] Loss: 0.0035 346 | Epoch [1/10], Iter [3380] Loss: 0.0040 347 | Epoch [1/10], Iter [3390] Loss: 0.0030 348 | Epoch [1/10], Iter [3400] Loss: 0.0034 349 | Epoch [1/10], Iter [3410] Loss: 0.0035 350 | Epoch [1/10], Iter [3420] Loss: 0.0047 351 | Epoch [1/10], Iter [3430] Loss: 0.0047 352 | Epoch [1/10], Iter [3440] Loss: 0.0043 353 | Epoch [1/10], Iter [3450] Loss: 0.0048 354 | Epoch [1/10], Iter [3460] Loss: 0.0035 355 | Epoch [1/10], Iter [3470] Loss: 0.0040 356 | Epoch [1/10], Iter [3480] Loss: 0.0043 357 | Epoch [1/10], Iter [3490] Loss: 0.0038 358 | Epoch [1/10], Iter [3500] Loss: 0.0047 359 | Epoch [1/10], Iter [3510] Loss: 0.0032 360 | Epoch [1/10], Iter [3520] Loss: 0.0036 361 | Epoch [1/10], Iter [3530] Loss: 0.0061 362 | Epoch [1/10], Iter [3540] Loss: 0.0038 363 | Epoch [1/10], Iter [3550] Loss: 0.0038 364 | Epoch [1/10], Iter [3560] Loss: 0.0041 365 | Epoch [1/10], Iter [3570] Loss: 0.0043 366 | Epoch [1/10], Iter [3580] Loss: 0.0030 367 | Epoch [1/10], Iter [3590] Loss: 0.0033 368 | Epoch [1/10], Iter [3600] Loss: 0.0042 369 | Epoch [1/10], Iter [3610] Loss: 0.0033 370 | Epoch [1/10], Iter [3620] Loss: 0.0037 371 | Epoch [1/10], Iter [3630] Loss: 0.0061 372 | Epoch [1/10], Iter [3640] Loss: 0.0052 373 | Epoch [1/10], Iter [3650] Loss: 0.0058 374 | Epoch [1/10], Iter [3660] Loss: 0.0047 375 | Epoch [1/10], Iter [3670] Loss: 0.0042 376 | Epoch [1/10], Iter [3680] Loss: 0.0036 377 | Epoch [1/10], Iter [3690] Loss: 0.0054 378 | Epoch [1/10], Iter [3700] Loss: 0.0073 379 | Epoch [1/10], Iter [3710] Loss: 0.0085 380 | Epoch [1/10], Iter [3720] Loss: 0.0034 381 | Epoch [1/10], Iter [3730] Loss: 0.0039 382 | Epoch [1/10], Iter [3740] Loss: 0.0048 383 | Epoch [1/10], Iter [3750] Loss: 0.0030 384 | Epoch [1/10], Iter [3760] Loss: 0.0043 385 | Epoch [1/10], Iter [3770] Loss: 0.0035 386 | Epoch [1/10], Iter [3780] Loss: 0.0047 387 | Epoch [1/10], Iter [3790] Loss: 0.0034 388 | Epoch [1/10], Iter [3800] Loss: 0.0041 389 | Epoch [1/10], Iter [3810] Loss: 0.0036 390 | Epoch [1/10], Iter [3820] Loss: 0.0040 391 | Epoch [1/10], Iter [3830] Loss: 0.0038 392 | Epoch [1/10], Iter [3840] Loss: 0.0033 393 | Epoch [1/10], Iter [3850] Loss: 0.0049 394 | Epoch [1/10], Iter [3860] Loss: 0.0031 395 | Epoch [1/10], Iter [3870] Loss: 0.0035 396 | Epoch [1/10], Iter [3880] Loss: 0.0066 397 | Epoch [1/10], Iter [3890] Loss: 0.0035 398 | Epoch [1/10], Iter [3900] Loss: 0.0040 399 | Epoch [1/10], Iter [3910] Loss: 0.0053 400 | Epoch [1/10], Iter [3920] Loss: 0.0062 401 | Epoch [1/10], Iter [3930] Loss: 0.0037 402 | Epoch [1/10], Iter [3940] Loss: 0.0050 403 | Epoch [1/10], Iter [3950] Loss: 0.0034 404 | Epoch [1/10], Iter [3960] Loss: 0.0034 405 | Epoch [1/10], Iter [3970] Loss: 0.0032 406 | Epoch [1/10], Iter [3980] Loss: 0.0053 407 | Epoch [1/10], Iter [3990] Loss: 0.0038 408 | Epoch [1/10], Iter [4000] Loss: 0.0045 409 | Epoch [1/10], Iter [4010] Loss: 0.0049 410 | Epoch [1/10], Iter [4020] Loss: 0.0051 411 | Epoch [1/10], Iter [4030] Loss: 0.0049 412 | Epoch [1/10], Iter [4040] Loss: 0.0036 413 | Epoch [1/10], Iter [4050] Loss: 0.0038 414 | Epoch [1/10], Iter [4060] Loss: 0.0041 415 | Epoch [1/10], Iter [4070] Loss: 0.0063 416 | Epoch [1/10], Iter [4080] Loss: 0.0034 417 | Epoch [1/10], Iter [4090] Loss: 0.0051 418 | Epoch [1/10], Iter [4100] Loss: 0.0035 419 | Epoch [1/10], Iter [4110] Loss: 0.0035 420 | Epoch [1/10], Iter [4120] Loss: 0.0041 421 | Epoch [1/10], Iter [4130] Loss: 0.0031 422 | Epoch [1/10], Iter [4140] Loss: 0.0041 423 | Epoch [1/10], Iter [4150] Loss: 0.0049 424 | Epoch [1/10], Iter [4160] Loss: 0.0043 425 | Epoch [1/10], Iter [4170] Loss: 0.0040 426 | Epoch [1/10], Iter [4180] Loss: 0.0035 427 | Epoch [1/10], Iter [4190] Loss: 0.0041 428 | Epoch [1/10], Iter [4200] Loss: 0.0044 429 | Epoch [1/10], Iter [4210] Loss: 0.0041 430 | Epoch [1/10], Iter [4220] Loss: 0.0035 431 | Epoch [1/10], Iter [4230] Loss: 0.0046 432 | Epoch [1/10], Iter [4240] Loss: 0.0035 433 | Epoch [1/10], Iter [4250] Loss: 0.0042 434 | Epoch [1/10], Iter [4260] Loss: 0.0043 435 | Epoch [1/10], Iter [4270] Loss: 0.0038 436 | Epoch [1/10], Iter [4280] Loss: 0.0041 437 | Epoch [1/10], Iter [4290] Loss: 0.0031 438 | Epoch [1/10], Iter [4300] Loss: 0.0046 439 | Epoch [1/10], Iter [4310] Loss: 0.0045 440 | Epoch [1/10], Iter [4320] Loss: 0.0030 441 | Epoch [1/10], Iter [4330] Loss: 0.0031 442 | Epoch [1/10], Iter [4340] Loss: 0.0034 443 | Epoch [1/10], Iter [4350] Loss: 0.0030 444 | Epoch [1/10], Iter [4360] Loss: 0.0071 445 | Epoch [1/10], Iter [4370] Loss: 0.0038 446 | Epoch [1/10], Iter [4380] Loss: 0.0047 447 | Epoch [1/10], Iter [4390] Loss: 0.0041 448 | Epoch [1/10], Iter [4400] Loss: 0.0048 449 | Epoch [1/10], Iter [4410] Loss: 0.0037 450 | Epoch [1/10], Iter [4420] Loss: 0.0029 451 | Epoch [1/10], Iter [4430] Loss: 0.0033 452 | Epoch [1/10], Iter [4440] Loss: 0.0045 453 | Epoch [1/10], Iter [4450] Loss: 0.0044 454 | Epoch [1/10], Iter [4460] Loss: 0.0058 455 | Epoch [1/10], Iter [4470] Loss: 0.0042 456 | Epoch [1/10], Iter [4480] Loss: 0.0042 457 | Epoch [1/10], Iter [4490] Loss: 0.0032 458 | Epoch [1/10], Iter [4500] Loss: 0.0060 459 | Epoch [1/10], Iter [4510] Loss: 0.0032 460 | Epoch [1/10], Iter [4520] Loss: 0.0036 461 | Epoch [1/10], Iter [4530] Loss: 0.0030 462 | Epoch [1/10], Iter [4540] Loss: 0.0055 463 | Epoch [1/10], Iter [4550] Loss: 0.0040 464 | Epoch [1/10], Iter [4560] Loss: 0.0037 465 | Epoch [1/10], Iter [4570] Loss: 0.0028 466 | Epoch [1/10], Iter [4580] Loss: 0.0037 467 | Epoch [1/10], Iter [4590] Loss: 0.0038 468 | Epoch [1/10], Iter [4600] Loss: 0.0061 469 | Epoch [1/10], Iter [4610] Loss: 0.0037 470 | Epoch [1/10], Iter [4620] Loss: 0.0044 471 | Epoch [1/10], Iter [4630] Loss: 0.0045 472 | Epoch [1/10], Iter [4640] Loss: 0.0030 473 | Epoch [1/10], Iter [4650] Loss: 0.0038 474 | Epoch [1/10], Iter [4660] Loss: 0.0039 475 | Epoch [1/10], Iter [4670] Loss: 0.0044 476 | Epoch [1/10], Iter [4680] Loss: 0.0038 477 | Epoch [1/10], Iter [4690] Loss: 0.0040 478 | Epoch [1/10], Iter [4700] Loss: 0.0027 479 | Epoch [1/10], Iter [4710] Loss: 0.0034 480 | Epoch [1/10], Iter [4720] Loss: 0.0036 481 | Epoch [1/10], Iter [4730] Loss: 0.0040 482 | Epoch [1/10], Iter [4740] Loss: 0.0048 483 | Epoch [1/10], Iter [4750] Loss: 0.0030 484 | Epoch [1/10], Iter [4760] Loss: 0.0041 485 | Epoch [1/10], Iter [4770] Loss: 0.0031 486 | Epoch [1/10], Iter [4780] Loss: 0.0032 487 | Epoch [1/10], Iter [4790] Loss: 0.0043 488 | Epoch [1/10], Iter [4800] Loss: 0.0055 489 | Epoch [1/10], Iter [4810] Loss: 0.0032 490 | Epoch [1/10], Iter [4820] Loss: 0.0055 491 | Epoch [1/10], Iter [4830] Loss: 0.0031 492 | Epoch [1/10], Iter [4840] Loss: 0.0036 493 | Epoch [1/10], Iter [4850] Loss: 0.0048 494 | Epoch [1/10], Iter [4860] Loss: 0.0034 495 | Epoch [1/10], Iter [4870] Loss: 0.0041 496 | Epoch [1/10], Iter [4880] Loss: 0.0036 497 | Epoch [1/10], Iter [4890] Loss: 0.0045 498 | Epoch [1/10], Iter [4900] Loss: 0.0035 499 | Epoch [1/10], Iter [4910] Loss: 0.0049 500 | Epoch [1/10], Iter [4920] Loss: 0.0037 501 | Epoch [1/10], Iter [4930] Loss: 0.0037 502 | Epoch [1/10], Iter [4940] Loss: 0.0037 503 | Epoch [1/10], Iter [4950] Loss: 0.0071 504 | Epoch [1/10], Iter [4960] Loss: 0.0029 505 | Epoch [1/10], Iter [4970] Loss: 0.0043 506 | Epoch [1/10], Iter [4980] Loss: 0.0031 507 | Epoch [1/10], Iter [4990] Loss: 0.0039 508 | Epoch [1/10], Iter [5000] Loss: 0.0034 509 | Epoch [1/10], Iter [5010] Loss: 0.0032 510 | Epoch [1/10], Iter [5020] Loss: 0.0031 511 | Epoch [1/10], Iter [5030] Loss: 0.0039 512 | Epoch [1/10], Iter [5040] Loss: 0.0032 513 | Epoch [1/10], Iter [5050] Loss: 0.0040 514 | Epoch [1/10], Iter [5060] Loss: 0.0039 515 | Epoch [1/10], Iter [5070] Loss: 0.0032 516 | Epoch [1/10], Iter [5080] Loss: 0.0042 517 | Epoch [1/10], Iter [5090] Loss: 0.0049 518 | Epoch [1/10], Iter [5100] Loss: 0.0031 519 | Epoch [1/10], Iter [5110] Loss: 0.0036 520 | Epoch [1/10], Iter [5120] Loss: 0.0031 521 | Epoch [1/10], Iter [5130] Loss: 0.0034 522 | Epoch [1/10], Iter [5140] Loss: 0.0035 523 | Epoch [1/10], Iter [5150] Loss: 0.0043 524 | Epoch [1/10], Iter [5160] Loss: 0.0047 525 | Epoch [1/10], Iter [5170] Loss: 0.0042 526 | Epoch [1/10], Iter [5180] Loss: 0.0037 527 | Epoch [1/10], Iter [5190] Loss: 0.0041 528 | Epoch [1/10], Iter [5200] Loss: 0.0029 529 | Epoch [1/10], Iter [5210] Loss: 0.0034 530 | Epoch [1/10], Iter [5220] Loss: 0.0038 531 | Epoch [1/10], Iter [5230] Loss: 0.0031 532 | Epoch [1/10], Iter [5240] Loss: 0.0034 533 | Epoch [1/10], Iter [5250] Loss: 0.0034 534 | Epoch [1/10], Iter [5260] Loss: 0.0047 535 | Epoch [1/10], Iter [5270] Loss: 0.0037 536 | Epoch [1/10], Iter [5280] Loss: 0.0037 537 | Epoch [1/10], Iter [5290] Loss: 0.0035 538 | Epoch [1/10], Iter [5300] Loss: 0.0038 539 | Epoch [1/10], Iter [5310] Loss: 0.0038 540 | Epoch [1/10], Iter [5320] Loss: 0.0046 541 | Epoch [1/10], Iter [5330] Loss: 0.0032 542 | Epoch [1/10], Iter [5340] Loss: 0.0031 543 | Epoch [1/10], Iter [5350] Loss: 0.0046 544 | Epoch [1/10], Iter [5360] Loss: 0.0036 545 | Epoch [1/10], Iter [5370] Loss: 0.0048 546 | Epoch [1/10], Iter [5380] Loss: 0.0035 547 | Epoch [1/10], Iter [5390] Loss: 0.0031 548 | Epoch [1/10], Iter [5400] Loss: 0.0042 549 | Epoch [1/10], Iter [5410] Loss: 0.0042 550 | Epoch [1/10], Iter [5420] Loss: 0.0032 551 | Epoch [1/10], Iter [5430] Loss: 0.0032 552 | Epoch [1/10], Iter [5440] Loss: 0.0027 553 | Epoch [1/10], Iter [5450] Loss: 0.0031 554 | Epoch [1/10], Iter [5460] Loss: 0.0036 555 | Epoch [1/10], Iter [5470] Loss: 0.0038 556 | Epoch [1/10], Iter [5480] Loss: 0.0034 557 | Epoch [1/10], Iter [5490] Loss: 0.0044 558 | Epoch [1/10], Iter [5500] Loss: 0.0029 559 | Epoch [1/10], Iter [5510] Loss: 0.0032 560 | Epoch [1/10], Iter [5520] Loss: 0.0031 561 | Epoch [1/10], Iter [5530] Loss: 0.0028 562 | Epoch [1/10], Iter [5540] Loss: 0.0046 563 | Epoch [1/10], Iter [5550] Loss: 0.0034 564 | Epoch [1/10], Iter [5560] Loss: 0.0033 565 | Epoch [1/10], Iter [5570] Loss: 0.0040 566 | Epoch [1/10], Iter [5580] Loss: 0.0057 567 | Epoch [1/10], Iter [5590] Loss: 0.0032 568 | Epoch [1/10], Iter [5600] Loss: 0.0036 569 | Epoch [1/10], Iter [5610] Loss: 0.0039 570 | Epoch [1/10], Iter [5620] Loss: 0.0031 571 | Epoch [1/10], Iter [5630] Loss: 0.0028 572 | Epoch [1/10], Iter [5640] Loss: 0.0035 573 | Epoch [1/10], Iter [5650] Loss: 0.0029 574 | Epoch [1/10], Iter [5660] Loss: 0.0043 575 | Epoch [1/10], Iter [5670] Loss: 0.0032 576 | Epoch [1/10], Iter [5680] Loss: 0.0035 577 | Epoch [1/10], Iter [5690] Loss: 0.0033 578 | Epoch [1/10], Iter [5700] Loss: 0.0043 579 | Epoch [1/10], Iter [5710] Loss: 0.0033 580 | Epoch [1/10], Iter [5720] Loss: 0.0036 581 | Epoch [1/10], Iter [5730] Loss: 0.0034 582 | Epoch [1/10], Iter [5740] Loss: 0.0026 583 | Epoch [1/10], Iter [5750] Loss: 0.0043 584 | Epoch [1/10], Iter [5760] Loss: 0.0044 585 | Epoch [1/10], Iter [5770] Loss: 0.0029 586 | Epoch [1/10], Iter [5780] Loss: 0.0036 587 | Epoch [1/10], Iter [5790] Loss: 0.0034 588 | Epoch [1/10], Iter [5800] Loss: 0.0029 589 | Epoch [1/10], Iter [5810] Loss: 0.0033 590 | Epoch [1/10], Iter [5820] Loss: 0.0039 591 | Epoch [1/10], Iter [5830] Loss: 0.0033 592 | Epoch [1/10], Iter [5840] Loss: 0.0034 593 | Epoch [1/10], Iter [5850] Loss: 0.0029 594 | Epoch [1/10], Iter [5860] Loss: 0.0028 595 | Epoch [1/10], Iter [5870] Loss: 0.0028 596 | Epoch [1/10], Iter [5880] Loss: 0.0027 597 | Epoch [1/10], Iter [5890] Loss: 0.0026 598 | Epoch [1/10], Iter [5900] Loss: 0.0034 599 | Epoch [1/10], Iter [5910] Loss: 0.0030 600 | Epoch [1/10], Iter [5920] Loss: 0.0037 601 | Epoch [1/10], Iter [5930] Loss: 0.0030 602 | Epoch [1/10], Iter [5940] Loss: 0.0038 603 | Epoch [1/10], Iter [5950] Loss: 0.0026 604 | Epoch [1/10], Iter [5960] Loss: 0.0039 605 | Epoch [1/10], Iter [5970] Loss: 0.0031 606 | Epoch [1/10], Iter [5980] Loss: 0.0032 607 | Epoch [1/10], Iter [5990] Loss: 0.0030 608 | Epoch [1/10], Iter [6000] Loss: 0.0035 609 | Epoch [1/10], Iter [6010] Loss: 0.0037 610 | Epoch [1/10], Iter [6020] Loss: 0.0032 611 | Epoch [1/10], Iter [6030] Loss: 0.0028 612 | Epoch [1/10], Iter [6040] Loss: 0.0044 613 | Epoch [1/10], Iter [6050] Loss: 0.0036 614 | Epoch [1/10], Iter [6060] Loss: 0.0033 615 | Epoch [1/10], Iter [6070] Loss: 0.0045 616 | Epoch [1/10], Iter [6080] Loss: 0.0081 617 | Epoch [1/10], Iter [6090] Loss: 0.0035 618 | Epoch [1/10], Iter [6100] Loss: 0.0034 619 | Epoch [1/10], Iter [6110] Loss: 0.0049 620 | Epoch [1/10], Iter [6120] Loss: 0.0063 621 | Epoch [1/10], Iter [6130] Loss: 0.0035 622 | Epoch [1/10], Iter [6140] Loss: 0.0030 623 | Epoch [1/10], Iter [6150] Loss: 0.0031 624 | Epoch [1/10], Iter [6160] Loss: 0.0037 625 | Epoch [1/10], Iter [6170] Loss: 0.0031 626 | Epoch [1/10], Iter [6180] Loss: 0.0033 627 | Epoch [1/10], Iter [6190] Loss: 0.0058 628 | Epoch [1/10], Iter [6200] Loss: 0.0033 629 | Epoch [1/10], Iter [6210] Loss: 0.0040 630 | Epoch [1/10], Iter [6220] Loss: 0.0033 631 | Epoch [1/10], Iter [6230] Loss: 0.0037 632 | Epoch [1/10], Iter [6240] Loss: 0.0037 633 | Epoch [1/10], Iter [6250] Loss: 0.0045 634 | Epoch [1/10], Iter [6260] Loss: 0.0033 635 | Epoch [1/10], Iter [6270] Loss: 0.0044 636 | Epoch [1/10], Iter [6280] Loss: 0.0039 637 | Epoch [1/10], Iter [6290] Loss: 0.0044 638 | Epoch [1/10], Iter [6300] Loss: 0.0038 639 | Epoch [1/10], Iter [6310] Loss: 0.0032 640 | Epoch [1/10], Iter [6320] Loss: 0.0037 641 | Epoch [1/10], Iter [6330] Loss: 0.0037 642 | Epoch [1/10], Iter [6340] Loss: 0.0021 643 | Epoch [1/10], Iter [6350] Loss: 0.0034 644 | Epoch [1/10], Iter [6360] Loss: 0.0037 645 | Epoch [1/10], Iter [6370] Loss: 0.0031 646 | Epoch [1/10], Iter [6380] Loss: 0.0027 647 | Epoch [1/10], Iter [6390] Loss: 0.0031 648 | Epoch [1/10], Iter [6400] Loss: 0.0029 649 | Epoch [1/10], Iter [6410] Loss: 0.0031 650 | Epoch [1/10], Iter [6420] Loss: 0.0033 651 | Epoch [1/10], Iter [6430] Loss: 0.0041 652 | Epoch [1/10], Iter [6440] Loss: 0.0044 653 | Epoch [1/10], Iter [6450] Loss: 0.0036 654 | Epoch [1/10], Iter [6460] Loss: 0.0033 655 | Epoch [1/10], Iter [6470] Loss: 0.0039 656 | Epoch [1/10], Iter [6480] Loss: 0.0044 657 | Epoch [1/10], Iter [6490] Loss: 0.0029 658 | Epoch [1/10], Iter [6500] Loss: 0.0033 659 | Epoch [1/10], Iter [6510] Loss: 0.0030 660 | Epoch [1/10], Iter [6520] Loss: 0.0040 661 | Epoch [1/10], Iter [6530] Loss: 0.0035 662 | Epoch [1/10], Iter [6540] Loss: 0.0027 663 | Epoch [1/10], Iter [6550] Loss: 0.0025 664 | Epoch [1/10], Iter [6560] Loss: 0.0032 665 | Epoch [1/10], Iter [6570] Loss: 0.0026 666 | Epoch [1/10], Iter [6580] Loss: 0.0027 667 | Epoch [1/10], Iter [6590] Loss: 0.0034 668 | Epoch [1/10], Iter [6600] Loss: 0.0039 669 | Epoch [1/10], Iter [6610] Loss: 0.0042 670 | Epoch [1/10], Iter [6620] Loss: 0.0031 671 | Epoch [1/10], Iter [6630] Loss: 0.0040 672 | Epoch [1/10], Iter [6640] Loss: 0.0029 673 | Epoch [1/10], Iter [6650] Loss: 0.0028 674 | Epoch [1/10], Iter [6660] Loss: 0.0028 675 | Epoch [1/10], Iter [6670] Loss: 0.0035 676 | Epoch [1/10], Iter [6680] Loss: 0.0025 677 | Epoch [1/10], Iter [6690] Loss: 0.0054 678 | Epoch [1/10], Iter [6700] Loss: 0.0035 679 | Epoch [1/10], Iter [6710] Loss: 0.0035 680 | Epoch [1/10], Iter [6720] Loss: 0.0035 681 | Epoch [1/10], Iter [6730] Loss: 0.0030 682 | Epoch [1/10], Iter [6740] Loss: 0.0037 683 | Epoch [1/10], Iter [6750] Loss: 0.0024 684 | Epoch [1/10], Iter [6760] Loss: 0.0034 685 | Epoch [1/10], Iter [6770] Loss: 0.0032 686 | Epoch [1/10], Iter [6780] Loss: 0.0025 687 | Epoch [1/10], Iter [6790] Loss: 0.0038 688 | Epoch [1/10], Iter [6800] Loss: 0.0030 689 | Epoch [1/10], Iter [6810] Loss: 0.0037 690 | Epoch [1/10], Iter [6820] Loss: 0.0025 691 | Epoch [1/10], Iter [6830] Loss: 0.0026 692 | Epoch [1/10], Iter [6840] Loss: 0.0048 693 | Epoch [1/10], Iter [6850] Loss: 0.0037 694 | Epoch [1/10], Iter [6860] Loss: 0.0034 695 | Epoch [1/10], Iter [6870] Loss: 0.0040 696 | Epoch [1/10], Iter [6880] Loss: 0.0034 697 | Epoch [1/10], Iter [6890] Loss: 0.0033 698 | Epoch [1/10], Iter [6900] Loss: 0.0039 699 | Epoch [1/10], Iter [6910] Loss: 0.0027 700 | Epoch [1/10], Iter [6920] Loss: 0.0029 701 | Epoch [1/10], Iter [6930] Loss: 0.0041 702 | Epoch [1/10], Iter [6940] Loss: 0.0033 703 | Epoch [1/10], Iter [6950] Loss: 0.0035 704 | Epoch [1/10], Iter [6960] Loss: 0.0030 705 | Epoch [1/10], Iter [6970] Loss: 0.0026 706 | Epoch [1/10], Iter [6980] Loss: 0.0031 707 | Epoch [1/10], Iter [6990] Loss: 0.0029 708 | Epoch [1/10], Iter [7000] Loss: 0.0036 709 | Epoch [1/10], Iter [7010] Loss: 0.0030 710 | Epoch [1/10], Iter [7020] Loss: 0.0038 711 | Epoch [1/10], Iter [7030] Loss: 0.0034 712 | Epoch [1/10], Iter [7040] Loss: 0.0042 713 | Epoch [1/10], Iter [7050] Loss: 0.0027 714 | Epoch [1/10], Iter [7060] Loss: 0.0034 715 | Epoch [1/10], Iter [7070] Loss: 0.0039 716 | Epoch [1/10], Iter [7080] Loss: 0.0028 717 | Epoch [1/10], Iter [7090] Loss: 0.0032 718 | Epoch [1/10], Iter [7100] Loss: 0.0034 719 | Epoch [1/10], Iter [7110] Loss: 0.0036 720 | Epoch [1/10], Iter [7120] Loss: 0.0035 721 | Epoch [1/10], Iter [7130] Loss: 0.0030 722 | Epoch [1/10], Iter [7140] Loss: 0.0029 723 | Epoch [1/10], Iter [7150] Loss: 0.0039 724 | Epoch [1/10], Iter [7160] Loss: 0.0034 725 | Epoch [1/10], Iter [7170] Loss: 0.0033 726 | Epoch [1/10], Iter [7180] Loss: 0.0046 727 | Epoch [1/10], Iter [7190] Loss: 0.0037 728 | Epoch [1/10], Iter [7200] Loss: 0.0034 729 | Epoch [1/10], Iter [7210] Loss: 0.0030 730 | Epoch [1/10], Iter [7220] Loss: 0.0021 731 | Epoch [1/10], Iter [7230] Loss: 0.0051 732 | Epoch [1/10], Iter [7240] Loss: 0.0024 733 | Epoch [1/10], Iter [7250] Loss: 0.0026 734 | Epoch [1/10], Iter [7260] Loss: 0.0026 735 | Epoch [1/10], Iter [7270] Loss: 0.0035 736 | Epoch [1/10], Iter [7280] Loss: 0.0061 737 | Epoch [1/10], Iter [7290] Loss: 0.0031 738 | Epoch [1/10], Iter [7300] Loss: 0.0033 739 | Epoch [1/10], Iter [7310] Loss: 0.0030 740 | Epoch [1/10], Iter [7320] Loss: 0.0033 741 | Epoch [1/10], Iter [7330] Loss: 0.0029 742 | Epoch [1/10], Iter [7340] Loss: 0.0036 743 | Epoch [1/10], Iter [7350] Loss: 0.0032 744 | Epoch [1/10], Iter [7360] Loss: 0.0030 745 | Epoch [1/10], Iter [7370] Loss: 0.0040 746 | Epoch [1/10], Iter [7380] Loss: 0.0025 747 | Epoch [1/10], Iter [7390] Loss: 0.0030 748 | Epoch [2/10], Iter [10] Loss: 0.0035 749 | Epoch [2/10], Iter [20] Loss: 0.0026 750 | Epoch [2/10], Iter [30] Loss: 0.0025 751 | Epoch [2/10], Iter [40] Loss: 0.0031 752 | Epoch [2/10], Iter [50] Loss: 0.0043 753 | Epoch [2/10], Iter [60] Loss: 0.0030 754 | Epoch [2/10], Iter [70] Loss: 0.0032 755 | Epoch [2/10], Iter [80] Loss: 0.0027 756 | Epoch [2/10], Iter [90] Loss: 0.0033 757 | Epoch [2/10], Iter [100] Loss: 0.0032 758 | Epoch [2/10], Iter [110] Loss: 0.0030 759 | Epoch [2/10], Iter [120] Loss: 0.0039 760 | Epoch [2/10], Iter [130] Loss: 0.0030 761 | Epoch [2/10], Iter [140] Loss: 0.0038 762 | Epoch [2/10], Iter [150] Loss: 0.0030 763 | Epoch [2/10], Iter [160] Loss: 0.0029 764 | Epoch [2/10], Iter [170] Loss: 0.0035 765 | Epoch [2/10], Iter [180] Loss: 0.0040 766 | Epoch [2/10], Iter [190] Loss: 0.0034 767 | Epoch [2/10], Iter [200] Loss: 0.0038 768 | Epoch [2/10], Iter [210] Loss: 0.0024 769 | Epoch [2/10], Iter [220] Loss: 0.0021 770 | Epoch [2/10], Iter [230] Loss: 0.0043 771 | Epoch [2/10], Iter [240] Loss: 0.0027 772 | Epoch [2/10], Iter [250] Loss: 0.0026 773 | Epoch [2/10], Iter [260] Loss: 0.0022 774 | Epoch [2/10], Iter [270] Loss: 0.0028 775 | Epoch [2/10], Iter [280] Loss: 0.0061 776 | Epoch [2/10], Iter [290] Loss: 0.0043 777 | Epoch [2/10], Iter [300] Loss: 0.0032 778 | Epoch [2/10], Iter [310] Loss: 0.0034 779 | Epoch [2/10], Iter [320] Loss: 0.0028 780 | Epoch [2/10], Iter [330] Loss: 0.0031 781 | Epoch [2/10], Iter [340] Loss: 0.0026 782 | Epoch [2/10], Iter [350] Loss: 0.0031 783 | Epoch [2/10], Iter [360] Loss: 0.0033 784 | Epoch [2/10], Iter [370] Loss: 0.0026 785 | Epoch [2/10], Iter [380] Loss: 0.0030 786 | Epoch [2/10], Iter [390] Loss: 0.0037 787 | Epoch [2/10], Iter [400] Loss: 0.0020 788 | Epoch [2/10], Iter [410] Loss: 0.0030 789 | Epoch [2/10], Iter [420] Loss: 0.0031 790 | Epoch [2/10], Iter [430] Loss: 0.0030 791 | Epoch [2/10], Iter [440] Loss: 0.0023 792 | Epoch [2/10], Iter [450] Loss: 0.0028 793 | Epoch [2/10], Iter [460] Loss: 0.0027 794 | Epoch [2/10], Iter [470] Loss: 0.0030 795 | Epoch [2/10], Iter [480] Loss: 0.0027 796 | Epoch [2/10], Iter [490] Loss: 0.0027 797 | Epoch [2/10], Iter [500] Loss: 0.0036 798 | Epoch [2/10], Iter [510] Loss: 0.0033 799 | Epoch [2/10], Iter [520] Loss: 0.0033 800 | Epoch [2/10], Iter [530] Loss: 0.0031 801 | Epoch [2/10], Iter [540] Loss: 0.0040 802 | Epoch [2/10], Iter [550] Loss: 0.0028 803 | Epoch [2/10], Iter [560] Loss: 0.0024 804 | Epoch [2/10], Iter [570] Loss: 0.0030 805 | Epoch [2/10], Iter [580] Loss: 0.0028 806 | Epoch [2/10], Iter [590] Loss: 0.0031 807 | Epoch [2/10], Iter [600] Loss: 0.0042 808 | Epoch [2/10], Iter [610] Loss: 0.0026 809 | Epoch [2/10], Iter [620] Loss: 0.0027 810 | Epoch [2/10], Iter [630] Loss: 0.0023 811 | Epoch [2/10], Iter [640] Loss: 0.0027 812 | Epoch [2/10], Iter [650] Loss: 0.0027 813 | Epoch [2/10], Iter [660] Loss: 0.0031 814 | Epoch [2/10], Iter [670] Loss: 0.0026 815 | Epoch [2/10], Iter [680] Loss: 0.0035 816 | Epoch [2/10], Iter [690] Loss: 0.0032 817 | Epoch [2/10], Iter [700] Loss: 0.0029 818 | Epoch [2/10], Iter [710] Loss: 0.0035 819 | Epoch [2/10], Iter [720] Loss: 0.0031 820 | Epoch [2/10], Iter [730] Loss: 0.0028 821 | Epoch [2/10], Iter [740] Loss: 0.0028 822 | Epoch [2/10], Iter [750] Loss: 0.0025 823 | Epoch [2/10], Iter [760] Loss: 0.0028 824 | Epoch [2/10], Iter [770] Loss: 0.0028 825 | Epoch [2/10], Iter [780] Loss: 0.0036 826 | Epoch [2/10], Iter [790] Loss: 0.0040 827 | Epoch [2/10], Iter [800] Loss: 0.0030 828 | Epoch [2/10], Iter [810] Loss: 0.0026 829 | Epoch [2/10], Iter [820] Loss: 0.0030 830 | Epoch [2/10], Iter [830] Loss: 0.0028 831 | Epoch [2/10], Iter [840] Loss: 0.0024 832 | Epoch [2/10], Iter [850] Loss: 0.0033 833 | Epoch [2/10], Iter [860] Loss: 0.0029 834 | Epoch [2/10], Iter [870] Loss: 0.0039 835 | Epoch [2/10], Iter [880] Loss: 0.0043 836 | Epoch [2/10], Iter [890] Loss: 0.0025 837 | Epoch [2/10], Iter [900] Loss: 0.0031 838 | Epoch [2/10], Iter [910] Loss: 0.0026 839 | Epoch [2/10], Iter [920] Loss: 0.0028 840 | Epoch [2/10], Iter [930] Loss: 0.0028 841 | Epoch [2/10], Iter [940] Loss: 0.0025 842 | Epoch [2/10], Iter [950] Loss: 0.0032 843 | Epoch [2/10], Iter [960] Loss: 0.0032 844 | Epoch [2/10], Iter [970] Loss: 0.0023 845 | Epoch [2/10], Iter [980] Loss: 0.0036 846 | Epoch [2/10], Iter [990] Loss: 0.0029 847 | Epoch [2/10], Iter [1000] Loss: 0.0032 848 | Epoch [2/10], Iter [1010] Loss: 0.0031 849 | Epoch [2/10], Iter [1020] Loss: 0.0030 850 | Epoch [2/10], Iter [1030] Loss: 0.0024 851 | Epoch [2/10], Iter [1040] Loss: 0.0026 852 | Epoch [2/10], Iter [1050] Loss: 0.0026 853 | Epoch [2/10], Iter [1060] Loss: 0.0027 854 | Epoch [2/10], Iter [1070] Loss: 0.0027 855 | Epoch [2/10], Iter [1080] Loss: 0.0028 856 | Epoch [2/10], Iter [1090] Loss: 0.0031 857 | Epoch [2/10], Iter [1100] Loss: 0.0034 858 | Epoch [2/10], Iter [1110] Loss: 0.0025 859 | Epoch [2/10], Iter [1120] Loss: 0.0028 860 | Epoch [2/10], Iter [1130] Loss: 0.0034 861 | Epoch [2/10], Iter [1140] Loss: 0.0029 862 | Epoch [2/10], Iter [1150] Loss: 0.0038 863 | Epoch [2/10], Iter [1160] Loss: 0.0025 864 | Epoch [2/10], Iter [1170] Loss: 0.0030 865 | Epoch [2/10], Iter [1180] Loss: 0.0043 866 | Epoch [2/10], Iter [1190] Loss: 0.0031 867 | Epoch [2/10], Iter [1200] Loss: 0.0031 868 | Epoch [2/10], Iter [1210] Loss: 0.0026 869 | Epoch [2/10], Iter [1220] Loss: 0.0048 870 | Epoch [2/10], Iter [1230] Loss: 0.0026 871 | Epoch [2/10], Iter [1240] Loss: 0.0034 872 | Epoch [2/10], Iter [1250] Loss: 0.0040 873 | Epoch [2/10], Iter [1260] Loss: 0.0033 874 | Epoch [2/10], Iter [1270] Loss: 0.0041 875 | Epoch [2/10], Iter [1280] Loss: 0.0037 876 | Epoch [2/10], Iter [1290] Loss: 0.0033 877 | Epoch [2/10], Iter [1300] Loss: 0.0034 878 | Epoch [2/10], Iter [1310] Loss: 0.0029 879 | Epoch [2/10], Iter [1320] Loss: 0.0026 880 | Epoch [2/10], Iter [1330] Loss: 0.0034 881 | Epoch [2/10], Iter [1340] Loss: 0.0046 882 | Epoch [2/10], Iter [1350] Loss: 0.0027 883 | Epoch [2/10], Iter [1360] Loss: 0.0022 884 | Epoch [2/10], Iter [1370] Loss: 0.0031 885 | Epoch [2/10], Iter [1380] Loss: 0.0022 886 | Epoch [2/10], Iter [1390] Loss: 0.0026 887 | Epoch [2/10], Iter [1400] Loss: 0.0033 888 | Epoch [2/10], Iter [1410] Loss: 0.0023 889 | Epoch [2/10], Iter [1420] Loss: 0.0025 890 | Epoch [2/10], Iter [1430] Loss: 0.0033 891 | Epoch [2/10], Iter [1440] Loss: 0.0027 892 | Epoch [2/10], Iter [1450] Loss: 0.0028 893 | Epoch [2/10], Iter [1460] Loss: 0.0032 894 | Epoch [2/10], Iter [1470] Loss: 0.0041 895 | Epoch [2/10], Iter [1480] Loss: 0.0026 896 | Epoch [2/10], Iter [1490] Loss: 0.0038 897 | Epoch [2/10], Iter [1500] Loss: 0.0024 898 | Epoch [2/10], Iter [1510] Loss: 0.0035 899 | Epoch [2/10], Iter [1520] Loss: 0.0029 900 | Epoch [2/10], Iter [1530] Loss: 0.0032 901 | Epoch [2/10], Iter [1540] Loss: 0.0025 902 | Epoch [2/10], Iter [1550] Loss: 0.0035 903 | Epoch [2/10], Iter [1560] Loss: 0.0027 904 | -------------------------------------------------------------------------------- /log2/nohup.out: -------------------------------------------------------------------------------- 1 | Traceback (most recent call last): 2 | File "train2.py", line 4, in 3 | from coco import load_dataset 4 | File "/home/achhe_ucdavis_edu/resnet-autoencoder/coco.py", line 1, in 5 | import torchvision.datasets 6 | ImportError: No module named torchvision.datasets 7 | start loading. 8 | load data success. 9 | load pretrained model success 10 | start training. 11 | Epoch [1/10], Iter [10] Loss: 0.2541 12 | Epoch [1/10], Iter [20] Loss: 0.2024 13 | Epoch [1/10], Iter [30] Loss: 0.2040 14 | Epoch [1/10], Iter [40] Loss: 0.1579 15 | Epoch [1/10], Iter [50] Loss: 0.1132 16 | Epoch [1/10], Iter [60] Loss: 0.1141 17 | Epoch [1/10], Iter [70] Loss: 0.1136 18 | Epoch [1/10], Iter [80] Loss: 0.1005 19 | Epoch [1/10], Iter [90] Loss: 0.0974 20 | Epoch [1/10], Iter [100] Loss: 0.0835 21 | Epoch [1/10], Iter [110] Loss: 0.0751 22 | Epoch [1/10], Iter [120] Loss: 0.0719 23 | Epoch [1/10], Iter [130] Loss: 0.0545 24 | Epoch [1/10], Iter [140] Loss: 0.0588 25 | Epoch [1/10], Iter [150] Loss: 0.0529 26 | Epoch [1/10], Iter [160] Loss: 0.0535 27 | Epoch [1/10], Iter [170] Loss: 0.0476 28 | Epoch [1/10], Iter [180] Loss: 0.0494 29 | Epoch [1/10], Iter [190] Loss: 0.0449 30 | Epoch [1/10], Iter [200] Loss: 0.0443 31 | Epoch [1/10], Iter [210] Loss: 0.0356 32 | Epoch [1/10], Iter [220] Loss: 0.0358 33 | Epoch [1/10], Iter [230] Loss: 0.0484 34 | Epoch [1/10], Iter [240] Loss: 0.0395 35 | Epoch [1/10], Iter [250] Loss: 0.0345 36 | Epoch [1/10], Iter [260] Loss: 0.0322 37 | Epoch [1/10], Iter [270] Loss: 0.0329 38 | Epoch [1/10], Iter [280] Loss: 0.0431 39 | Epoch [1/10], Iter [290] Loss: 0.0416 40 | Epoch [1/10], Iter [300] Loss: 0.0294 41 | Epoch [1/10], Iter [310] Loss: 0.0359 42 | Epoch [1/10], Iter [320] Loss: 0.0320 43 | Epoch [1/10], Iter [330] Loss: 0.0302 44 | Epoch [1/10], Iter [340] Loss: 0.0254 45 | Epoch [1/10], Iter [350] Loss: 0.0255 46 | Epoch [1/10], Iter [360] Loss: 0.0261 47 | Epoch [1/10], Iter [370] Loss: 0.0250 48 | Epoch [1/10], Iter [380] Loss: 0.0243 49 | Epoch [1/10], Iter [390] Loss: 0.0313 50 | Epoch [1/10], Iter [400] Loss: 0.0213 51 | Epoch [1/10], Iter [410] Loss: 0.0277 52 | Epoch [1/10], Iter [420] Loss: 0.0238 53 | Epoch [1/10], Iter [430] Loss: 0.0242 54 | Epoch [1/10], Iter [440] Loss: 0.0234 55 | Epoch [1/10], Iter [450] Loss: 0.0230 56 | Epoch [1/10], Iter [460] Loss: 0.0229 57 | Epoch [1/10], Iter [470] Loss: 0.0228 58 | Epoch [1/10], Iter [480] Loss: 0.0227 59 | Epoch [1/10], Iter [490] Loss: 0.0199 60 | Epoch [1/10], Iter [500] Loss: 0.0222 61 | Epoch [1/10], Iter [510] Loss: 0.0203 62 | Epoch [1/10], Iter [520] Loss: 0.0200 63 | Epoch [1/10], Iter [530] Loss: 0.0165 64 | Epoch [1/10], Iter [540] Loss: 0.0242 65 | Epoch [1/10], Iter [550] Loss: 0.0196 66 | Epoch [1/10], Iter [560] Loss: 0.0153 67 | Epoch [1/10], Iter [570] Loss: 0.0183 68 | Epoch [1/10], Iter [580] Loss: 0.0210 69 | Epoch [1/10], Iter [590] Loss: 0.0197 70 | Epoch [1/10], Iter [600] Loss: 0.0218 71 | Epoch [1/10], Iter [610] Loss: 0.0150 72 | Epoch [1/10], Iter [620] Loss: 0.0146 73 | Epoch [1/10], Iter [630] Loss: 0.0166 74 | Epoch [1/10], Iter [640] Loss: 0.0163 75 | Epoch [1/10], Iter [650] Loss: 0.0181 76 | Epoch [1/10], Iter [660] Loss: 0.0187 77 | Epoch [1/10], Iter [670] Loss: 0.0176 78 | Epoch [1/10], Iter [680] Loss: 0.0193 79 | Epoch [1/10], Iter [690] Loss: 0.0184 80 | Epoch [1/10], Iter [700] Loss: 0.0176 81 | Epoch [1/10], Iter [710] Loss: 0.0172 82 | Epoch [1/10], Iter [720] Loss: 0.0146 83 | Epoch [1/10], Iter [730] Loss: 0.0145 84 | Epoch [1/10], Iter [740] Loss: 0.0171 85 | Epoch [1/10], Iter [750] Loss: 0.0135 86 | Epoch [1/10], Iter [760] Loss: 0.0159 87 | Epoch [1/10], Iter [770] Loss: 0.0153 88 | Epoch [1/10], Iter [780] Loss: 0.0159 89 | Epoch [1/10], Iter [790] Loss: 0.0130 90 | Epoch [1/10], Iter [800] Loss: 0.0169 91 | Epoch [1/10], Iter [810] Loss: 0.0139 92 | Epoch [1/10], Iter [820] Loss: 0.0152 93 | Epoch [1/10], Iter [830] Loss: 0.0142 94 | Epoch [1/10], Iter [840] Loss: 0.0121 95 | Epoch [1/10], Iter [850] Loss: 0.0138 96 | Epoch [1/10], Iter [860] Loss: 0.0137 97 | Epoch [1/10], Iter [870] Loss: 0.0159 98 | Epoch [1/10], Iter [880] Loss: 0.0177 99 | Epoch [1/10], Iter [890] Loss: 0.0122 100 | Epoch [1/10], Iter [900] Loss: 0.0156 101 | Epoch [1/10], Iter [910] Loss: 0.0135 102 | Epoch [1/10], Iter [920] Loss: 0.0124 103 | Epoch [1/10], Iter [930] Loss: 0.0150 104 | Epoch [1/10], Iter [940] Loss: 0.0134 105 | Epoch [1/10], Iter [950] Loss: 0.0133 106 | Epoch [1/10], Iter [960] Loss: 0.0148 107 | Epoch [1/10], Iter [970] Loss: 0.0119 108 | Epoch [1/10], Iter [980] Loss: 0.0165 109 | Epoch [1/10], Iter [990] Loss: 0.0131 110 | Epoch [1/10], Iter [1000] Loss: 0.0137 111 | Epoch [1/10], Iter [1010] Loss: 0.0165 112 | Epoch [1/10], Iter [1020] Loss: 0.0135 113 | Epoch [1/10], Iter [1030] Loss: 0.0122 114 | Epoch [1/10], Iter [1040] Loss: 0.0163 115 | Epoch [1/10], Iter [1050] Loss: 0.0104 116 | Epoch [1/10], Iter [1060] Loss: 0.0113 117 | Epoch [1/10], Iter [1070] Loss: 0.0127 118 | Epoch [1/10], Iter [1080] Loss: 0.0112 119 | Epoch [1/10], Iter [1090] Loss: 0.0129 120 | Epoch [1/10], Iter [1100] Loss: 0.0131 121 | Epoch [1/10], Iter [1110] Loss: 0.0127 122 | Epoch [1/10], Iter [1120] Loss: 0.0119 123 | Epoch [1/10], Iter [1130] Loss: 0.0157 124 | Epoch [1/10], Iter [1140] Loss: 0.0128 125 | Epoch [1/10], Iter [1150] Loss: 0.0145 126 | Epoch [1/10], Iter [1160] Loss: 0.0115 127 | Epoch [1/10], Iter [1170] Loss: 0.0114 128 | Epoch [1/10], Iter [1180] Loss: 0.0130 129 | Epoch [1/10], Iter [1190] Loss: 0.0142 130 | Epoch [1/10], Iter [1200] Loss: 0.0117 131 | Epoch [1/10], Iter [1210] Loss: 0.0103 132 | Epoch [1/10], Iter [1220] Loss: 0.0148 133 | Epoch [1/10], Iter [1230] Loss: 0.0098 134 | Epoch [1/10], Iter [1240] Loss: 0.0118 135 | Epoch [1/10], Iter [1250] Loss: 0.0115 136 | Epoch [1/10], Iter [1260] Loss: 0.0119 137 | Epoch [1/10], Iter [1270] Loss: 0.0134 138 | Epoch [1/10], Iter [1280] Loss: 0.0144 139 | Epoch [1/10], Iter [1290] Loss: 0.0112 140 | Epoch [1/10], Iter [1300] Loss: 0.0119 141 | Epoch [1/10], Iter [1310] Loss: 0.0101 142 | Epoch [1/10], Iter [1320] Loss: 0.0097 143 | Epoch [1/10], Iter [1330] Loss: 0.0106 144 | Epoch [1/10], Iter [1340] Loss: 0.0131 145 | Epoch [1/10], Iter [1350] Loss: 0.0103 146 | Epoch [1/10], Iter [1360] Loss: 0.0103 147 | Epoch [1/10], Iter [1370] Loss: 0.0102 148 | Epoch [1/10], Iter [1380] Loss: 0.0077 149 | Epoch [1/10], Iter [1390] Loss: 0.0099 150 | Epoch [1/10], Iter [1400] Loss: 0.0112 151 | Epoch [1/10], Iter [1410] Loss: 0.0096 152 | Epoch [1/10], Iter [1420] Loss: 0.0099 153 | Epoch [1/10], Iter [1430] Loss: 0.0132 154 | Epoch [1/10], Iter [1440] Loss: 0.0102 155 | Epoch [1/10], Iter [1450] Loss: 0.0103 156 | Epoch [1/10], Iter [1460] Loss: 0.0100 157 | Epoch [1/10], Iter [1470] Loss: 0.0138 158 | Epoch [1/10], Iter [1480] Loss: 0.0107 159 | Epoch [1/10], Iter [1490] Loss: 0.0126 160 | Epoch [1/10], Iter [1500] Loss: 0.0090 161 | Epoch [1/10], Iter [1510] Loss: 0.0108 162 | Epoch [1/10], Iter [1520] Loss: 0.0101 163 | Epoch [1/10], Iter [1530] Loss: 0.0100 164 | Epoch [1/10], Iter [1540] Loss: 0.0084 165 | Epoch [1/10], Iter [1550] Loss: 0.0135 166 | Epoch [1/10], Iter [1560] Loss: 0.0102 167 | Epoch [1/10], Iter [1570] Loss: 0.0108 168 | Epoch [1/10], Iter [1580] Loss: 0.0092 169 | Epoch [1/10], Iter [1590] Loss: 0.0100 170 | Epoch [1/10], Iter [1600] Loss: 0.0108 171 | Epoch [1/10], Iter [1610] Loss: 0.0104 172 | Epoch [1/10], Iter [1620] Loss: 0.0079 173 | Epoch [1/10], Iter [1630] Loss: 0.0098 174 | Epoch [1/10], Iter [1640] Loss: 0.0139 175 | Epoch [1/10], Iter [1650] Loss: 0.0105 176 | Epoch [1/10], Iter [1660] Loss: 0.0075 177 | Epoch [1/10], Iter [1670] Loss: 0.0104 178 | Epoch [1/10], Iter [1680] Loss: 0.0091 179 | Epoch [1/10], Iter [1690] Loss: 0.0123 180 | Epoch [1/10], Iter [1700] Loss: 0.0091 181 | Epoch [1/10], Iter [1710] Loss: 0.0081 182 | Epoch [1/10], Iter [1720] Loss: 0.0102 183 | Epoch [1/10], Iter [1730] Loss: 0.0095 184 | Epoch [1/10], Iter [1740] Loss: 0.0154 185 | Epoch [1/10], Iter [1750] Loss: 0.0101 186 | Epoch [1/10], Iter [1760] Loss: 0.0081 187 | Epoch [1/10], Iter [1770] Loss: 0.0124 188 | Epoch [1/10], Iter [1780] Loss: 0.0077 189 | Epoch [1/10], Iter [1790] Loss: 0.0108 190 | Epoch [1/10], Iter [1800] Loss: 0.0107 191 | Epoch [1/10], Iter [1810] Loss: 0.0095 192 | Epoch [1/10], Iter [1820] Loss: 0.0097 193 | Epoch [1/10], Iter [1830] Loss: 0.0116 194 | Epoch [1/10], Iter [1840] Loss: 0.0080 195 | Epoch [1/10], Iter [1850] Loss: 0.0105 196 | Epoch [1/10], Iter [1860] Loss: 0.0142 197 | Epoch [1/10], Iter [1870] Loss: 0.0104 198 | Epoch [1/10], Iter [1880] Loss: 0.0103 199 | Epoch [1/10], Iter [1890] Loss: 0.0097 200 | Epoch [1/10], Iter [1900] Loss: 0.0095 201 | Epoch [1/10], Iter [1910] Loss: 0.0096 202 | Epoch [1/10], Iter [1920] Loss: 0.0098 203 | Epoch [1/10], Iter [1930] Loss: 0.0100 204 | Epoch [1/10], Iter [1940] Loss: 0.0084 205 | Epoch [1/10], Iter [1950] Loss: 0.0081 206 | Epoch [1/10], Iter [1960] Loss: 0.0117 207 | Epoch [1/10], Iter [1970] Loss: 0.0078 208 | Epoch [1/10], Iter [1980] Loss: 0.0125 209 | Epoch [1/10], Iter [1990] Loss: 0.0086 210 | Epoch [1/10], Iter [2000] Loss: 0.0074 211 | Epoch [1/10], Iter [2010] Loss: 0.0104 212 | Epoch [1/10], Iter [2020] Loss: 0.0084 213 | Epoch [1/10], Iter [2030] Loss: 0.0073 214 | Epoch [1/10], Iter [2040] Loss: 0.0097 215 | Epoch [1/10], Iter [2050] Loss: 0.0102 216 | Epoch [1/10], Iter [2060] Loss: 0.0103 217 | Epoch [1/10], Iter [2070] Loss: 0.0087 218 | Epoch [1/10], Iter [2080] Loss: 0.0103 219 | Epoch [1/10], Iter [2090] Loss: 0.0139 220 | Epoch [1/10], Iter [2100] Loss: 0.0097 221 | Epoch [1/10], Iter [2110] Loss: 0.0098 222 | Epoch [1/10], Iter [2120] Loss: 0.0069 223 | Epoch [1/10], Iter [2130] Loss: 0.0063 224 | Epoch [1/10], Iter [2140] Loss: 0.0081 225 | Epoch [1/10], Iter [2150] Loss: 0.0103 226 | Epoch [1/10], Iter [2160] Loss: 0.0072 227 | Epoch [1/10], Iter [2170] Loss: 0.0092 228 | Epoch [1/10], Iter [2180] Loss: 0.0092 229 | Epoch [1/10], Iter [2190] Loss: 0.0101 230 | Epoch [1/10], Iter [2200] Loss: 0.0071 231 | Epoch [1/10], Iter [2210] Loss: 0.0085 232 | Epoch [1/10], Iter [2220] Loss: 0.0071 233 | Epoch [1/10], Iter [2230] Loss: 0.0078 234 | Epoch [1/10], Iter [2240] Loss: 0.0074 235 | Epoch [1/10], Iter [2250] Loss: 0.0105 236 | Epoch [1/10], Iter [2260] Loss: 0.0094 237 | Epoch [1/10], Iter [2270] Loss: 0.0074 238 | Epoch [1/10], Iter [2280] Loss: 0.0077 239 | Epoch [1/10], Iter [2290] Loss: 0.0100 240 | Epoch [1/10], Iter [2300] Loss: 0.0077 241 | Epoch [1/10], Iter [2310] Loss: 0.0065 242 | Epoch [1/10], Iter [2320] Loss: 0.0068 243 | Epoch [1/10], Iter [2330] Loss: 0.0093 244 | Epoch [1/10], Iter [2340] Loss: 0.0083 245 | Epoch [1/10], Iter [2350] Loss: 0.0069 246 | Epoch [1/10], Iter [2360] Loss: 0.0082 247 | Epoch [1/10], Iter [2370] Loss: 0.0078 248 | Epoch [1/10], Iter [2380] Loss: 0.0086 249 | Epoch [1/10], Iter [2390] Loss: 0.0069 250 | Epoch [1/10], Iter [2400] Loss: 0.0075 251 | Epoch [1/10], Iter [2410] Loss: 0.0094 252 | Epoch [1/10], Iter [2420] Loss: 0.0071 253 | Epoch [1/10], Iter [2430] Loss: 0.0073 254 | Epoch [1/10], Iter [2440] Loss: 0.0069 255 | Epoch [1/10], Iter [2450] Loss: 0.0063 256 | Epoch [1/10], Iter [2460] Loss: 0.0055 257 | Epoch [1/10], Iter [2470] Loss: 0.0097 258 | Epoch [1/10], Iter [2480] Loss: 0.0077 259 | Epoch [1/10], Iter [2490] Loss: 0.0086 260 | Epoch [1/10], Iter [2500] Loss: 0.0076 261 | Epoch [1/10], Iter [2510] Loss: 0.0077 262 | Epoch [1/10], Iter [2520] Loss: 0.0074 263 | Epoch [1/10], Iter [2530] Loss: 0.0108 264 | Epoch [1/10], Iter [2540] Loss: 0.0076 265 | Epoch [1/10], Iter [2550] Loss: 0.0085 266 | Epoch [1/10], Iter [2560] Loss: 0.0073 267 | Epoch [1/10], Iter [2570] Loss: 0.0072 268 | Epoch [1/10], Iter [2580] Loss: 0.0105 269 | Epoch [1/10], Iter [2590] Loss: 0.0075 270 | Epoch [1/10], Iter [2600] Loss: 0.0080 271 | Epoch [1/10], Iter [2610] Loss: 0.0086 272 | Epoch [1/10], Iter [2620] Loss: 0.0081 273 | Epoch [1/10], Iter [2630] Loss: 0.0078 274 | Epoch [1/10], Iter [2640] Loss: 0.0088 275 | Epoch [1/10], Iter [2650] Loss: 0.0081 276 | Epoch [1/10], Iter [2660] Loss: 0.0082 277 | Epoch [1/10], Iter [2670] Loss: 0.0076 278 | Epoch [1/10], Iter [2680] Loss: 0.0059 279 | Epoch [1/10], Iter [2690] Loss: 0.0067 280 | Epoch [1/10], Iter [2700] Loss: 0.0068 281 | Epoch [1/10], Iter [2710] Loss: 0.0062 282 | Epoch [1/10], Iter [2720] Loss: 0.0075 283 | Epoch [1/10], Iter [2730] Loss: 0.0071 284 | Epoch [1/10], Iter [2740] Loss: 0.0101 285 | Epoch [1/10], Iter [2750] Loss: 0.0068 286 | Epoch [1/10], Iter [2760] Loss: 0.0073 287 | Epoch [1/10], Iter [2770] Loss: 0.0077 288 | Epoch [1/10], Iter [2780] Loss: 0.0083 289 | Epoch [1/10], Iter [2790] Loss: 0.0059 290 | Epoch [1/10], Iter [2800] Loss: 0.0084 291 | Epoch [1/10], Iter [2810] Loss: 0.0075 292 | Epoch [1/10], Iter [2820] Loss: 0.0089 293 | Epoch [1/10], Iter [2830] Loss: 0.0075 294 | Epoch [1/10], Iter [2840] Loss: 0.0085 295 | Epoch [1/10], Iter [2850] Loss: 0.0128 296 | Epoch [1/10], Iter [2860] Loss: 0.0072 297 | Epoch [1/10], Iter [2870] Loss: 0.0071 298 | Epoch [1/10], Iter [2880] Loss: 0.0089 299 | Epoch [1/10], Iter [2890] Loss: 0.0116 300 | Epoch [1/10], Iter [2900] Loss: 0.0085 301 | Epoch [1/10], Iter [2910] Loss: 0.0064 302 | Epoch [1/10], Iter [2920] Loss: 0.0058 303 | Epoch [1/10], Iter [2930] Loss: 0.0070 304 | Epoch [1/10], Iter [2940] Loss: 0.0063 305 | Epoch [1/10], Iter [2950] Loss: 0.0077 306 | Epoch [1/10], Iter [2960] Loss: 0.0073 307 | Epoch [1/10], Iter [2970] Loss: 0.0077 308 | Epoch [1/10], Iter [2980] Loss: 0.0076 309 | Epoch [1/10], Iter [2990] Loss: 0.0083 310 | Epoch [1/10], Iter [3000] Loss: 0.0077 311 | Epoch [1/10], Iter [3010] Loss: 0.0084 312 | Epoch [1/10], Iter [3020] Loss: 0.0067 313 | Epoch [1/10], Iter [3030] Loss: 0.0072 314 | Epoch [1/10], Iter [3040] Loss: 0.0079 315 | Epoch [1/10], Iter [3050] Loss: 0.0090 316 | Epoch [1/10], Iter [3060] Loss: 0.0071 317 | Epoch [1/10], Iter [3070] Loss: 0.0086 318 | Epoch [1/10], Iter [3080] Loss: 0.0061 319 | Epoch [1/10], Iter [3090] Loss: 0.0088 320 | Epoch [1/10], Iter [3100] Loss: 0.0068 321 | Epoch [1/10], Iter [3110] Loss: 0.0070 322 | Epoch [1/10], Iter [3120] Loss: 0.0061 323 | Epoch [1/10], Iter [3130] Loss: 0.0056 324 | Epoch [1/10], Iter [3140] Loss: 0.0051 325 | Epoch [1/10], Iter [3150] Loss: 0.0072 326 | Epoch [1/10], Iter [3160] Loss: 0.0055 327 | Epoch [1/10], Iter [3170] Loss: 0.0068 328 | Epoch [1/10], Iter [3180] Loss: 0.0071 329 | Epoch [1/10], Iter [3190] Loss: 0.0076 330 | Epoch [1/10], Iter [3200] Loss: 0.0062 331 | Epoch [1/10], Iter [3210] Loss: 0.0062 332 | Epoch [1/10], Iter [3220] Loss: 0.0072 333 | Epoch [1/10], Iter [3230] Loss: 0.0093 334 | Epoch [1/10], Iter [3240] Loss: 0.0051 335 | Epoch [1/10], Iter [3250] Loss: 0.0060 336 | Epoch [1/10], Iter [3260] Loss: 0.0079 337 | Epoch [1/10], Iter [3270] Loss: 0.0060 338 | Epoch [1/10], Iter [3280] Loss: 0.0059 339 | Epoch [1/10], Iter [3290] Loss: 0.0058 340 | Epoch [1/10], Iter [3300] Loss: 0.0074 341 | Epoch [1/10], Iter [3310] Loss: 0.0062 342 | Epoch [1/10], Iter [3320] Loss: 0.0069 343 | Epoch [1/10], Iter [3330] Loss: 0.0068 344 | Epoch [1/10], Iter [3340] Loss: 0.0068 345 | Epoch [1/10], Iter [3350] Loss: 0.0061 346 | Epoch [1/10], Iter [3360] Loss: 0.0056 347 | Epoch [1/10], Iter [3370] Loss: 0.0059 348 | Epoch [1/10], Iter [3380] Loss: 0.0060 349 | Epoch [1/10], Iter [3390] Loss: 0.0053 350 | Epoch [1/10], Iter [3400] Loss: 0.0056 351 | Epoch [1/10], Iter [3410] Loss: 0.0054 352 | Epoch [1/10], Iter [3420] Loss: 0.0080 353 | Epoch [1/10], Iter [3430] Loss: 0.0073 354 | Epoch [1/10], Iter [3440] Loss: 0.0073 355 | Epoch [1/10], Iter [3450] Loss: 0.0069 356 | Epoch [1/10], Iter [3460] Loss: 0.0061 357 | Epoch [1/10], Iter [3470] Loss: 0.0068 358 | Epoch [1/10], Iter [3480] Loss: 0.0067 359 | Epoch [1/10], Iter [3490] Loss: 0.0062 360 | Epoch [1/10], Iter [3500] Loss: 0.0074 361 | Epoch [1/10], Iter [3510] Loss: 0.0053 362 | Epoch [1/10], Iter [3520] Loss: 0.0061 363 | Epoch [1/10], Iter [3530] Loss: 0.0091 364 | Epoch [1/10], Iter [3540] Loss: 0.0062 365 | Epoch [1/10], Iter [3550] Loss: 0.0064 366 | Epoch [1/10], Iter [3560] Loss: 0.0066 367 | Epoch [1/10], Iter [3570] Loss: 0.0070 368 | Epoch [1/10], Iter [3580] Loss: 0.0051 369 | Epoch [1/10], Iter [3590] Loss: 0.0058 370 | Epoch [1/10], Iter [3600] Loss: 0.0062 371 | Epoch [1/10], Iter [3610] Loss: 0.0051 372 | Epoch [1/10], Iter [3620] Loss: 0.0063 373 | Epoch [1/10], Iter [3630] Loss: 0.0087 374 | Epoch [1/10], Iter [3640] Loss: 0.0079 375 | Epoch [1/10], Iter [3650] Loss: 0.0080 376 | Epoch [1/10], Iter [3660] Loss: 0.0075 377 | Epoch [1/10], Iter [3670] Loss: 0.0065 378 | Epoch [1/10], Iter [3680] Loss: 0.0059 379 | Epoch [1/10], Iter [3690] Loss: 0.0080 380 | Epoch [1/10], Iter [3700] Loss: 0.0099 381 | Epoch [1/10], Iter [3710] Loss: 0.0112 382 | Epoch [1/10], Iter [3720] Loss: 0.0049 383 | Epoch [1/10], Iter [3730] Loss: 0.0061 384 | Epoch [1/10], Iter [3740] Loss: 0.0068 385 | Epoch [1/10], Iter [3750] Loss: 0.0046 386 | Epoch [1/10], Iter [3760] Loss: 0.0066 387 | Epoch [1/10], Iter [3770] Loss: 0.0056 388 | Epoch [1/10], Iter [3780] Loss: 0.0065 389 | Epoch [1/10], Iter [3790] Loss: 0.0059 390 | Epoch [1/10], Iter [3800] Loss: 0.0068 391 | Epoch [1/10], Iter [3810] Loss: 0.0055 392 | Epoch [1/10], Iter [3820] Loss: 0.0064 393 | Epoch [1/10], Iter [3830] Loss: 0.0067 394 | Epoch [1/10], Iter [3840] Loss: 0.0051 395 | Epoch [1/10], Iter [3850] Loss: 0.0066 396 | Epoch [1/10], Iter [3860] Loss: 0.0052 397 | Epoch [1/10], Iter [3870] Loss: 0.0056 398 | Epoch [1/10], Iter [3880] Loss: 0.0100 399 | Epoch [1/10], Iter [3890] Loss: 0.0058 400 | Epoch [1/10], Iter [3900] Loss: 0.0068 401 | Epoch [1/10], Iter [3910] Loss: 0.0071 402 | Epoch [1/10], Iter [3920] Loss: 0.0100 403 | Epoch [1/10], Iter [3930] Loss: 0.0064 404 | Epoch [1/10], Iter [3940] Loss: 0.0087 405 | Epoch [1/10], Iter [3950] Loss: 0.0061 406 | Epoch [1/10], Iter [3960] Loss: 0.0061 407 | Epoch [1/10], Iter [3970] Loss: 0.0053 408 | Epoch [1/10], Iter [3980] Loss: 0.0074 409 | Epoch [1/10], Iter [3990] Loss: 0.0063 410 | Epoch [1/10], Iter [4000] Loss: 0.0068 411 | Epoch [1/10], Iter [4010] Loss: 0.0067 412 | Epoch [1/10], Iter [4020] Loss: 0.0079 413 | Epoch [1/10], Iter [4030] Loss: 0.0069 414 | Epoch [1/10], Iter [4040] Loss: 0.0062 415 | Epoch [1/10], Iter [4050] Loss: 0.0059 416 | Epoch [1/10], Iter [4060] Loss: 0.0065 417 | Epoch [1/10], Iter [4070] Loss: 0.0100 418 | Epoch [1/10], Iter [4080] Loss: 0.0054 419 | Epoch [1/10], Iter [4090] Loss: 0.0069 420 | Epoch [1/10], Iter [4100] Loss: 0.0060 421 | Epoch [1/10], Iter [4110] Loss: 0.0057 422 | Epoch [1/10], Iter [4120] Loss: 0.0057 423 | Epoch [1/10], Iter [4130] Loss: 0.0053 424 | Epoch [1/10], Iter [4140] Loss: 0.0068 425 | Epoch [1/10], Iter [4150] Loss: 0.0077 426 | Epoch [1/10], Iter [4160] Loss: 0.0065 427 | Epoch [1/10], Iter [4170] Loss: 0.0066 428 | Epoch [1/10], Iter [4180] Loss: 0.0053 429 | Epoch [1/10], Iter [4190] Loss: 0.0058 430 | Epoch [1/10], Iter [4200] Loss: 0.0066 431 | Epoch [1/10], Iter [4210] Loss: 0.0065 432 | Epoch [1/10], Iter [4220] Loss: 0.0053 433 | Epoch [1/10], Iter [4230] Loss: 0.0071 434 | Epoch [1/10], Iter [4240] Loss: 0.0058 435 | Epoch [1/10], Iter [4250] Loss: 0.0065 436 | Epoch [1/10], Iter [4260] Loss: 0.0068 437 | Epoch [1/10], Iter [4270] Loss: 0.0053 438 | Epoch [1/10], Iter [4280] Loss: 0.0060 439 | Epoch [1/10], Iter [4290] Loss: 0.0047 440 | Epoch [1/10], Iter [4300] Loss: 0.0075 441 | Epoch [1/10], Iter [4310] Loss: 0.0055 442 | Epoch [1/10], Iter [4320] Loss: 0.0050 443 | Epoch [1/10], Iter [4330] Loss: 0.0052 444 | Epoch [1/10], Iter [4340] Loss: 0.0057 445 | Epoch [1/10], Iter [4350] Loss: 0.0048 446 | Epoch [1/10], Iter [4360] Loss: 0.0091 447 | Epoch [1/10], Iter [4370] Loss: 0.0062 448 | Epoch [1/10], Iter [4380] Loss: 0.0075 449 | Epoch [1/10], Iter [4390] Loss: 0.0068 450 | Epoch [1/10], Iter [4400] Loss: 0.0072 451 | Epoch [1/10], Iter [4410] Loss: 0.0053 452 | Epoch [1/10], Iter [4420] Loss: 0.0048 453 | Epoch [1/10], Iter [4430] Loss: 0.0051 454 | Epoch [1/10], Iter [4440] Loss: 0.0071 455 | Epoch [1/10], Iter [4450] Loss: 0.0076 456 | Epoch [1/10], Iter [4460] Loss: 0.0084 457 | Epoch [1/10], Iter [4470] Loss: 0.0067 458 | Epoch [1/10], Iter [4480] Loss: 0.0065 459 | Epoch [1/10], Iter [4490] Loss: 0.0053 460 | Epoch [1/10], Iter [4500] Loss: 0.0100 461 | Epoch [1/10], Iter [4510] Loss: 0.0058 462 | Epoch [1/10], Iter [4520] Loss: 0.0051 463 | Epoch [1/10], Iter [4530] Loss: 0.0051 464 | Epoch [1/10], Iter [4540] Loss: 0.0081 465 | Epoch [1/10], Iter [4550] Loss: 0.0057 466 | Epoch [1/10], Iter [4560] Loss: 0.0059 467 | Epoch [1/10], Iter [4570] Loss: 0.0048 468 | Epoch [1/10], Iter [4580] Loss: 0.0054 469 | Epoch [1/10], Iter [4590] Loss: 0.0061 470 | Epoch [1/10], Iter [4600] Loss: 0.0072 471 | Epoch [1/10], Iter [4610] Loss: 0.0057 472 | Epoch [1/10], Iter [4620] Loss: 0.0069 473 | Epoch [1/10], Iter [4630] Loss: 0.0067 474 | Epoch [1/10], Iter [4640] Loss: 0.0041 475 | Epoch [1/10], Iter [4650] Loss: 0.0059 476 | Epoch [1/10], Iter [4660] Loss: 0.0066 477 | Epoch [1/10], Iter [4670] Loss: 0.0069 478 | Epoch [1/10], Iter [4680] Loss: 0.0065 479 | Epoch [1/10], Iter [4690] Loss: 0.0061 480 | Epoch [1/10], Iter [4700] Loss: 0.0043 481 | Epoch [1/10], Iter [4710] Loss: 0.0049 482 | Epoch [1/10], Iter [4720] Loss: 0.0060 483 | Epoch [1/10], Iter [4730] Loss: 0.0062 484 | Epoch [1/10], Iter [4740] Loss: 0.0059 485 | Epoch [1/10], Iter [4750] Loss: 0.0046 486 | Epoch [1/10], Iter [4760] Loss: 0.0063 487 | Epoch [1/10], Iter [4770] Loss: 0.0050 488 | Epoch [1/10], Iter [4780] Loss: 0.0051 489 | Epoch [1/10], Iter [4790] Loss: 0.0061 490 | Epoch [1/10], Iter [4800] Loss: 0.0080 491 | Epoch [1/10], Iter [4810] Loss: 0.0049 492 | Epoch [1/10], Iter [4820] Loss: 0.0086 493 | Epoch [1/10], Iter [4830] Loss: 0.0047 494 | Epoch [1/10], Iter [4840] Loss: 0.0054 495 | Epoch [1/10], Iter [4850] Loss: 0.0070 496 | Epoch [1/10], Iter [4860] Loss: 0.0055 497 | Epoch [1/10], Iter [4870] Loss: 0.0061 498 | Epoch [1/10], Iter [4880] Loss: 0.0057 499 | Epoch [1/10], Iter [4890] Loss: 0.0069 500 | Epoch [1/10], Iter [4900] Loss: 0.0056 501 | Epoch [1/10], Iter [4910] Loss: 0.0076 502 | Epoch [1/10], Iter [4920] Loss: 0.0053 503 | Epoch [1/10], Iter [4930] Loss: 0.0056 504 | Epoch [1/10], Iter [4940] Loss: 0.0059 505 | Epoch [1/10], Iter [4950] Loss: 0.0102 506 | Epoch [1/10], Iter [4960] Loss: 0.0042 507 | Epoch [1/10], Iter [4970] Loss: 0.0068 508 | Epoch [1/10], Iter [4980] Loss: 0.0049 509 | Epoch [1/10], Iter [4990] Loss: 0.0057 510 | Epoch [1/10], Iter [5000] Loss: 0.0048 511 | Epoch [1/10], Iter [5010] Loss: 0.0048 512 | Epoch [1/10], Iter [5020] Loss: 0.0052 513 | Epoch [1/10], Iter [5030] Loss: 0.0063 514 | Epoch [1/10], Iter [5040] Loss: 0.0054 515 | Epoch [1/10], Iter [5050] Loss: 0.0062 516 | Epoch [1/10], Iter [5060] Loss: 0.0058 517 | Epoch [1/10], Iter [5070] Loss: 0.0052 518 | Epoch [1/10], Iter [5080] Loss: 0.0061 519 | Epoch [1/10], Iter [5090] Loss: 0.0053 520 | Epoch [1/10], Iter [5100] Loss: 0.0053 521 | Epoch [1/10], Iter [5110] Loss: 0.0054 522 | Epoch [1/10], Iter [5120] Loss: 0.0049 523 | Epoch [1/10], Iter [5130] Loss: 0.0045 524 | Epoch [1/10], Iter [5140] Loss: 0.0053 525 | Epoch [1/10], Iter [5150] Loss: 0.0065 526 | Epoch [1/10], Iter [5160] Loss: 0.0081 527 | Epoch [1/10], Iter [5170] Loss: 0.0071 528 | Epoch [1/10], Iter [5180] Loss: 0.0059 529 | Epoch [1/10], Iter [5190] Loss: 0.0065 530 | Epoch [1/10], Iter [5200] Loss: 0.0042 531 | Epoch [1/10], Iter [5210] Loss: 0.0057 532 | Epoch [1/10], Iter [5220] Loss: 0.0056 533 | Epoch [1/10], Iter [5230] Loss: 0.0039 534 | Epoch [1/10], Iter [5240] Loss: 0.0050 535 | Epoch [1/10], Iter [5250] Loss: 0.0053 536 | Epoch [1/10], Iter [5260] Loss: 0.0065 537 | Epoch [1/10], Iter [5270] Loss: 0.0054 538 | Epoch [1/10], Iter [5280] Loss: 0.0050 539 | Epoch [1/10], Iter [5290] Loss: 0.0054 540 | Epoch [1/10], Iter [5300] Loss: 0.0051 541 | Epoch [1/10], Iter [5310] Loss: 0.0064 542 | Epoch [1/10], Iter [5320] Loss: 0.0052 543 | Epoch [1/10], Iter [5330] Loss: 0.0045 544 | Epoch [1/10], Iter [5340] Loss: 0.0045 545 | Epoch [1/10], Iter [5350] Loss: 0.0066 546 | Epoch [1/10], Iter [5360] Loss: 0.0051 547 | Epoch [1/10], Iter [5370] Loss: 0.0061 548 | Epoch [1/10], Iter [5380] Loss: 0.0051 549 | Epoch [1/10], Iter [5390] Loss: 0.0046 550 | Epoch [1/10], Iter [5400] Loss: 0.0055 551 | Epoch [1/10], Iter [5410] Loss: 0.0061 552 | Epoch [1/10], Iter [5420] Loss: 0.0053 553 | Epoch [1/10], Iter [5430] Loss: 0.0046 554 | Epoch [1/10], Iter [5440] Loss: 0.0042 555 | Epoch [1/10], Iter [5450] Loss: 0.0042 556 | Epoch [1/10], Iter [5460] Loss: 0.0057 557 | Epoch [1/10], Iter [5470] Loss: 0.0057 558 | Epoch [1/10], Iter [5480] Loss: 0.0051 559 | Epoch [1/10], Iter [5490] Loss: 0.0063 560 | Epoch [1/10], Iter [5500] Loss: 0.0044 561 | Epoch [1/10], Iter [5510] Loss: 0.0053 562 | Epoch [1/10], Iter [5520] Loss: 0.0043 563 | Epoch [1/10], Iter [5530] Loss: 0.0043 564 | Epoch [1/10], Iter [5540] Loss: 0.0063 565 | Epoch [1/10], Iter [5550] Loss: 0.0051 566 | Epoch [1/10], Iter [5560] Loss: 0.0050 567 | Epoch [1/10], Iter [5570] Loss: 0.0053 568 | Epoch [1/10], Iter [5580] Loss: 0.0076 569 | Epoch [1/10], Iter [5590] Loss: 0.0049 570 | Epoch [1/10], Iter [5600] Loss: 0.0051 571 | Epoch [1/10], Iter [5610] Loss: 0.0061 572 | Epoch [1/10], Iter [5620] Loss: 0.0045 573 | Epoch [1/10], Iter [5630] Loss: 0.0047 574 | Epoch [1/10], Iter [5640] Loss: 0.0055 575 | Epoch [1/10], Iter [5650] Loss: 0.0045 576 | Epoch [1/10], Iter [5660] Loss: 0.0066 577 | Epoch [1/10], Iter [5670] Loss: 0.0050 578 | Epoch [1/10], Iter [5680] Loss: 0.0054 579 | Epoch [1/10], Iter [5690] Loss: 0.0047 580 | Epoch [1/10], Iter [5700] Loss: 0.0058 581 | Epoch [1/10], Iter [5710] Loss: 0.0042 582 | Epoch [1/10], Iter [5720] Loss: 0.0044 583 | Epoch [1/10], Iter [5730] Loss: 0.0043 584 | Epoch [1/10], Iter [5740] Loss: 0.0043 585 | Epoch [1/10], Iter [5750] Loss: 0.0065 586 | Epoch [1/10], Iter [5760] Loss: 0.0063 587 | Epoch [1/10], Iter [5770] Loss: 0.0040 588 | Epoch [1/10], Iter [5780] Loss: 0.0049 589 | Epoch [1/10], Iter [5790] Loss: 0.0047 590 | Epoch [1/10], Iter [5800] Loss: 0.0044 591 | Epoch [1/10], Iter [5810] Loss: 0.0047 592 | Epoch [1/10], Iter [5820] Loss: 0.0055 593 | Epoch [1/10], Iter [5830] Loss: 0.0045 594 | Epoch [1/10], Iter [5840] Loss: 0.0047 595 | Epoch [1/10], Iter [5850] Loss: 0.0044 596 | Epoch [1/10], Iter [5860] Loss: 0.0043 597 | Epoch [1/10], Iter [5870] Loss: 0.0041 598 | Epoch [1/10], Iter [5880] Loss: 0.0039 599 | Epoch [1/10], Iter [5890] Loss: 0.0040 600 | Epoch [1/10], Iter [5900] Loss: 0.0050 601 | Epoch [1/10], Iter [5910] Loss: 0.0047 602 | Epoch [1/10], Iter [5920] Loss: 0.0040 603 | Epoch [1/10], Iter [5930] Loss: 0.0045 604 | Epoch [1/10], Iter [5940] Loss: 0.0052 605 | Epoch [1/10], Iter [5950] Loss: 0.0039 606 | Epoch [1/10], Iter [5960] Loss: 0.0063 607 | Epoch [1/10], Iter [5970] Loss: 0.0053 608 | Epoch [1/10], Iter [5980] Loss: 0.0045 609 | Epoch [1/10], Iter [5990] Loss: 0.0043 610 | Epoch [1/10], Iter [6000] Loss: 0.0056 611 | Epoch [1/10], Iter [6010] Loss: 0.0058 612 | Epoch [1/10], Iter [6020] Loss: 0.0050 613 | Epoch [1/10], Iter [6030] Loss: 0.0041 614 | Epoch [1/10], Iter [6040] Loss: 0.0049 615 | Epoch [1/10], Iter [6050] Loss: 0.0055 616 | Epoch [1/10], Iter [6060] Loss: 0.0050 617 | Epoch [1/10], Iter [6070] Loss: 0.0055 618 | Epoch [1/10], Iter [6080] Loss: 0.0095 619 | Epoch [1/10], Iter [6090] Loss: 0.0051 620 | Epoch [1/10], Iter [6100] Loss: 0.0051 621 | Epoch [1/10], Iter [6110] Loss: 0.0058 622 | Epoch [1/10], Iter [6120] Loss: 0.0081 623 | Epoch [1/10], Iter [6130] Loss: 0.0050 624 | Epoch [1/10], Iter [6140] Loss: 0.0041 625 | Epoch [1/10], Iter [6150] Loss: 0.0047 626 | Epoch [1/10], Iter [6160] Loss: 0.0054 627 | Epoch [1/10], Iter [6170] Loss: 0.0048 628 | Epoch [1/10], Iter [6180] Loss: 0.0049 629 | Epoch [1/10], Iter [6190] Loss: 0.0065 630 | Epoch [1/10], Iter [6200] Loss: 0.0045 631 | Epoch [1/10], Iter [6210] Loss: 0.0061 632 | Epoch [1/10], Iter [6220] Loss: 0.0050 633 | Epoch [1/10], Iter [6230] Loss: 0.0067 634 | Epoch [1/10], Iter [6240] Loss: 0.0047 635 | Epoch [1/10], Iter [6250] Loss: 0.0067 636 | Epoch [1/10], Iter [6260] Loss: 0.0041 637 | Epoch [1/10], Iter [6270] Loss: 0.0047 638 | Epoch [1/10], Iter [6280] Loss: 0.0047 639 | Epoch [1/10], Iter [6290] Loss: 0.0050 640 | Epoch [1/10], Iter [6300] Loss: 0.0050 641 | Epoch [1/10], Iter [6310] Loss: 0.0048 642 | Epoch [1/10], Iter [6320] Loss: 0.0055 643 | Epoch [1/10], Iter [6330] Loss: 0.0055 644 | Epoch [1/10], Iter [6340] Loss: 0.0030 645 | Epoch [1/10], Iter [6350] Loss: 0.0050 646 | Epoch [1/10], Iter [6360] Loss: 0.0054 647 | Epoch [1/10], Iter [6370] Loss: 0.0041 648 | Epoch [1/10], Iter [6380] Loss: 0.0045 649 | Epoch [1/10], Iter [6390] Loss: 0.0037 650 | Epoch [1/10], Iter [6400] Loss: 0.0041 651 | Epoch [1/10], Iter [6410] Loss: 0.0049 652 | Epoch [1/10], Iter [6420] Loss: 0.0054 653 | Epoch [1/10], Iter [6430] Loss: 0.0057 654 | Epoch [1/10], Iter [6440] Loss: 0.0060 655 | Epoch [1/10], Iter [6450] Loss: 0.0053 656 | Epoch [1/10], Iter [6460] Loss: 0.0056 657 | Epoch [1/10], Iter [6470] Loss: 0.0054 658 | Epoch [1/10], Iter [6480] Loss: 0.0053 659 | Epoch [1/10], Iter [6490] Loss: 0.0042 660 | Epoch [1/10], Iter [6500] Loss: 0.0043 661 | Epoch [1/10], Iter [6510] Loss: 0.0036 662 | Epoch [1/10], Iter [6520] Loss: 0.0048 663 | Epoch [1/10], Iter [6530] Loss: 0.0051 664 | Epoch [1/10], Iter [6540] Loss: 0.0042 665 | Epoch [1/10], Iter [6550] Loss: 0.0031 666 | Epoch [1/10], Iter [6560] Loss: 0.0048 667 | Epoch [1/10], Iter [6570] Loss: 0.0041 668 | Epoch [1/10], Iter [6580] Loss: 0.0042 669 | Epoch [1/10], Iter [6590] Loss: 0.0052 670 | Epoch [1/10], Iter [6600] Loss: 0.0061 671 | Epoch [1/10], Iter [6610] Loss: 0.0058 672 | Epoch [1/10], Iter [6620] Loss: 0.0050 673 | Epoch [1/10], Iter [6630] Loss: 0.0062 674 | Epoch [1/10], Iter [6640] Loss: 0.0046 675 | Epoch [1/10], Iter [6650] Loss: 0.0045 676 | Epoch [1/10], Iter [6660] Loss: 0.0037 677 | Epoch [1/10], Iter [6670] Loss: 0.0047 678 | Epoch [1/10], Iter [6680] Loss: 0.0035 679 | Epoch [1/10], Iter [6690] Loss: 0.0063 680 | Epoch [1/10], Iter [6700] Loss: 0.0049 681 | Epoch [1/10], Iter [6710] Loss: 0.0052 682 | Epoch [1/10], Iter [6720] Loss: 0.0051 683 | Epoch [1/10], Iter [6730] Loss: 0.0048 684 | Epoch [1/10], Iter [6740] Loss: 0.0052 685 | Epoch [1/10], Iter [6750] Loss: 0.0036 686 | Epoch [1/10], Iter [6760] Loss: 0.0046 687 | Epoch [1/10], Iter [6770] Loss: 0.0042 688 | Epoch [1/10], Iter [6780] Loss: 0.0041 689 | Epoch [1/10], Iter [6790] Loss: 0.0065 690 | Epoch [1/10], Iter [6800] Loss: 0.0049 691 | Epoch [1/10], Iter [6810] Loss: 0.0052 692 | Epoch [1/10], Iter [6820] Loss: 0.0041 693 | Epoch [1/10], Iter [6830] Loss: 0.0047 694 | Epoch [1/10], Iter [6840] Loss: 0.0048 695 | Epoch [1/10], Iter [6850] Loss: 0.0050 696 | Epoch [1/10], Iter [6860] Loss: 0.0051 697 | Epoch [1/10], Iter [6870] Loss: 0.0048 698 | Epoch [1/10], Iter [6880] Loss: 0.0047 699 | Epoch [1/10], Iter [6890] Loss: 0.0047 700 | Epoch [1/10], Iter [6900] Loss: 0.0043 701 | Epoch [1/10], Iter [6910] Loss: 0.0034 702 | Epoch [1/10], Iter [6920] Loss: 0.0043 703 | Epoch [1/10], Iter [6930] Loss: 0.0057 704 | Epoch [1/10], Iter [6940] Loss: 0.0046 705 | Epoch [1/10], Iter [6950] Loss: 0.0047 706 | Epoch [1/10], Iter [6960] Loss: 0.0038 707 | Epoch [1/10], Iter [6970] Loss: 0.0039 708 | Epoch [1/10], Iter [6980] Loss: 0.0040 709 | Epoch [1/10], Iter [6990] Loss: 0.0036 710 | Epoch [1/10], Iter [7000] Loss: 0.0044 711 | Epoch [1/10], Iter [7010] Loss: 0.0043 712 | Epoch [1/10], Iter [7020] Loss: 0.0053 713 | Epoch [1/10], Iter [7030] Loss: 0.0043 714 | Epoch [1/10], Iter [7040] Loss: 0.0044 715 | Epoch [1/10], Iter [7050] Loss: 0.0036 716 | Epoch [1/10], Iter [7060] Loss: 0.0041 717 | Epoch [1/10], Iter [7070] Loss: 0.0050 718 | Epoch [1/10], Iter [7080] Loss: 0.0042 719 | Epoch [1/10], Iter [7090] Loss: 0.0047 720 | Epoch [1/10], Iter [7100] Loss: 0.0054 721 | Epoch [1/10], Iter [7110] Loss: 0.0052 722 | Epoch [1/10], Iter [7120] Loss: 0.0044 723 | Epoch [1/10], Iter [7130] Loss: 0.0048 724 | Epoch [1/10], Iter [7140] Loss: 0.0040 725 | Epoch [1/10], Iter [7150] Loss: 0.0049 726 | Epoch [1/10], Iter [7160] Loss: 0.0043 727 | Epoch [1/10], Iter [7170] Loss: 0.0048 728 | Epoch [1/10], Iter [7180] Loss: 0.0062 729 | Epoch [1/10], Iter [7190] Loss: 0.0052 730 | Epoch [1/10], Iter [7200] Loss: 0.0042 731 | Epoch [1/10], Iter [7210] Loss: 0.0036 732 | Epoch [1/10], Iter [7220] Loss: 0.0033 733 | Epoch [1/10], Iter [7230] Loss: 0.0059 734 | Epoch [1/10], Iter [7240] Loss: 0.0033 735 | Epoch [1/10], Iter [7250] Loss: 0.0040 736 | Epoch [1/10], Iter [7260] Loss: 0.0039 737 | Epoch [1/10], Iter [7270] Loss: 0.0043 738 | Epoch [1/10], Iter [7280] Loss: 0.0050 739 | Epoch [1/10], Iter [7290] Loss: 0.0046 740 | Epoch [1/10], Iter [7300] Loss: 0.0055 741 | Epoch [1/10], Iter [7310] Loss: 0.0045 742 | Epoch [1/10], Iter [7320] Loss: 0.0044 743 | Epoch [1/10], Iter [7330] Loss: 0.0041 744 | Epoch [1/10], Iter [7340] Loss: 0.0049 745 | Epoch [1/10], Iter [7350] Loss: 0.0040 746 | Epoch [1/10], Iter [7360] Loss: 0.0046 747 | Epoch [1/10], Iter [7370] Loss: 0.0058 748 | Epoch [1/10], Iter [7380] Loss: 0.0039 749 | Epoch [1/10], Iter [7390] Loss: 0.0048 750 | Epoch [2/10], Iter [10] Loss: 0.0053 751 | Epoch [2/10], Iter [20] Loss: 0.0040 752 | Epoch [2/10], Iter [30] Loss: 0.0038 753 | Epoch [2/10], Iter [40] Loss: 0.0044 754 | Epoch [2/10], Iter [50] Loss: 0.0044 755 | Epoch [2/10], Iter [60] Loss: 0.0038 756 | Epoch [2/10], Iter [70] Loss: 0.0040 757 | Epoch [2/10], Iter [80] Loss: 0.0037 758 | Epoch [2/10], Iter [90] Loss: 0.0045 759 | Epoch [2/10], Iter [100] Loss: 0.0041 760 | Epoch [2/10], Iter [110] Loss: 0.0041 761 | Epoch [2/10], Iter [120] Loss: 0.0044 762 | Epoch [2/10], Iter [130] Loss: 0.0041 763 | Epoch [2/10], Iter [140] Loss: 0.0054 764 | Epoch [2/10], Iter [150] Loss: 0.0036 765 | Epoch [2/10], Iter [160] Loss: 0.0040 766 | Epoch [2/10], Iter [170] Loss: 0.0043 767 | Epoch [2/10], Iter [180] Loss: 0.0052 768 | Epoch [2/10], Iter [190] Loss: 0.0046 769 | Epoch [2/10], Iter [200] Loss: 0.0044 770 | Epoch [2/10], Iter [210] Loss: 0.0036 771 | Epoch [2/10], Iter [220] Loss: 0.0031 772 | Epoch [2/10], Iter [230] Loss: 0.0059 773 | Epoch [2/10], Iter [240] Loss: 0.0042 774 | Epoch [2/10], Iter [250] Loss: 0.0037 775 | Epoch [2/10], Iter [260] Loss: 0.0029 776 | Epoch [2/10], Iter [270] Loss: 0.0038 777 | Epoch [2/10], Iter [280] Loss: 0.0069 778 | Epoch [2/10], Iter [290] Loss: 0.0058 779 | Epoch [2/10], Iter [300] Loss: 0.0048 780 | Epoch [2/10], Iter [310] Loss: 0.0053 781 | Epoch [2/10], Iter [320] Loss: 0.0040 782 | Epoch [2/10], Iter [330] Loss: 0.0044 783 | Epoch [2/10], Iter [340] Loss: 0.0037 784 | Epoch [2/10], Iter [350] Loss: 0.0045 785 | Epoch [2/10], Iter [360] Loss: 0.0043 786 | Epoch [2/10], Iter [370] Loss: 0.0036 787 | Epoch [2/10], Iter [380] Loss: 0.0041 788 | Epoch [2/10], Iter [390] Loss: 0.0048 789 | Epoch [2/10], Iter [400] Loss: 0.0030 790 | Epoch [2/10], Iter [410] Loss: 0.0042 791 | Epoch [2/10], Iter [420] Loss: 0.0048 792 | Epoch [2/10], Iter [430] Loss: 0.0044 793 | Epoch [2/10], Iter [440] Loss: 0.0032 794 | Epoch [2/10], Iter [450] Loss: 0.0042 795 | Epoch [2/10], Iter [460] Loss: 0.0039 796 | Epoch [2/10], Iter [470] Loss: 0.0038 797 | Epoch [2/10], Iter [480] Loss: 0.0037 798 | Epoch [2/10], Iter [490] Loss: 0.0040 799 | Epoch [2/10], Iter [500] Loss: 0.0051 800 | Epoch [2/10], Iter [510] Loss: 0.0056 801 | Epoch [2/10], Iter [520] Loss: 0.0047 802 | Epoch [2/10], Iter [530] Loss: 0.0038 803 | Epoch [2/10], Iter [540] Loss: 0.0052 804 | Epoch [2/10], Iter [550] Loss: 0.0038 805 | Epoch [2/10], Iter [560] Loss: 0.0033 806 | Epoch [2/10], Iter [570] Loss: 0.0039 807 | Epoch [2/10], Iter [580] Loss: 0.0047 808 | Epoch [2/10], Iter [590] Loss: 0.0041 809 | Epoch [2/10], Iter [600] Loss: 0.0050 810 | Epoch [2/10], Iter [610] Loss: 0.0039 811 | Epoch [2/10], Iter [620] Loss: 0.0037 812 | Epoch [2/10], Iter [630] Loss: 0.0032 813 | Epoch [2/10], Iter [640] Loss: 0.0036 814 | Epoch [2/10], Iter [650] Loss: 0.0039 815 | Epoch [2/10], Iter [660] Loss: 0.0042 816 | Epoch [2/10], Iter [670] Loss: 0.0038 817 | Epoch [2/10], Iter [680] Loss: 0.0055 818 | Epoch [2/10], Iter [690] Loss: 0.0050 819 | Epoch [2/10], Iter [700] Loss: 0.0041 820 | Epoch [2/10], Iter [710] Loss: 0.0050 821 | Epoch [2/10], Iter [720] Loss: 0.0036 822 | Epoch [2/10], Iter [730] Loss: 0.0040 823 | Epoch [2/10], Iter [740] Loss: 0.0041 824 | Epoch [2/10], Iter [750] Loss: 0.0033 825 | Epoch [2/10], Iter [760] Loss: 0.0038 826 | Epoch [2/10], Iter [770] Loss: 0.0038 827 | Epoch [2/10], Iter [780] Loss: 0.0051 828 | Epoch [2/10], Iter [790] Loss: 0.0043 829 | Epoch [2/10], Iter [800] Loss: 0.0042 830 | Epoch [2/10], Iter [810] Loss: 0.0036 831 | Epoch [2/10], Iter [820] Loss: 0.0042 832 | Epoch [2/10], Iter [830] Loss: 0.0041 833 | Epoch [2/10], Iter [840] Loss: 0.0033 834 | Epoch [2/10], Iter [850] Loss: 0.0039 835 | Epoch [2/10], Iter [860] Loss: 0.0040 836 | Epoch [2/10], Iter [870] Loss: 0.0048 837 | Epoch [2/10], Iter [880] Loss: 0.0057 838 | Epoch [2/10], Iter [890] Loss: 0.0032 839 | Epoch [2/10], Iter [900] Loss: 0.0040 840 | Epoch [2/10], Iter [910] Loss: 0.0037 841 | Epoch [2/10], Iter [920] Loss: 0.0036 842 | Epoch [2/10], Iter [930] Loss: 0.0037 843 | Epoch [2/10], Iter [940] Loss: 0.0035 844 | Epoch [2/10], Iter [950] Loss: 0.0037 845 | Epoch [2/10], Iter [960] Loss: 0.0043 846 | Epoch [2/10], Iter [970] Loss: 0.0034 847 | Epoch [2/10], Iter [980] Loss: 0.0051 848 | Epoch [2/10], Iter [990] Loss: 0.0040 849 | Epoch [2/10], Iter [1000] Loss: 0.0039 850 | Epoch [2/10], Iter [1010] Loss: 0.0043 851 | Epoch [2/10], Iter [1020] Loss: 0.0044 852 | Epoch [2/10], Iter [1030] Loss: 0.0037 853 | Epoch [2/10], Iter [1040] Loss: 0.0043 854 | Epoch [2/10], Iter [1050] Loss: 0.0037 855 | Epoch [2/10], Iter [1060] Loss: 0.0035 856 | Epoch [2/10], Iter [1070] Loss: 0.0036 857 | Epoch [2/10], Iter [1080] Loss: 0.0038 858 | Epoch [2/10], Iter [1090] Loss: 0.0043 859 | Epoch [2/10], Iter [1100] Loss: 0.0038 860 | Epoch [2/10], Iter [1110] Loss: 0.0038 861 | Epoch [2/10], Iter [1120] Loss: 0.0039 862 | Epoch [2/10], Iter [1130] Loss: 0.0052 863 | Epoch [2/10], Iter [1140] Loss: 0.0040 864 | Epoch [2/10], Iter [1150] Loss: 0.0055 865 | Epoch [2/10], Iter [1160] Loss: 0.0035 866 | Epoch [2/10], Iter [1170] Loss: 0.0038 867 | Epoch [2/10], Iter [1180] Loss: 0.0063 868 | Epoch [2/10], Iter [1190] Loss: 0.0043 869 | Epoch [2/10], Iter [1200] Loss: 0.0042 870 | Epoch [2/10], Iter [1210] Loss: 0.0037 871 | Epoch [2/10], Iter [1220] Loss: 0.0054 872 | Epoch [2/10], Iter [1230] Loss: 0.0034 873 | Epoch [2/10], Iter [1240] Loss: 0.0043 874 | Epoch [2/10], Iter [1250] Loss: 0.0046 875 | Epoch [2/10], Iter [1260] Loss: 0.0031 876 | Epoch [2/10], Iter [1270] Loss: 0.0042 877 | Epoch [2/10], Iter [1280] Loss: 0.0049 878 | Epoch [2/10], Iter [1290] Loss: 0.0038 879 | Epoch [2/10], Iter [1300] Loss: 0.0036 880 | Epoch [2/10], Iter [1310] Loss: 0.0038 881 | Epoch [2/10], Iter [1320] Loss: 0.0037 882 | Epoch [2/10], Iter [1330] Loss: 0.0039 883 | Epoch [2/10], Iter [1340] Loss: 0.0040 884 | Epoch [2/10], Iter [1350] Loss: 0.0039 885 | Epoch [2/10], Iter [1360] Loss: 0.0031 886 | Epoch [2/10], Iter [1370] Loss: 0.0035 887 | Epoch [2/10], Iter [1380] Loss: 0.0027 888 | Epoch [2/10], Iter [1390] Loss: 0.0038 889 | Epoch [2/10], Iter [1400] Loss: 0.0052 890 | Epoch [2/10], Iter [1410] Loss: 0.0033 891 | Epoch [2/10], Iter [1420] Loss: 0.0036 892 | Epoch [2/10], Iter [1430] Loss: 0.0045 893 | Epoch [2/10], Iter [1440] Loss: 0.0034 894 | Epoch [2/10], Iter [1450] Loss: 0.0034 895 | Epoch [2/10], Iter [1460] Loss: 0.0041 896 | Epoch [2/10], Iter [1470] Loss: 0.0051 897 | Epoch [2/10], Iter [1480] Loss: 0.0036 898 | Epoch [2/10], Iter [1490] Loss: 0.0049 899 | Epoch [2/10], Iter [1500] Loss: 0.0031 900 | Epoch [2/10], Iter [1510] Loss: 0.0046 901 | Epoch [2/10], Iter [1520] Loss: 0.0042 902 | Epoch [2/10], Iter [1530] Loss: 0.0041 903 | Epoch [2/10], Iter [1540] Loss: 0.0035 904 | Epoch [2/10], Iter [1550] Loss: 0.0042 905 | Epoch [2/10], Iter [1560] Loss: 0.0038 906 | Epoch [2/10], Iter [1570] Loss: 0.0050 907 | Epoch [2/10], Iter [1580] Loss: 0.0034 908 | Epoch [2/10], Iter [1590] Loss: 0.0045 909 | Epoch [2/10], Iter [1600] Loss: 0.0036 910 | Epoch [2/10], Iter [1610] Loss: 0.0042 911 | Epoch [2/10], Iter [1620] Loss: 0.0030 912 | Epoch [2/10], Iter [1630] Loss: 0.0045 913 | Epoch [2/10], Iter [1640] Loss: 0.0042 914 | Epoch [2/10], Iter [1650] Loss: 0.0045 915 | Epoch [2/10], Iter [1660] Loss: 0.0032 916 | Epoch [2/10], Iter [1670] Loss: 0.0050 917 | Epoch [2/10], Iter [1680] Loss: 0.0037 918 | Epoch [2/10], Iter [1690] Loss: 0.0041 919 | Epoch [2/10], Iter [1700] Loss: 0.0038 920 | Epoch [2/10], Iter [1710] Loss: 0.0032 921 | Epoch [2/10], Iter [1720] Loss: 0.0049 922 | Epoch [2/10], Iter [1730] Loss: 0.0045 923 | Epoch [2/10], Iter [1740] Loss: 0.0063 924 | Epoch [2/10], Iter [1750] Loss: 0.0047 925 | Epoch [2/10], Iter [1760] Loss: 0.0038 926 | Epoch [2/10], Iter [1770] Loss: 0.0039 927 | Epoch [2/10], Iter [1780] Loss: 0.0036 928 | Epoch [2/10], Iter [1790] Loss: 0.0043 929 | Epoch [2/10], Iter [1800] Loss: 0.0052 930 | Epoch [2/10], Iter [1810] Loss: 0.0039 931 | Epoch [2/10], Iter [1820] Loss: 0.0036 932 | Epoch [2/10], Iter [1830] Loss: 0.0042 933 | Epoch [2/10], Iter [1840] Loss: 0.0033 934 | Epoch [2/10], Iter [1850] Loss: 0.0047 935 | Epoch [2/10], Iter [1860] Loss: 0.0042 936 | Epoch [2/10], Iter [1870] Loss: 0.0044 937 | Epoch [2/10], Iter [1880] Loss: 0.0039 938 | Epoch [2/10], Iter [1890] Loss: 0.0039 939 | Epoch [2/10], Iter [1900] Loss: 0.0044 940 | Epoch [2/10], Iter [1910] Loss: 0.0045 941 | Epoch [2/10], Iter [1920] Loss: 0.0041 942 | Epoch [2/10], Iter [1930] Loss: 0.0048 943 | Epoch [2/10], Iter [1940] Loss: 0.0038 944 | Epoch [2/10], Iter [1950] Loss: 0.0037 945 | Epoch [2/10], Iter [1960] Loss: 0.0037 946 | Epoch [2/10], Iter [1970] Loss: 0.0038 947 | Epoch [2/10], Iter [1980] Loss: 0.0058 948 | Epoch [2/10], Iter [1990] Loss: 0.0037 949 | Epoch [2/10], Iter [2000] Loss: 0.0031 950 | Epoch [2/10], Iter [2010] Loss: 0.0038 951 | Epoch [2/10], Iter [2020] Loss: 0.0033 952 | Epoch [2/10], Iter [2030] Loss: 0.0032 953 | Epoch [2/10], Iter [2040] Loss: 0.0041 954 | Epoch [2/10], Iter [2050] Loss: 0.0037 955 | Epoch [2/10], Iter [2060] Loss: 0.0042 956 | Epoch [2/10], Iter [2070] Loss: 0.0041 957 | Epoch [2/10], Iter [2080] Loss: 0.0040 958 | Epoch [2/10], Iter [2090] Loss: 0.0045 959 | Epoch [2/10], Iter [2100] Loss: 0.0046 960 | Epoch [2/10], Iter [2110] Loss: 0.0042 961 | Epoch [2/10], Iter [2120] Loss: 0.0036 962 | Epoch [2/10], Iter [2130] Loss: 0.0036 963 | Epoch [2/10], Iter [2140] Loss: 0.0037 964 | Epoch [2/10], Iter [2150] Loss: 0.0033 965 | Epoch [2/10], Iter [2160] Loss: 0.0039 966 | Epoch [2/10], Iter [2170] Loss: 0.0042 967 | Epoch [2/10], Iter [2180] Loss: 0.0041 968 | Epoch [2/10], Iter [2190] Loss: 0.0050 969 | Epoch [2/10], Iter [2200] Loss: 0.0033 970 | Epoch [2/10], Iter [2210] Loss: 0.0035 971 | Epoch [2/10], Iter [2220] Loss: 0.0029 972 | Epoch [2/10], Iter [2230] Loss: 0.0036 973 | Epoch [2/10], Iter [2240] Loss: 0.0032 974 | Epoch [2/10], Iter [2250] Loss: 0.0039 975 | Epoch [2/10], Iter [2260] Loss: 0.0039 976 | Epoch [2/10], Iter [2270] Loss: 0.0034 977 | Epoch [2/10], Iter [2280] Loss: 0.0030 978 | Epoch [2/10], Iter [2290] Loss: 0.0049 979 | Epoch [2/10], Iter [2300] Loss: 0.0031 980 | Epoch [2/10], Iter [2310] Loss: 0.0029 981 | Epoch [2/10], Iter [2320] Loss: 0.0030 982 | Epoch [2/10], Iter [2330] Loss: 0.0040 983 | Epoch [2/10], Iter [2340] Loss: 0.0040 984 | Epoch [2/10], Iter [2350] Loss: 0.0031 985 | Epoch [2/10], Iter [2360] Loss: 0.0044 986 | Epoch [2/10], Iter [2370] Loss: 0.0033 987 | Epoch [2/10], Iter [2380] Loss: 0.0038 988 | Epoch [2/10], Iter [2390] Loss: 0.0032 989 | Epoch [2/10], Iter [2400] Loss: 0.0035 990 | Epoch [2/10], Iter [2410] Loss: 0.0037 991 | Epoch [2/10], Iter [2420] Loss: 0.0032 992 | Epoch [2/10], Iter [2430] Loss: 0.0035 993 | Epoch [2/10], Iter [2440] Loss: 0.0034 994 | Epoch [2/10], Iter [2450] Loss: 0.0030 995 | Epoch [2/10], Iter [2460] Loss: 0.0025 996 | Epoch [2/10], Iter [2470] Loss: 0.0034 997 | Epoch [2/10], Iter [2480] Loss: 0.0038 998 | Epoch [2/10], Iter [2490] Loss: 0.0044 999 | Epoch [2/10], Iter [2500] Loss: 0.0036 1000 | Epoch [2/10], Iter [2510] Loss: 0.0035 1001 | Epoch [2/10], Iter [2520] Loss: 0.0042 1002 | Epoch [2/10], Iter [2530] Loss: 0.0056 1003 | Epoch [2/10], Iter [2540] Loss: 0.0038 1004 | Epoch [2/10], Iter [2550] Loss: 0.0042 1005 | Epoch [2/10], Iter [2560] Loss: 0.0035 1006 | Epoch [2/10], Iter [2570] Loss: 0.0031 1007 | Epoch [2/10], Iter [2580] Loss: 0.0042 1008 | Epoch [2/10], Iter [2590] Loss: 0.0033 1009 | Epoch [2/10], Iter [2600] Loss: 0.0036 1010 | Epoch [2/10], Iter [2610] Loss: 0.0041 1011 | Epoch [2/10], Iter [2620] Loss: 0.0035 1012 | Epoch [2/10], Iter [2630] Loss: 0.0038 1013 | Epoch [2/10], Iter [2640] Loss: 0.0044 1014 | Epoch [2/10], Iter [2650] Loss: 0.0044 1015 | Epoch [2/10], Iter [2660] Loss: 0.0043 1016 | Epoch [2/10], Iter [2670] Loss: 0.0039 1017 | Epoch [2/10], Iter [2680] Loss: 0.0028 1018 | Epoch [2/10], Iter [2690] Loss: 0.0037 1019 | Epoch [2/10], Iter [2700] Loss: 0.0036 1020 | Epoch [2/10], Iter [2710] Loss: 0.0034 1021 | Epoch [2/10], Iter [2720] Loss: 0.0037 1022 | Epoch [2/10], Iter [2730] Loss: 0.0034 1023 | Epoch [2/10], Iter [2740] Loss: 0.0035 1024 | Epoch [2/10], Iter [2750] Loss: 0.0037 1025 | Epoch [2/10], Iter [2760] Loss: 0.0037 1026 | Epoch [2/10], Iter [2770] Loss: 0.0037 1027 | Epoch [2/10], Iter [2780] Loss: 0.0031 1028 | Epoch [2/10], Iter [2790] Loss: 0.0023 1029 | Epoch [2/10], Iter [2800] Loss: 0.0040 1030 | Epoch [2/10], Iter [2810] Loss: 0.0036 1031 | Epoch [2/10], Iter [2820] Loss: 0.0042 1032 | Epoch [2/10], Iter [2830] Loss: 0.0039 1033 | Epoch [2/10], Iter [2840] Loss: 0.0041 1034 | Epoch [2/10], Iter [2850] Loss: 0.0037 1035 | Epoch [2/10], Iter [2860] Loss: 0.0034 1036 | Epoch [2/10], Iter [2870] Loss: 0.0035 1037 | Epoch [2/10], Iter [2880] Loss: 0.0043 1038 | Epoch [2/10], Iter [2890] Loss: 0.0037 1039 | Epoch [2/10], Iter [2900] Loss: 0.0041 1040 | Epoch [2/10], Iter [2910] Loss: 0.0029 1041 | Epoch [2/10], Iter [2920] Loss: 0.0030 1042 | Epoch [2/10], Iter [2930] Loss: 0.0030 1043 | Epoch [2/10], Iter [2940] Loss: 0.0028 1044 | Epoch [2/10], Iter [2950] Loss: 0.0035 1045 | Epoch [2/10], Iter [2960] Loss: 0.0029 1046 | Epoch [2/10], Iter [2970] Loss: 0.0036 1047 | Epoch [2/10], Iter [2980] Loss: 0.0036 1048 | Epoch [2/10], Iter [2990] Loss: 0.0041 1049 | Epoch [2/10], Iter [3000] Loss: 0.0040 1050 | Epoch [2/10], Iter [3010] Loss: 0.0032 1051 | Epoch [2/10], Iter [3020] Loss: 0.0035 1052 | Epoch [2/10], Iter [3030] Loss: 0.0035 1053 | Epoch [2/10], Iter [3040] Loss: 0.0036 1054 | Epoch [2/10], Iter [3050] Loss: 0.0035 1055 | Epoch [2/10], Iter [3060] Loss: 0.0036 1056 | Epoch [2/10], Iter [3070] Loss: 0.0034 1057 | Epoch [2/10], Iter [3080] Loss: 0.0028 1058 | Epoch [2/10], Iter [3090] Loss: 0.0043 1059 | Epoch [2/10], Iter [3100] Loss: 0.0034 1060 | Epoch [2/10], Iter [3110] Loss: 0.0038 1061 | Epoch [2/10], Iter [3120] Loss: 0.0035 1062 | Epoch [2/10], Iter [3130] Loss: 0.0032 1063 | Epoch [2/10], Iter [3140] Loss: 0.0032 1064 | Epoch [2/10], Iter [3150] Loss: 0.0045 1065 | Epoch [2/10], Iter [3160] Loss: 0.0029 1066 | Epoch [2/10], Iter [3170] Loss: 0.0041 1067 | Epoch [2/10], Iter [3180] Loss: 0.0036 1068 | Epoch [2/10], Iter [3190] Loss: 0.0041 1069 | Epoch [2/10], Iter [3200] Loss: 0.0032 1070 | Epoch [2/10], Iter [3210] Loss: 0.0033 1071 | Epoch [2/10], Iter [3220] Loss: 0.0041 1072 | Epoch [2/10], Iter [3230] Loss: 0.0039 1073 | Epoch [2/10], Iter [3240] Loss: 0.0027 1074 | Epoch [2/10], Iter [3250] Loss: 0.0029 1075 | Epoch [2/10], Iter [3260] Loss: 0.0042 1076 | Epoch [2/10], Iter [3270] Loss: 0.0031 1077 | Epoch [2/10], Iter [3280] Loss: 0.0034 1078 | Epoch [2/10], Iter [3290] Loss: 0.0035 1079 | Epoch [2/10], Iter [3300] Loss: 0.0037 1080 | Epoch [2/10], Iter [3310] Loss: 0.0035 1081 | Epoch [2/10], Iter [3320] Loss: 0.0038 1082 | Epoch [2/10], Iter [3330] Loss: 0.0036 1083 | Epoch [2/10], Iter [3340] Loss: 0.0041 1084 | Epoch [2/10], Iter [3350] Loss: 0.0034 1085 | Epoch [2/10], Iter [3360] Loss: 0.0031 1086 | Epoch [2/10], Iter [3370] Loss: 0.0031 1087 | Epoch [2/10], Iter [3380] Loss: 0.0032 1088 | Epoch [2/10], Iter [3390] Loss: 0.0028 1089 | Epoch [2/10], Iter [3400] Loss: 0.0028 1090 | Epoch [2/10], Iter [3410] Loss: 0.0030 1091 | Epoch [2/10], Iter [3420] Loss: 0.0041 1092 | Epoch [2/10], Iter [3430] Loss: 0.0032 1093 | Epoch [2/10], Iter [3440] Loss: 0.0028 1094 | Epoch [2/10], Iter [3450] Loss: 0.0034 1095 | Epoch [2/10], Iter [3460] Loss: 0.0038 1096 | Epoch [2/10], Iter [3470] Loss: 0.0038 1097 | Epoch [2/10], Iter [3480] Loss: 0.0032 1098 | Epoch [2/10], Iter [3490] Loss: 0.0030 1099 | Epoch [2/10], Iter [3500] Loss: 0.0045 1100 | Epoch [2/10], Iter [3510] Loss: 0.0030 1101 | Epoch [2/10], Iter [3520] Loss: 0.0031 1102 | Epoch [2/10], Iter [3530] Loss: 0.0039 1103 | Epoch [2/10], Iter [3540] Loss: 0.0032 1104 | Epoch [2/10], Iter [3550] Loss: 0.0034 1105 | Epoch [2/10], Iter [3560] Loss: 0.0032 1106 | Epoch [2/10], Iter [3570] Loss: 0.0036 1107 | Epoch [2/10], Iter [3580] Loss: 0.0028 1108 | Epoch [2/10], Iter [3590] Loss: 0.0028 1109 | Epoch [2/10], Iter [3600] Loss: 0.0036 1110 | Epoch [2/10], Iter [3610] Loss: 0.0026 1111 | Epoch [2/10], Iter [3620] Loss: 0.0035 1112 | Epoch [2/10], Iter [3630] Loss: 0.0045 1113 | Epoch [2/10], Iter [3640] Loss: 0.0031 1114 | Epoch [2/10], Iter [3650] Loss: 0.0038 1115 | Epoch [2/10], Iter [3660] Loss: 0.0034 1116 | Epoch [2/10], Iter [3670] Loss: 0.0036 1117 | Epoch [2/10], Iter [3680] Loss: 0.0031 1118 | Epoch [2/10], Iter [3690] Loss: 0.0044 1119 | Epoch [2/10], Iter [3700] Loss: 0.0041 1120 | Epoch [2/10], Iter [3710] Loss: 0.0039 1121 | Epoch [2/10], Iter [3720] Loss: 0.0024 1122 | Epoch [2/10], Iter [3730] Loss: 0.0029 1123 | Epoch [2/10], Iter [3740] Loss: 0.0039 1124 | Epoch [2/10], Iter [3750] Loss: 0.0025 1125 | Epoch [2/10], Iter [3760] Loss: 0.0038 1126 | Epoch [2/10], Iter [3770] Loss: 0.0033 1127 | Epoch [2/10], Iter [3780] Loss: 0.0027 1128 | Epoch [2/10], Iter [3790] Loss: 0.0032 1129 | Epoch [2/10], Iter [3800] Loss: 0.0037 1130 | Epoch [2/10], Iter [3810] Loss: 0.0032 1131 | Epoch [2/10], Iter [3820] Loss: 0.0037 1132 | Epoch [2/10], Iter [3830] Loss: 0.0038 1133 | Epoch [2/10], Iter [3840] Loss: 0.0029 1134 | Epoch [2/10], Iter [3850] Loss: 0.0032 1135 | Epoch [2/10], Iter [3860] Loss: 0.0030 1136 | Epoch [2/10], Iter [3870] Loss: 0.0028 1137 | Epoch [2/10], Iter [3880] Loss: 0.0048 1138 | Epoch [2/10], Iter [3890] Loss: 0.0031 1139 | Epoch [2/10], Iter [3900] Loss: 0.0036 1140 | Epoch [2/10], Iter [3910] Loss: 0.0042 1141 | Epoch [2/10], Iter [3920] Loss: 0.0041 1142 | Epoch [2/10], Iter [3930] Loss: 0.0030 1143 | Epoch [2/10], Iter [3940] Loss: 0.0034 1144 | Epoch [2/10], Iter [3950] Loss: 0.0036 1145 | Epoch [2/10], Iter [3960] Loss: 0.0032 1146 | Epoch [2/10], Iter [3970] Loss: 0.0033 1147 | Epoch [2/10], Iter [3980] Loss: 0.0028 1148 | Epoch [2/10], Iter [3990] Loss: 0.0037 1149 | Epoch [2/10], Iter [4000] Loss: 0.0030 1150 | Epoch [2/10], Iter [4010] Loss: 0.0037 1151 | Epoch [2/10], Iter [4020] Loss: 0.0036 1152 | Epoch [2/10], Iter [4030] Loss: 0.0030 1153 | Epoch [2/10], Iter [4040] Loss: 0.0032 1154 | Epoch [2/10], Iter [4050] Loss: 0.0034 1155 | Epoch [2/10], Iter [4060] Loss: 0.0039 1156 | Epoch [2/10], Iter [4070] Loss: 0.0042 1157 | Epoch [2/10], Iter [4080] Loss: 0.0031 1158 | Epoch [2/10], Iter [4090] Loss: 0.0029 1159 | Epoch [2/10], Iter [4100] Loss: 0.0036 1160 | Epoch [2/10], Iter [4110] Loss: 0.0032 1161 | Epoch [2/10], Iter [4120] Loss: 0.0031 1162 | Epoch [2/10], Iter [4130] Loss: 0.0029 1163 | Epoch [2/10], Iter [4140] Loss: 0.0031 1164 | Epoch [2/10], Iter [4150] Loss: 0.0033 1165 | Epoch [2/10], Iter [4160] Loss: 0.0034 1166 | Epoch [2/10], Iter [4170] Loss: 0.0039 1167 | Epoch [2/10], Iter [4180] Loss: 0.0030 1168 | Epoch [2/10], Iter [4190] Loss: 0.0037 1169 | Epoch [2/10], Iter [4200] Loss: 0.0034 1170 | Epoch [2/10], Iter [4210] Loss: 0.0039 1171 | Epoch [2/10], Iter [4220] Loss: 0.0033 1172 | Epoch [2/10], Iter [4230] Loss: 0.0035 1173 | Epoch [2/10], Iter [4240] Loss: 0.0033 1174 | Epoch [2/10], Iter [4250] Loss: 0.0039 1175 | Epoch [2/10], Iter [4260] Loss: 0.0043 1176 | Epoch [2/10], Iter [4270] Loss: 0.0037 1177 | Epoch [2/10], Iter [4280] Loss: 0.0034 1178 | Epoch [2/10], Iter [4290] Loss: 0.0027 1179 | Epoch [2/10], Iter [4300] Loss: 0.0033 1180 | Epoch [2/10], Iter [4310] Loss: 0.0030 1181 | Epoch [2/10], Iter [4320] Loss: 0.0027 1182 | Epoch [2/10], Iter [4330] Loss: 0.0028 1183 | Epoch [2/10], Iter [4340] Loss: 0.0031 1184 | Epoch [2/10], Iter [4350] Loss: 0.0027 1185 | Epoch [2/10], Iter [4360] Loss: 0.0035 1186 | Epoch [2/10], Iter [4370] Loss: 0.0031 1187 | Epoch [2/10], Iter [4380] Loss: 0.0044 1188 | Epoch [2/10], Iter [4390] Loss: 0.0041 1189 | Epoch [2/10], Iter [4400] Loss: 0.0041 1190 | Epoch [2/10], Iter [4410] Loss: 0.0027 1191 | Epoch [2/10], Iter [4420] Loss: 0.0027 1192 | Epoch [2/10], Iter [4430] Loss: 0.0030 1193 | Epoch [2/10], Iter [4440] Loss: 0.0037 1194 | Epoch [2/10], Iter [4450] Loss: 0.0047 1195 | Epoch [2/10], Iter [4460] Loss: 0.0031 1196 | Epoch [2/10], Iter [4470] Loss: 0.0039 1197 | Epoch [2/10], Iter [4480] Loss: 0.0037 1198 | Epoch [2/10], Iter [4490] Loss: 0.0030 1199 | Epoch [2/10], Iter [4500] Loss: 0.0043 1200 | Epoch [2/10], Iter [4510] Loss: 0.0034 1201 | Epoch [2/10], Iter [4520] Loss: 0.0033 1202 | Epoch [2/10], Iter [4530] Loss: 0.0031 1203 | Epoch [2/10], Iter [4540] Loss: 0.0032 1204 | Epoch [2/10], Iter [4550] Loss: 0.0030 1205 | Epoch [2/10], Iter [4560] Loss: 0.0035 1206 | Epoch [2/10], Iter [4570] Loss: 0.0027 1207 | Epoch [2/10], Iter [4580] Loss: 0.0030 1208 | Epoch [2/10], Iter [4590] Loss: 0.0036 1209 | Epoch [2/10], Iter [4600] Loss: 0.0033 1210 | Epoch [2/10], Iter [4610] Loss: 0.0032 1211 | Epoch [2/10], Iter [4620] Loss: 0.0036 1212 | Epoch [2/10], Iter [4630] Loss: 0.0037 1213 | Epoch [2/10], Iter [4640] Loss: 0.0025 1214 | Epoch [2/10], Iter [4650] Loss: 0.0039 1215 | Epoch [2/10], Iter [4660] Loss: 0.0029 1216 | Epoch [2/10], Iter [4670] Loss: 0.0032 1217 | Epoch [2/10], Iter [4680] Loss: 0.0035 1218 | Epoch [2/10], Iter [4690] Loss: 0.0034 1219 | Epoch [2/10], Iter [4700] Loss: 0.0025 1220 | Epoch [2/10], Iter [4710] Loss: 0.0030 1221 | Epoch [2/10], Iter [4720] Loss: 0.0031 1222 | Epoch [2/10], Iter [4730] Loss: 0.0033 1223 | Epoch [2/10], Iter [4740] Loss: 0.0035 1224 | Epoch [2/10], Iter [4750] Loss: 0.0029 1225 | Epoch [2/10], Iter [4760] Loss: 0.0033 1226 | Epoch [2/10], Iter [4770] Loss: 0.0031 1227 | Epoch [2/10], Iter [4780] Loss: 0.0031 1228 | Epoch [2/10], Iter [4790] Loss: 0.0036 1229 | Epoch [2/10], Iter [4800] Loss: 0.0046 1230 | Epoch [2/10], Iter [4810] Loss: 0.0030 1231 | Epoch [2/10], Iter [4820] Loss: 0.0045 1232 | Epoch [2/10], Iter [4830] Loss: 0.0032 1233 | Epoch [2/10], Iter [4840] Loss: 0.0033 1234 | Epoch [2/10], Iter [4850] Loss: 0.0046 1235 | Epoch [2/10], Iter [4860] Loss: 0.0029 1236 | Epoch [2/10], Iter [4870] Loss: 0.0034 1237 | Epoch [2/10], Iter [4880] Loss: 0.0029 1238 | Epoch [2/10], Iter [4890] Loss: 0.0032 1239 | Epoch [2/10], Iter [4900] Loss: 0.0032 1240 | Epoch [2/10], Iter [4910] Loss: 0.0046 1241 | Epoch [2/10], Iter [4920] Loss: 0.0036 1242 | Epoch [2/10], Iter [4930] Loss: 0.0037 1243 | Epoch [2/10], Iter [4940] Loss: 0.0039 1244 | Epoch [2/10], Iter [4950] Loss: 0.0046 1245 | Epoch [2/10], Iter [4960] Loss: 0.0027 1246 | Epoch [2/10], Iter [4970] Loss: 0.0036 1247 | Epoch [2/10], Iter [4980] Loss: 0.0029 1248 | Epoch [2/10], Iter [4990] Loss: 0.0029 1249 | Epoch [2/10], Iter [5000] Loss: 0.0027 1250 | Epoch [2/10], Iter [5010] Loss: 0.0030 1251 | Epoch [2/10], Iter [5020] Loss: 0.0027 1252 | Epoch [2/10], Iter [5030] Loss: 0.0038 1253 | Epoch [2/10], Iter [5040] Loss: 0.0032 1254 | Epoch [2/10], Iter [5050] Loss: 0.0035 1255 | Epoch [2/10], Iter [5060] Loss: 0.0034 1256 | Epoch [2/10], Iter [5070] Loss: 0.0034 1257 | Epoch [2/10], Iter [5080] Loss: 0.0038 1258 | Epoch [2/10], Iter [5090] Loss: 0.0034 1259 | Epoch [2/10], Iter [5100] Loss: 0.0028 1260 | Epoch [2/10], Iter [5110] Loss: 0.0034 1261 | Epoch [2/10], Iter [5120] Loss: 0.0031 1262 | Epoch [2/10], Iter [5130] Loss: 0.0030 1263 | Epoch [2/10], Iter [5140] Loss: 0.0034 1264 | Epoch [2/10], Iter [5150] Loss: 0.0041 1265 | Epoch [2/10], Iter [5160] Loss: 0.0039 1266 | Epoch [2/10], Iter [5170] Loss: 0.0035 1267 | Epoch [2/10], Iter [5180] Loss: 0.0037 1268 | Epoch [2/10], Iter [5190] Loss: 0.0038 1269 | Epoch [2/10], Iter [5200] Loss: 0.0028 1270 | Epoch [2/10], Iter [5210] Loss: 0.0030 1271 | Epoch [2/10], Iter [5220] Loss: 0.0037 1272 | Epoch [2/10], Iter [5230] Loss: 0.0025 1273 | Epoch [2/10], Iter [5240] Loss: 0.0031 1274 | Epoch [2/10], Iter [5250] Loss: 0.0030 1275 | Epoch [2/10], Iter [5260] Loss: 0.0032 1276 | Epoch [2/10], Iter [5270] Loss: 0.0035 1277 | Epoch [2/10], Iter [5280] Loss: 0.0039 1278 | Epoch [2/10], Iter [5290] Loss: 0.0041 1279 | Epoch [2/10], Iter [5300] Loss: 0.0036 1280 | Epoch [2/10], Iter [5310] Loss: 0.0033 1281 | Epoch [2/10], Iter [5320] Loss: 0.0032 1282 | Epoch [2/10], Iter [5330] Loss: 0.0027 1283 | Epoch [2/10], Iter [5340] Loss: 0.0026 1284 | Epoch [2/10], Iter [5350] Loss: 0.0042 1285 | Epoch [2/10], Iter [5360] Loss: 0.0030 1286 | Epoch [2/10], Iter [5370] Loss: 0.0043 1287 | Epoch [2/10], Iter [5380] Loss: 0.0033 1288 | Epoch [2/10], Iter [5390] Loss: 0.0029 1289 | Epoch [2/10], Iter [5400] Loss: 0.0032 1290 | Epoch [2/10], Iter [5410] Loss: 0.0034 1291 | Epoch [2/10], Iter [5420] Loss: 0.0033 1292 | Epoch [2/10], Iter [5430] Loss: 0.0029 1293 | Epoch [2/10], Iter [5440] Loss: 0.0026 1294 | Epoch [2/10], Iter [5450] Loss: 0.0027 1295 | Epoch [2/10], Iter [5460] Loss: 0.0040 1296 | Epoch [2/10], Iter [5470] Loss: 0.0031 1297 | Epoch [2/10], Iter [5480] Loss: 0.0026 1298 | Epoch [2/10], Iter [5490] Loss: 0.0041 1299 | Epoch [2/10], Iter [5500] Loss: 0.0026 1300 | Epoch [2/10], Iter [5510] Loss: 0.0030 1301 | Epoch [2/10], Iter [5520] Loss: 0.0027 1302 | Epoch [2/10], Iter [5530] Loss: 0.0027 1303 | Epoch [2/10], Iter [5540] Loss: 0.0033 1304 | Epoch [2/10], Iter [5550] Loss: 0.0028 1305 | Epoch [2/10], Iter [5560] Loss: 0.0029 1306 | Epoch [2/10], Iter [5570] Loss: 0.0033 1307 | Epoch [2/10], Iter [5580] Loss: 0.0029 1308 | Epoch [2/10], Iter [5590] Loss: 0.0028 1309 | Epoch [2/10], Iter [5600] Loss: 0.0030 1310 | Epoch [2/10], Iter [5610] Loss: 0.0037 1311 | Epoch [2/10], Iter [5620] Loss: 0.0031 1312 | Epoch [2/10], Iter [5630] Loss: 0.0027 1313 | Epoch [2/10], Iter [5640] Loss: 0.0036 1314 | Epoch [2/10], Iter [5650] Loss: 0.0027 1315 | Epoch [2/10], Iter [5660] Loss: 0.0033 1316 | Epoch [2/10], Iter [5670] Loss: 0.0034 1317 | Epoch [2/10], Iter [5680] Loss: 0.0034 1318 | Epoch [2/10], Iter [5690] Loss: 0.0029 1319 | Epoch [2/10], Iter [5700] Loss: 0.0036 1320 | Epoch [2/10], Iter [5710] Loss: 0.0026 1321 | Epoch [2/10], Iter [5720] Loss: 0.0028 1322 | Epoch [2/10], Iter [5730] Loss: 0.0027 1323 | Epoch [2/10], Iter [5740] Loss: 0.0023 1324 | Epoch [2/10], Iter [5750] Loss: 0.0033 1325 | Epoch [2/10], Iter [5760] Loss: 0.0037 1326 | Epoch [2/10], Iter [5770] Loss: 0.0028 1327 | Epoch [2/10], Iter [5780] Loss: 0.0039 1328 | Epoch [2/10], Iter [5790] Loss: 0.0037 1329 | Epoch [2/10], Iter [5800] Loss: 0.0030 1330 | Epoch [2/10], Iter [5810] Loss: 0.0030 1331 | Epoch [2/10], Iter [5820] Loss: 0.0039 1332 | Epoch [2/10], Iter [5830] Loss: 0.0026 1333 | Epoch [2/10], Iter [5840] Loss: 0.0034 1334 | Epoch [2/10], Iter [5850] Loss: 0.0031 1335 | Epoch [2/10], Iter [5860] Loss: 0.0028 1336 | Epoch [2/10], Iter [5870] Loss: 0.0028 1337 | Epoch [2/10], Iter [5880] Loss: 0.0025 1338 | Epoch [2/10], Iter [5890] Loss: 0.0026 1339 | Epoch [2/10], Iter [5900] Loss: 0.0031 1340 | Epoch [2/10], Iter [5910] Loss: 0.0028 1341 | Epoch [2/10], Iter [5920] Loss: 0.0027 1342 | Epoch [2/10], Iter [5930] Loss: 0.0027 1343 | Epoch [2/10], Iter [5940] Loss: 0.0030 1344 | Epoch [2/10], Iter [5950] Loss: 0.0026 1345 | Epoch [2/10], Iter [5960] Loss: 0.0038 1346 | Epoch [2/10], Iter [5970] Loss: 0.0032 1347 | Epoch [2/10], Iter [5980] Loss: 0.0033 1348 | Epoch [2/10], Iter [5990] Loss: 0.0029 1349 | Epoch [2/10], Iter [6000] Loss: 0.0034 1350 | Epoch [2/10], Iter [6010] Loss: 0.0028 1351 | Epoch [2/10], Iter [6020] Loss: 0.0033 1352 | Epoch [2/10], Iter [6030] Loss: 0.0025 1353 | Epoch [2/10], Iter [6040] Loss: 0.0031 1354 | Epoch [2/10], Iter [6050] Loss: 0.0033 1355 | Epoch [2/10], Iter [6060] Loss: 0.0031 1356 | Epoch [2/10], Iter [6070] Loss: 0.0031 1357 | Epoch [2/10], Iter [6080] Loss: 0.0043 1358 | Epoch [2/10], Iter [6090] Loss: 0.0030 1359 | Epoch [2/10], Iter [6100] Loss: 0.0035 1360 | Epoch [2/10], Iter [6110] Loss: 0.0033 1361 | Epoch [2/10], Iter [6120] Loss: 0.0037 1362 | Epoch [2/10], Iter [6130] Loss: 0.0035 1363 | Epoch [2/10], Iter [6140] Loss: 0.0028 1364 | Epoch [2/10], Iter [6150] Loss: 0.0030 1365 | Epoch [2/10], Iter [6160] Loss: 0.0035 1366 | Epoch [2/10], Iter [6170] Loss: 0.0026 1367 | Epoch [2/10], Iter [6180] Loss: 0.0034 1368 | Epoch [2/10], Iter [6190] Loss: 0.0038 1369 | Epoch [2/10], Iter [6200] Loss: 0.0027 1370 | Epoch [2/10], Iter [6210] Loss: 0.0035 1371 | Epoch [2/10], Iter [6220] Loss: 0.0032 1372 | Epoch [2/10], Iter [6230] Loss: 0.0034 1373 | Epoch [2/10], Iter [6240] Loss: 0.0033 1374 | Epoch [2/10], Iter [6250] Loss: 0.0036 1375 | Epoch [2/10], Iter [6260] Loss: 0.0031 1376 | Epoch [2/10], Iter [6270] Loss: 0.0030 1377 | Epoch [2/10], Iter [6280] Loss: 0.0033 1378 | Epoch [2/10], Iter [6290] Loss: 0.0032 1379 | Epoch [2/10], Iter [6300] Loss: 0.0033 1380 | Epoch [2/10], Iter [6310] Loss: 0.0032 1381 | Epoch [2/10], Iter [6320] Loss: 0.0037 1382 | Epoch [2/10], Iter [6330] Loss: 0.0034 1383 | Epoch [2/10], Iter [6340] Loss: 0.0020 1384 | Epoch [2/10], Iter [6350] Loss: 0.0028 1385 | Epoch [2/10], Iter [6360] Loss: 0.0038 1386 | Epoch [2/10], Iter [6370] Loss: 0.0025 1387 | Epoch [2/10], Iter [6380] Loss: 0.0032 1388 | Epoch [2/10], Iter [6390] Loss: 0.0024 1389 | Epoch [2/10], Iter [6400] Loss: 0.0028 1390 | Epoch [2/10], Iter [6410] Loss: 0.0025 1391 | Epoch [2/10], Iter [6420] Loss: 0.0033 1392 | Epoch [2/10], Iter [6430] Loss: 0.0038 1393 | Epoch [2/10], Iter [6440] Loss: 0.0032 1394 | Epoch [2/10], Iter [6450] Loss: 0.0035 1395 | Epoch [2/10], Iter [6460] Loss: 0.0035 1396 | Epoch [2/10], Iter [6470] Loss: 0.0036 1397 | Epoch [2/10], Iter [6480] Loss: 0.0038 1398 | Epoch [2/10], Iter [6490] Loss: 0.0029 1399 | Epoch [2/10], Iter [6500] Loss: 0.0032 1400 | Epoch [2/10], Iter [6510] Loss: 0.0024 1401 | Epoch [2/10], Iter [6520] Loss: 0.0027 1402 | Epoch [2/10], Iter [6530] Loss: 0.0031 1403 | Epoch [2/10], Iter [6540] Loss: 0.0028 1404 | Epoch [2/10], Iter [6550] Loss: 0.0022 1405 | Epoch [2/10], Iter [6560] Loss: 0.0031 1406 | Epoch [2/10], Iter [6570] Loss: 0.0026 1407 | Epoch [2/10], Iter [6580] Loss: 0.0025 1408 | Epoch [2/10], Iter [6590] Loss: 0.0032 1409 | Epoch [2/10], Iter [6600] Loss: 0.0036 1410 | Epoch [2/10], Iter [6610] Loss: 0.0029 1411 | Epoch [2/10], Iter [6620] Loss: 0.0032 1412 | Epoch [2/10], Iter [6630] Loss: 0.0037 1413 | Epoch [2/10], Iter [6640] Loss: 0.0026 1414 | Epoch [2/10], Iter [6650] Loss: 0.0029 1415 | Epoch [2/10], Iter [6660] Loss: 0.0025 1416 | Epoch [2/10], Iter [6670] Loss: 0.0037 1417 | Epoch [2/10], Iter [6680] Loss: 0.0028 1418 | Epoch [2/10], Iter [6690] Loss: 0.0047 1419 | Epoch [2/10], Iter [6700] Loss: 0.0031 1420 | Epoch [2/10], Iter [6710] Loss: 0.0034 1421 | Epoch [2/10], Iter [6720] Loss: 0.0038 1422 | Epoch [2/10], Iter [6730] Loss: 0.0034 1423 | Epoch [2/10], Iter [6740] Loss: 0.0035 1424 | Epoch [2/10], Iter [6750] Loss: 0.0025 1425 | Epoch [2/10], Iter [6760] Loss: 0.0033 1426 | Epoch [2/10], Iter [6770] Loss: 0.0026 1427 | Epoch [2/10], Iter [6780] Loss: 0.0028 1428 | Epoch [2/10], Iter [6790] Loss: 0.0039 1429 | Epoch [2/10], Iter [6800] Loss: 0.0032 1430 | Epoch [2/10], Iter [6810] Loss: 0.0034 1431 | Epoch [2/10], Iter [6820] Loss: 0.0027 1432 | Epoch [2/10], Iter [6830] Loss: 0.0031 1433 | Epoch [2/10], Iter [6840] Loss: 0.0030 1434 | Epoch [2/10], Iter [6850] Loss: 0.0034 1435 | Epoch [2/10], Iter [6860] Loss: 0.0032 1436 | Epoch [2/10], Iter [6870] Loss: 0.0029 1437 | Epoch [2/10], Iter [6880] Loss: 0.0034 1438 | Epoch [2/10], Iter [6890] Loss: 0.0032 1439 | Epoch [2/10], Iter [6900] Loss: 0.0030 1440 | Epoch [2/10], Iter [6910] Loss: 0.0023 1441 | Epoch [2/10], Iter [6920] Loss: 0.0031 1442 | Epoch [2/10], Iter [6930] Loss: 0.0040 1443 | start loading. 1444 | load data success. 1445 | load pretrained model success 1446 | start training. 1447 | Traceback (most recent call last): 1448 | File "train3.py", line 60, in 1449 | tmp1, tmp2, tmp3 = model(image) 1450 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/module.py", line 357, in __call__ 1451 | result = self.forward(*input, **kwargs) 1452 | File "/home/achhe_ucdavis_edu/resnet-autoencoder/autoencoder3.py", line 198, in forward 1453 | tmp2 = self.decoder(tmp1, img.size()) 1454 | File "/home/achhe_ucdavis_edu/resnet-autoencoder/autoencoder3.py", line 188, in decoder 1455 | x = self.uplayer1(x) 1456 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/module.py", line 357, in __call__ 1457 | result = self.forward(*input, **kwargs) 1458 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/container.py", line 67, in forward 1459 | input = module(input) 1460 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/module.py", line 357, in __call__ 1461 | result = self.forward(*input, **kwargs) 1462 | File "/home/achhe_ucdavis_edu/resnet-autoencoder/autoencoder3.py", line 79, in forward 1463 | out = self.conv1(x) 1464 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/module.py", line 357, in __call__ 1465 | result = self.forward(*input, **kwargs) 1466 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/modules/conv.py", line 282, in forward 1467 | self.padding, self.dilation, self.groups) 1468 | File "/home/chongruo/miniconda2/envs/pytorch-python3.5/lib/python3.5/site-packages/torch/nn/functional.py", line 90, in conv2d 1469 | return f(input, weight, bias) 1470 | RuntimeError: Given groups=1, weight[512, 2048, 1, 1], so expected input[16, 512, 28, 28] to have 2048 channels, but got 512 channels instead 1471 | --------------------------------------------------------------------------------