你好 测试如何进行

[neverstoplearn]

你好 测试如何进行

[neverstoplearn]

测试命令：python3 test_CIFAR10.py --model_path model.pth --out_dir ./output/ --data-dir cifar-data
测试结果：
Files already downloaded and verified
Files already downloaded and verified
0.03137254901960784
tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

AT_models_test_acc: 0.1
AT_fgsm_acc: 0.1
AT_pgd_acc_10: 0.0993
AT_pgd_acc_20: 0.0988
AT_pgd_acc_50: 0.0982
AT_pgd_cw_acc_20: 0.0962

[neverstoplearn]

提供的模型测试结果：
Files already downloaded and verified
Files already downloaded and verified
0.03137254901960784
tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

tensor([[[0.0314]],

    [[0.0314]],

    [[0.0314]]], device='cuda:0')

AT_models_test_acc: 0.8623
AT_fgsm_acc: 0.6898
AT_pgd_acc_10: 0.5762
AT_pgd_acc_20: 0.5648
AT_pgd_acc_50: 0.5615
AT_pgd_cw_acc_20: 0.5572

[neverstoplearn]

最后部分的训练日志：
[2024/05/29 17:02:24] - Epoch Seconds LR Train Loss Train Acc
[2024/05/29 17:26:05] - 107 1420.5 0.0010 1.6196 0.7278
[2024/05/29 17:29:14] - Test Loss Test Acc PGD Loss PGD Acc
[2024/05/29 17:29:14] - 0.7389 0.8615 1.0344 0.5653
[2024/05/29 17:29:14] - Test Acc PGD Acc
[2024/05/29 17:29:14] - 0.8610 0.5705
[2024/05/29 17:29:14] - Epoch Seconds LR Train Loss Train Acc
[2024/05/29 17:53:02] - 108 1428.2 0.0010 1.6174 0.7296
[2024/05/29 17:56:11] - Test Loss Test Acc PGD Loss PGD Acc
[2024/05/29 17:56:11] - 0.7401 0.8616 1.0359 0.5660
[2024/05/29 17:56:11] - Test Acc PGD Acc
[2024/05/29 17:56:11] - 0.8610 0.5705
[2024/05/29 17:56:11] - Epoch Seconds LR Train Loss Train Acc
[2024/05/29 18:19:51] - 109 1420.3 0.0010 1.6146 0.7315
[2024/05/29 18:23:00] - Test Loss Test Acc PGD Loss PGD Acc
[2024/05/29 18:23:00] - 0.7411 0.8624 1.0370 0.5640
[2024/05/29 18:23:00] - Test Acc PGD Acc
[2024/05/29 18:23:00] - 0.8610 0.5705
[2024/05/29 18:23:00] - [0.429, 0.4686, 0.4878, 0.4705, 0.4762, 0.4847, 0.4606, 0.4771, 0.4831, 0.4787, 0.4827, 0.48, 0.4709, 0.4779, 0.4864, 0.4693, 0.4756, 0.4886, 0.4652, 0.474, 0.4783, 0.4411, 0.4663, 0.4778, 0.4774, 0.4775, 0.4886, 0.4868, 0.4701, 0.4604, 0.4673, 0.481, 0.4837, 0.468, 0.4558, 0.488, 0.4464, 0.4546, 0.4825, 0.4444, 0.491, 0.5636, 0.5642, 0.5697, 0.5679, 0.5651, 0.5692, 0.5705, 0.5691, 0.5698, 0.5666, 0.5642, 0.5657, 0.5653, 0.566, 0.564]
[2024/05/29 18:23:00] - [0.7417, 0.7615, 0.7675, 0.7737, 0.7557, 0.7811, 0.7466, 0.7684, 0.7771, 0.7584, 0.7805, 0.7571, 0.769, 0.7786, 0.7652, 0.7675, 0.7725, 0.7804, 0.7598, 0.7809, 0.7715, 0.7449, 0.7657, 0.7828, 0.7654, 0.7831, 0.7858, 0.7611, 0.7744, 0.7637, 0.7823, 0.7724, 0.7803, 0.752, 0.7637, 0.7801, 0.7788, 0.7857, 0.7817, 0.7598, 0.7931, 0.8448, 0.8501, 0.8495, 0.8561, 0.8539, 0.8598, 0.861, 0.8593, 0.8594, 0.8618, 0.8625, 0.8609, 0.8615, 0.8616, 0.8624]

[jiaxiaojunQAQ]

你需要check一下模型的参数是否正常取出，有些是保存在state_dict里面

[neverstoplearn]

你需要check一下模型的参数是否正常取出，有些是保存在state_dict里面

已解决 感谢