Sākums Izpētīt Palīdzība
Pierakstīties
AIModelWatermark
/
classification-main
2
0
Atdalīts 0
Faili Problēmas 0 Izmaiņu pieprasījumi 0 Vikivietne
Koks: bf0ccfaab8
Atzari Tagi
classificati...
/
bash_run.sh

bash_run.sh 3.6 KB
Vēsture Neapstrādāts

12345678910111213141516171819202122232425262728293031 
  1. # 模型训练
    2. 

  2. python run.py --model Alexnet --input_size 32 --save_path ./checkpoints/Alexnet/clean/best.pt --save_path_last ./checkpoints/Alexnet/clean/last.pt --epoch 10 --data_path ./dataset --dataset_name CIFAR-10
    3. 

  3. # 嵌入白盒水印模型训练
    4. 

  4. python run.py --model Alexnet --input_size 32 --save_path ./checkpoints/Alexnet/white_box_embed/best.pt --save_path_last ./checkpoints/Alexnet/white_box_embed/last.pt --epoch 10 --data_path ./dataset --dataset_name CIFAR-10 --white_box_embed True
    5. 

  5. 

  6. # For 用于训练不同模型，以及保存相应的路径
    7. 

  7. # -------------------------------------------------------------------------------------------------------------------- #
    8. 

  8. python run.py --model 'resnet' --save_path './checkpoints/resnet/watermarking/best.pt' --save_path_last './checkpoints/resnet/watermarking/last.pt' --epoch 100
    9. 

  9. python run.py --model 'VGG19' --save_path './checkpoints/VGG19/watermarking/best.pt' --save_path_last './checkpoints/VGG19/watermarking/last.pt' --epoch 100
    10. 

  10. python run.py --model 'Alexnet' --input_size 112 --save_path './checkpoints/Alexnet/watermarking/best.pt' --save_path_last './checkpoints/Alexnet/watermarking/last.pt' --epoch 100
    11. 

  11. python run.py --model 'mobilenetv2' --save_path './checkpoints/mobilenetv2/watermarking/best.pt' --save_path_last './checkpoints/mobilenetv2/watermarking/last.pt' --epoch 100
    12. 

  12. python run.py --model 'GoogleNet' --input_size 32 --save_path './checkpoints/GoogleNet/watermarking/best.pt' --save_path_last './checkpoints/GoogleNet/watermarking/last.pt' --epoch 100
    13. 

  13. python run.py --model 'badnet' --input_size 32 --save_path './checkpoints/badnet/watermarking/best.pt' --save_path_last './checkpoints/badnet/watermarking/last.pt' --epoch 100
    14. 

  14. python run.py --model 'efficientnet' --input_size 32 --save_path './checkpoints/efficientnetv2_s/watermarking/best.pt' --save_path_last './checkpoints/efficientnetv2_s/watermarking/last.pt' --epoch 100
    15. 

  15. 

  16. 

  17. 

  18. # For 用于剪枝模型，剪枝后微调训练，保存剪枝后模型路径，以及验证微调模型准确性
    19. 

  19. # -------------------------------------------------------------------------------------------------------------------- #
    20. 

  20. python run.py --model 'resnet'  --prune True --prune_weight './checkpoints/resnet/watermarking/best.pt' --prune_save './checkpoints/resnet/watermarking/prune_best.pt' --epoch 40
    21. 

  21. python run.py --model 'VGG19'  --prune True --prune_weight './checkpoints/VGG19/watermarking/best.pt' --prune_save './checkpoints/VGG19/watermarking/prune_best.pt' --epoch 40
    22. 

  22. python run.py --model 'Alexnet'  --prune True --input_size 112 --prune_weight './checkpoints/Alexnet/watermarking/best.pt' --prune_save './checkpoints/Alexnet/watermarking/prune_best.pt' --epoch 40
    23. 

  23. python run.py --model 'mobilenetv2'  --prune True --prune_weight './checkpoints/mobilenetv2/watermarking/best.pt' --prune_save './checkpoints/mobilenetv2/watermarking/prune_best.pt' --epoch 40
    24. 

  24. python run.py --model 'GoogleNet' --input_size 32 --prune True --prune_weight './checkpoints/GoogleNet/watermarking/best.pt' --prune_save './checkpoints/GoogleNet/watermarking/prune_best.pt' --epoch 40
    25. 

  25. python run.py --model 'badnet' --input_size 32 --prune True --prune_weight './checkpoints/badnet/watermarking/best.pt' --prune_save './checkpoints/badnet/watermarking/prune_best.pt' --epoch 40
    26. 

  26. python run.py --model 'efficientnet' --input_size 32 --prune True --prune_weight './checkpoints/efficientnetv2_s/watermarking/best.pt' --prune_save './checkpoints/efficientnetv2_s/watermarking/prune_best.pt' --epoch 40
    27. 

  27. 

  28. 

  29. 

  30. # For 用于剪枝模型后边缘部署
    31. 

  31. # -------------------------------------------------------------------------------------------------------------------- #
    32.