Closed yiwliu closed 3 years ago
您好!感谢您对我的研究成果感兴趣。
根据以往的经验和其他使用者的反馈,一般使用已设定好的默认的参数,构建正常的训练数据集、验证数据集,就可以取得很好的效果。 唯一需要注意的是:不能用不适用计算的显卡(比如Nvidia 2080等只适合打游戏的显卡),是否您用了这类计算资源呢?
期待您更多的反馈信息
祝好
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月26日(星期一) 晚上8:10 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "Subscribed"<subscribed@noreply.github.com>; 主题: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
作者你好,我想问一下关于这篇论文如何具体构造数据集,如何进行训练。 鉴于本人水平有限,使用该repo的代码进行实验的时候不能取得很好的效果。
关于如何设置学习率、如何构建训练数据集、验证数据集,具体如何训练的问题希望能够稍微指导一下。
感谢
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.
非常感谢您的回复!!! 我使用的GPU型号为 GeForece RTX 2080Ti,应该不存在使用游戏显卡进行训练的问题。
以下具体描述一下本人的实验的相关配置,以及目前的实验结果。
实验配置 系统:ubuntu 18.04 CUDA version 11.0 python 3.7.9 torch 1.4.0 torchaudio 0.4.0 torchvision 0.4.0 使用的代码为您提供的SiaStegNet 按照您的README中提供的启动方式进行启动,使用的除了train以及valid的cover和stego目录(4个目录)参数进行了替换,model使用kenet
实验数据集构建 本人构建数据集按照fridrich论文SRNet中的实验数据集进行了构造, 使用了BOSSbase以及BOWS一共20000张图片,选取BOSSbase中的4000张作为valid cover,剩下的16000张作为 train cover。 利用fridrich官网提供的S-UNIWARD算法,以0.4的payload生成了以上(16000:4000)对应的stego用于训练。
附件是训练了200多个epoch的结果,以下是其中的一部分信息: 2020-10-27 15:46:52.%f train.py:267[27649] INFO Epoch: 239 2020-10-27 15:46:52.%f train.py:268[27649] INFO Train 2020-10-27 15:48:09.%f train.py:215[27649] INFO Train epoch: 239 [200/1000] Accuracy: 50.59% Loss: 0.720610 2020-10-27 15:49:27.%f train.py:215[27649] INFO Train epoch: 239 [400/1000] Accuracy: 50.73% Loss: 0.720714 2020-10-27 15:50:45.%f train.py:215[27649] INFO Train epoch: 239 [600/1000] Accuracy: 49.02% Loss: 0.721326 2020-10-27 15:52:02.%f train.py:215[27649] INFO Train epoch: 239 [800/1000] Accuracy: 49.77% Loss: 0.720646 2020-10-27 15:53:19.%f train.py:215[27649] INFO Train epoch: 239 [1000/1000] Accuracy: 50.11% Loss: 0.720887 2020-10-27 15:53:19.%f train.py:270[27649] INFO Time: 109008.01739406586 2020-10-27 15:53:19.%f train.py:271[27649] INFO Test 2020-10-27 15:53:42.%f train.py:254[27649] INFO Test set: Loss: 0.7220, Accuracy: 50.04%) 2020-10-27 15:53:42.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-27 15:53:42.%f train.py:283[27649] INFO Time: 109030.6237590313
如果您有时间非常感谢指导!!!
无论如何非常感谢您的回复,祝您工作顺利,万事如意!
SiaStg notifications@github.com 于2020年10月27日周二 下午2:27写道:
您好!感谢您对我的研究成果感兴趣。
根据以往的经验和其他使用者的反馈,一般使用已设定好的默认的参数,构建正常的训练数据集、验证数据集,就可以取得很好的效果。 唯一需要注意的是:不能用不适用计算的显卡(比如Nvidia 2080等只适合打游戏的显卡),是否您用了这类计算资源呢?
期待您更多的反馈信息
祝好
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月26日(星期一) 晚上8:10 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "Subscribed"<subscribed@noreply.github.com>; 主题: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
作者你好,我想问一下关于这篇论文如何具体构造数据集,如何进行训练。 鉴于本人水平有限,使用该repo的代码进行实验的时候不能取得很好的效果。
关于如何设置学习率、如何构建训练数据集、验证数据集,具体如何训练的问题希望能够稍微指导一下。
感谢
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/SiaStg/SiaStegNet/issues/1#issuecomment-717016277, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALXJEEQM3DXN26KTCXJUTF3SMZR67ANCNFSM4S7JJCIA .
2020-10-26 09:36:28.%f env.py:27[27649] INFO Using a generated random seed 28362477 2020-10-26 09:36:28.%f train.py:85[27649] INFO Command Line Arguments: Namespace(alpha=0.1, batch_size=32, ckpt_dir='./kenet_result', cuda=True, epoch=500, eps=1e-08, finetune=None, gpu_id=0, log_interval=200, lr=0.001, lr_str=2, margin=1.0, model='kenet', num_workers=0, random_crop=False, random_crop_train=False, seed=-1, train_cover_dir='/mnt/sda4/datasets/netdata/train/cover', train_stego_dir='/mnt/sda4/datasets/netdata/train/stego', val_cover_dir='/mnt/sda4/datasets/netdata/valid/cover', val_stego_dir='/mnt/sda4/datasets/netdata/valid/stego', wd=0.0001) 2020-10-26 09:36:28.%f train.py:91[27649] INFO Building data loader 2020-10-26 09:36:29.%f dataloader.py:105[27649] INFO Training set length is 32000 2020-10-26 09:36:29.%f dataloader.py:106[27649] INFO Training epoch length is 1000 2020-10-26 09:36:29.%f dataloader.py:152[27649] INFO Testing set length is 8000
客气客气
不建议用2080/2080Ti哦 “GeForce RTX™ 2080 Ti 是 NVIDIA 全新推出的旗舰款显卡,其在游戏逼真度和性能表现方面堪称革命性的技术创举。”
能否方便用搭载其他计算资源的机器试一下呢?
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月27日(星期二) 下午4:01 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
非常感谢您的回复!!! 我使用的GPU型号为 GeForece RTX 2080Ti,应该不存在使用游戏显卡进行训练的问题。
以下具体描述一下本人的实验的相关配置,以及目前的实验结果。
实验配置 系统:ubuntu 18.04 CUDA version 11.0 python 3.7.9 torch 1.4.0 torchaudio 0.4.0 torchvision 0.4.0 使用的代码为您提供的SiaStegNet 按照您的README中提供的启动方式进行启动,使用的除了train以及valid的cover和stego目录(4个目录)参数进行了替换,model使用kenet
实验数据集构建 本人构建数据集按照fridrish论文SRNet中的实验数据集进行了构造, 使用了BOSSbase以及BOWS一共20000张图片,选取BOSSbase中的4000张作为valid cover,剩下的16000张作为 train cover。利用fridrish官网提供的 S-UNIWARD算法,以0.4的payload生成了以上(16000:4000)对应的stego用于训练。
附件是训练了200多个epoch的结果,以下是其中的一部分信息: 2020-10-27 15:46:52.%f train.py:267[27649] INFO Epoch: 239 2020-10-27 15:46:52.%f train.py:268[27649] INFO Train 2020-10-27 15:48:09.%f train.py:215[27649] INFO Train epoch: 239 [200/1000] Accuracy: 50.59% Loss: 0.720610 2020-10-27 15:49:27.%f train.py:215[27649] INFO Train epoch: 239 [400/1000] Accuracy: 50.73% Loss: 0.720714 2020-10-27 15:50:45.%f train.py:215[27649] INFO Train epoch: 239 [600/1000] Accuracy: 49.02% Loss: 0.721326 2020-10-27 15:52:02.%f train.py:215[27649] INFO Train epoch: 239 [800/1000] Accuracy: 49.77% Loss: 0.720646 2020-10-27 15:53:19.%f train.py:215[27649] INFO Train epoch: 239 [1000/1000] Accuracy: 50.11% Loss: 0.720887 2020-10-27 15:53:19.%f train.py:270[27649] INFO Time: 109008.01739406586 2020-10-27 15:53:19.%f train.py:271[27649] INFO Test 2020-10-27 15:53:42.%f train.py:254[27649] INFO Test set: Loss: 0.7220, Accuracy: 50.04%) 2020-10-27 15:53:42.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-27 15:53:42.%f train.py:283[27649] INFO Time: 109030.6237590313
如果您有时间非常感谢指导!!!
无论如何非常感谢您的回复,祝您工作顺利,完事如意!
SiaStg <notifications@github.com> 于2020年10月27日周二 下午2:27写道:
> 您好!感谢您对我的研究成果感兴趣。 > > > 根据以往的经验和其他使用者的反馈,一般使用已设定好的默认的参数,构建正常的训练数据集、验证数据集,就可以取得很好的效果。 > 唯一需要注意的是:不能用不适用计算的显卡(比如Nvidia 2080等只适合打游戏的显卡),是否您用了这类计算资源呢? > > > 期待您更多的反馈信息 > > > 祝好 > > > > > ------------------ 原始邮件 ------------------ > 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; > 发送时间: 2020年10月26日(星期一) 晚上8:10 > 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; > 抄送: "Subscribed"<subscribed@noreply.github.com>; > 主题: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) > > > > > > > 作者你好,我想问一下关于这篇论文如何具体构造数据集,如何进行训练。 > 鉴于本人水平有限,使用该repo的代码进行实验的时候不能取得很好的效果。 > > 关于如何设置学习率、如何构建训练数据集、验证数据集,具体如何训练的问题希望能够稍微指导一下。 > > 感谢 > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > <https://github.com/SiaStg/SiaStegNet/issues/1#issuecomment-717016277>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ALXJEEQM3DXN26KTCXJUTF3SMZR67ANCNFSM4S7JJCIA> > . >
2020-10-26 09:36:28.%f env.py:27[27649] INFO Using a generated random seed 28362477 2020-10-26 09:36:28.%f train.py:85[27649] INFO Command Line Arguments: Namespace(alpha=0.1, batch_size=32, ckpt_dir='./kenet_result', cuda=True, epoch=500, eps=1e-08, finetune=None, gpu_id=0, log_interval=200, lr=0.001, lr_str=2, margin=1.0, model='kenet', num_workers=0, random_crop=False, random_crop_train=False, seed=-1, train_cover_dir='/mnt/sda4/datasets/netdata/train/cover', train_stego_dir='/mnt/sda4/datasets/netdata/train/stego', val_cover_dir='/mnt/sda4/datasets/netdata/valid/cover', val_stego_dir='/mnt/sda4/datasets/netdata/valid/stego', wd=0.0001) 2020-10-26 09:36:28.%f train.py:91[27649] INFO Building data loader 2020-10-26 09:36:29.%f dataloader.py:105[27649] INFO Training set length is 32000 2020-10-26 09:36:29.%f dataloader.py:106[27649] INFO Training epoch length is 1000 2020-10-26 09:36:29.%f dataloader.py:152[27649] INFO Testing set length is 8000 2020-10-26 09:36:29.%f train.py:109[27649] INFO Building model 2020-10-26 09:36:31.%f train.py:267[27649] INFO Epoch: 1 2020-10-26 09:36:31.%f train.py:268[27649] INFO Train 2020-10-26 09:37:43.%f train.py:215[27649] INFO Train epoch: 1 [200/1000] Accuracy: 50.25% Loss: 0.734632 2020-10-26 09:38:57.%f train.py:215[27649] INFO Train epoch: 1 [400/1000] Accuracy: 49.83% Loss: 0.724080 2020-10-26 09:41:11.%f train.py:215[27649] INFO Train epoch: 1 [600/1000] Accuracy: 49.33% Loss: 0.724804 2020-10-26 09:43:47.%f train.py:215[27649] INFO Train epoch: 1 [800/1000] Accuracy: 50.06% Loss: 0.723864 2020-10-26 09:46:29.%f train.py:215[27649] INFO Train epoch: 1 [1000/1000] Accuracy: 50.41% Loss: 0.723972 2020-10-26 09:46:29.%f train.py:270[27649] INFO Time: 598.1813097000122 2020-10-26 09:46:29.%f train.py:271[27649] INFO Test 2020-10-26 09:47:13.%f train.py:254[27649] INFO Test set: Loss: 0.7251, Accuracy: 50.12%) 2020-10-26 09:47:13.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 09:47:13.%f train.py:283[27649] INFO Time: 641.4804406166077 2020-10-26 09:47:13.%f train.py:267[27649] INFO Epoch: 2 2020-10-26 09:47:13.%f train.py:268[27649] INFO Train 2020-10-26 09:49:56.%f train.py:215[27649] INFO Train epoch: 2 [200/1000] Accuracy: 50.08% Loss: 0.724571 2020-10-26 09:52:45.%f train.py:215[27649] INFO Train epoch: 2 [400/1000] Accuracy: 51.30% Loss: 0.723082 2020-10-26 09:55:23.%f train.py:215[27649] INFO Train epoch: 2 [600/1000] Accuracy: 48.69% Loss: 0.724915 2020-10-26 09:58:04.%f train.py:215[27649] INFO Train epoch: 2 [800/1000] Accuracy: 50.47% Loss: 0.722644 2020-10-26 10:00:44.%f train.py:215[27649] INFO Train epoch: 2 [1000/1000] Accuracy: 50.42% Loss: 0.722950 2020-10-26 10:00:45.%f train.py:270[27649] INFO Time: 1453.9544608592987 2020-10-26 10:00:45.%f train.py:271[27649] INFO Test 2020-10-26 10:01:32.%f train.py:254[27649] INFO Test set: Loss: 0.7250, Accuracy: 50.05%) 2020-10-26 10:01:32.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:01:32.%f train.py:283[27649] INFO Time: 1500.8485856056213 2020-10-26 10:01:32.%f train.py:267[27649] INFO Epoch: 3 2020-10-26 10:01:32.%f train.py:268[27649] INFO Train 2020-10-26 10:04:10.%f train.py:215[27649] INFO Train epoch: 3 [200/1000] Accuracy: 49.70% Loss: 0.724378 2020-10-26 10:06:42.%f train.py:215[27649] INFO Train epoch: 3 [400/1000] Accuracy: 49.97% Loss: 0.724390 2020-10-26 10:09:20.%f train.py:215[27649] INFO Train epoch: 3 [600/1000] Accuracy: 49.53% Loss: 0.724867 2020-10-26 10:12:00.%f train.py:215[27649] INFO Train epoch: 3 [800/1000] Accuracy: 49.56% Loss: 0.723496 2020-10-26 10:14:39.%f train.py:215[27649] INFO Train epoch: 3 [1000/1000] Accuracy: 50.38% Loss: 0.722862 2020-10-26 10:14:40.%f train.py:270[27649] INFO Time: 2288.688045501709 2020-10-26 10:14:40.%f train.py:271[27649] INFO Test 2020-10-26 10:15:27.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.06%) 2020-10-26 10:15:27.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:15:27.%f train.py:283[27649] INFO Time: 2336.0412130355835 2020-10-26 10:15:27.%f train.py:267[27649] INFO Epoch: 4 2020-10-26 10:15:27.%f train.py:268[27649] INFO Train 2020-10-26 10:18:08.%f train.py:215[27649] INFO Train epoch: 4 [200/1000] Accuracy: 50.41% Loss: 0.723008 2020-10-26 10:20:53.%f train.py:215[27649] INFO Train epoch: 4 [400/1000] Accuracy: 50.16% Loss: 0.724553 2020-10-26 10:23:33.%f train.py:215[27649] INFO Train epoch: 4 [600/1000] Accuracy: 51.03% Loss: 0.723544 2020-10-26 10:26:14.%f train.py:215[27649] INFO Train epoch: 4 [800/1000] Accuracy: 50.28% Loss: 0.723467 2020-10-26 10:29:00.%f train.py:215[27649] INFO Train epoch: 4 [1000/1000] Accuracy: 48.58% Loss: 0.724647 2020-10-26 10:29:01.%f train.py:270[27649] INFO Time: 3149.5196056365967 2020-10-26 10:29:01.%f train.py:271[27649] INFO Test 2020-10-26 10:29:44.%f train.py:254[27649] INFO Test set: Loss: 0.7248, Accuracy: 50.11%) 2020-10-26 10:29:44.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:29:44.%f train.py:283[27649] INFO Time: 3192.6355996131897 2020-10-26 10:29:44.%f train.py:267[27649] INFO Epoch: 5 2020-10-26 10:29:44.%f train.py:268[27649] INFO Train 2020-10-26 10:32:30.%f train.py:215[27649] INFO Train epoch: 5 [200/1000] Accuracy: 50.56% Loss: 0.722997 2020-10-26 10:35:11.%f train.py:215[27649] INFO Train epoch: 5 [400/1000] Accuracy: 50.25% Loss: 0.723275 2020-10-26 10:37:51.%f train.py:215[27649] INFO Train epoch: 5 [600/1000] Accuracy: 49.91% Loss: 0.723445 2020-10-26 10:40:34.%f train.py:215[27649] INFO Train epoch: 5 [800/1000] Accuracy: 50.12% Loss: 0.723526 2020-10-26 10:42:59.%f train.py:215[27649] INFO Train epoch: 5 [1000/1000] Accuracy: 51.03% Loss: 0.723624 2020-10-26 10:42:59.%f train.py:270[27649] INFO Time: 3987.87735247612 2020-10-26 10:42:59.%f train.py:271[27649] INFO Test 2020-10-26 10:43:31.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.12%) 2020-10-26 10:43:31.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:43:31.%f train.py:283[27649] INFO Time: 4019.8337297439575 2020-10-26 10:43:31.%f train.py:267[27649] INFO Epoch: 6 2020-10-26 10:43:31.%f train.py:268[27649] INFO Train 2020-10-26 10:46:15.%f train.py:215[27649] INFO Train epoch: 6 [200/1000] Accuracy: 50.52% Loss: 0.722977 2020-10-26 10:48:55.%f train.py:215[27649] INFO Train epoch: 6 [400/1000] Accuracy: 50.16% Loss: 0.723495 2020-10-26 10:51:35.%f train.py:215[27649] INFO Train epoch: 6 [600/1000] Accuracy: 49.84% Loss: 0.723730 2020-10-26 10:54:18.%f train.py:215[27649] INFO Train epoch: 6 [800/1000] Accuracy: 49.83% Loss: 0.723103 2020-10-26 10:56:58.%f train.py:215[27649] INFO Train epoch: 6 [1000/1000] Accuracy: 49.44% Loss: 0.724185 2020-10-26 10:56:58.%f train.py:270[27649] INFO Time: 4827.371083021164 2020-10-26 10:56:58.%f train.py:271[27649] INFO Test 2020-10-26 10:57:46.%f train.py:254[27649] INFO Test set: Loss: 0.7251, Accuracy: 50.02%) 2020-10-26 10:57:46.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:57:46.%f train.py:283[27649] INFO Time: 4874.3904139995575 2020-10-26 10:57:46.%f train.py:267[27649] INFO Epoch: 7 2020-10-26 10:57:46.%f train.py:268[27649] INFO Train 2020-10-26 11:00:26.%f train.py:215[27649] INFO Train epoch: 7 [200/1000] Accuracy: 50.39% Loss: 0.723030 2020-10-26 11:03:04.%f train.py:215[27649] INFO Train epoch: 7 [400/1000] Accuracy: 49.62% Loss: 0.723639 2020-10-26 11:04:22.%f train.py:215[27649] INFO Train epoch: 7 [600/1000] Accuracy: 50.14% Loss: 0.723232 2020-10-26 11:06:20.%f train.py:215[27649] INFO Train epoch: 7 [800/1000] Accuracy: 50.88% Loss: 0.723464 2020-10-26 11:09:00.%f train.py:215[27649] INFO Train epoch: 7 [1000/1000] Accuracy: 50.33% Loss: 0.723632 2020-10-26 11:09:00.%f train.py:270[27649] INFO Time: 5549.245084285736 2020-10-26 11:09:00.%f train.py:271[27649] INFO Test 2020-10-26 11:09:47.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.14%) 2020-10-26 11:09:47.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:09:47.%f train.py:283[27649] INFO Time: 5596.314600229263 2020-10-26 11:09:47.%f train.py:267[27649] INFO Epoch: 8 2020-10-26 11:09:47.%f train.py:268[27649] INFO Train 2020-10-26 11:12:28.%f train.py:215[27649] INFO Train epoch: 8 [200/1000] Accuracy: 50.12% Loss: 0.722704 2020-10-26 11:15:11.%f train.py:215[27649] INFO Train epoch: 8 [400/1000] Accuracy: 49.34% Loss: 0.723899 2020-10-26 11:17:51.%f train.py:215[27649] INFO Train epoch: 8 [600/1000] Accuracy: 49.66% Loss: 0.723716 2020-10-26 11:20:31.%f train.py:215[27649] INFO Train epoch: 8 [800/1000] Accuracy: 49.61% Loss: 0.723263 2020-10-26 11:23:15.%f train.py:215[27649] INFO Train epoch: 8 [1000/1000] Accuracy: 50.39% Loss: 0.722728 2020-10-26 11:23:15.%f train.py:270[27649] INFO Time: 6404.300854444504 2020-10-26 11:23:15.%f train.py:271[27649] INFO Test 2020-10-26 11:23:59.%f train.py:254[27649] INFO Test set: Loss: 0.7250, Accuracy: 50.06%) 2020-10-26 11:23:59.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:23:59.%f train.py:283[27649] INFO Time: 6447.4487590789795 2020-10-26 11:23:59.%f train.py:267[27649] INFO Epoch: 9 2020-10-26 11:23:59.%f train.py:268[27649] INFO Train 2020-10-26 11:26:44.%f train.py:215[27649] INFO Train epoch: 9 [200/1000] Accuracy: 50.09% Loss: 0.723151 2020-10-26 11:29:24.%f train.py:215[27649] INFO Train epoch: 9 [400/1000] Accuracy: 50.05% Loss: 0.723475 2020-10-26 11:32:04.%f train.py:215[27649] INFO Train epoch: 9 [600/1000] Accuracy: 50.33% Loss: 0.723219 2020-10-26 11:34:47.%f train.py:215[27649] INFO Train epoch: 9 [800/1000] Accuracy: 49.06% Loss: 0.723540 2020-10-26 11:37:25.%f train.py:215[27649] INFO Train epoch: 9 [1000/1000] Accuracy: 50.19% Loss: 0.723136 2020-10-26 11:37:26.%f train.py:270[27649] INFO Time: 7254.803579568863 2020-10-26 11:37:26.%f train.py:271[27649] INFO Test 2020-10-26 11:38:13.%f train.py:254[27649] INFO Test set: Loss: 0.7253, Accuracy: 50.00%) 2020-10-26 11:38:13.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:38:13.%f train.py:283[27649] INFO Time: 7301.76851773262 2020-10-26 11:38:13.%f train.py:267[27649] INFO Epoch: 10 2020-10-26 11:38:13.%f train.py:268[27649] INFO Train 2020-10-26 11:40:53.%f train.py:215[27649] INFO Train epoch: 10 [200/1000] Accuracy: 50.08% Loss: 0.722869 2020-10-26 11:43:33.%f train.py:215[27649] INFO Train epoch: 10 [400/1000] Accuracy: 49.64% Loss: 0.723510 2020-10-26 11:46:17.%f train.py:215[27649] INFO Train epoch: 10 [600/1000] Accuracy: 50.70% Loss: 0.722605 2020-10-26 11:48:57.%f train.py:215[27649] INFO Train epoch: 10 [800/1000] Accuracy: 49.70% Loss: 0.722730 2020-10-26 11:51:37.%f train.py:215[27649] INFO Train epoch: 10 [1000/1000] Accuracy: 49.05% Loss: 0.723217 2020-10-26 11:51:38.%f train.py:270[27649] INFO Time: 8106.941040277481 2020-10-26 11:51:38.%f train.py:271[27649] INFO Test 2020-10-26 11:52:25.%f train.py:254[27649] INFO Test set: Loss: 0.7246, Accuracy: 49.99%) 2020-10-26 11:52:25.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:52:25.%f train.py:283[27649] INFO Time: 8153.689094305038 2020-10-26 11:52:25.%f train.py:267[27649] INFO Epoch: 11 2020-10-26 11:52:25.%f train.py:268[27649] INFO Train 2020-10-26 11:55:04.%f train.py:215[27649] INFO Train epoch: 11 [200/1000] Accuracy: 48.84% Loss: 0.722997 2020-10-26 11:57:43.%f train.py:215[27649] INFO Train epoch: 11 [400/1000] Accuracy: 49.75% Loss: 0.722570 2020-10-26 12:00:25.%f train.py:215[27649] INFO Train epoch: 11 [600/1000] Accuracy: 50.05% Loss: 0.722979 2020-10-26 12:03:03.%f train.py:215[27649] INFO Train epoch: 11 [800/1000] Accuracy: 50.08% Loss: 0.722205 2020-10-26 12:05:40.%f train.py:215[27649] INFO Train epoch: 11 [1000/1000] Accuracy: 50.39% Loss: 0.722835 2020-10-26 12:05:41.%f train.py:270[27649] INFO Time: 8949.71140408516 2020-10-26 12:05:41.%f train.py:271[27649] INFO Test 2020-10-26 12:06:27.%f train.py:254[27649] INFO Test set: Loss: 0.7241, Accuracy: 50.09%) 2020-10-26 12:06:27.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 12:06:27.%f train.py:283[27649] INFO Time: 8996.346478939056 2020-10-26 12:06:27.%f train.py:267[27649] INFO Epoch: 12 2020-10-26 12:06:27.%f train.py:268[27649] INFO Train 2020-10-26 12:09:04.%f train.py:215[27649] INFO Train epoch: 12 [200/1000] Accuracy: 50.06% Loss: 0.722660 2020-10-26 12:11:44.%f train.py:215[27649] INFO Train epoch: 12 [400/1000] Accuracy: 49.77% Loss: 0.723204 2020-10-26 12:14:22.%f train.py:215[27649] INFO Train epoch: 12 [600/1000] Accuracy: 49.12% Loss: 0.723967 2020-10-26 12:17:00.%f train.py:215[27649] INFO Train epoch: 12 [800/1000] Accuracy: 51.12% Loss: 0.722024 2020-10-26 12:19:41.%f train.py:215[27649] INFO Train epoch: 12 [1000/1000] Accuracy: 50.50% Loss: 0.722844 2020-10-26 12:19:41.%f train.py:270[27649] INFO Time: 9790.267852544785 2020-10-26 12:19:41.%f train.py:271[27649] INFO Test 2020-10-26 12:20:24.%f train.py:254[27649] INFO Test set: Loss: 0.7245, Accuracy: 50.18%) 2020-10-26 12:20:24.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:20:24.%f train.py:283[27649] INFO Time: 9833.335562229156 2020-10-26 12:20:24.%f train.py:267[27649] INFO Epoch: 13 2020-10-26 12:20:24.%f train.py:268[27649] INFO Train 2020-10-26 12:23:05.%f train.py:215[27649] INFO Train epoch: 13 [200/1000] Accuracy: 50.50% Loss: 0.722402 2020-10-26 12:25:43.%f train.py:215[27649] INFO Train epoch: 13 [400/1000] Accuracy: 49.67% Loss: 0.722641 2020-10-26 12:28:21.%f train.py:215[27649] INFO Train epoch: 13 [600/1000] Accuracy: 50.34% Loss: 0.722604 2020-10-26 12:31:00.%f train.py:215[27649] INFO Train epoch: 13 [800/1000] Accuracy: 49.55% Loss: 0.723070 2020-10-26 12:33:41.%f train.py:215[27649] INFO Train epoch: 13 [1000/1000] Accuracy: 50.64% Loss: 0.721970 2020-10-26 12:33:41.%f train.py:270[27649] INFO Time: 10630.292489290237 2020-10-26 12:33:41.%f train.py:271[27649] INFO Test 2020-10-26 12:34:28.%f train.py:254[27649] INFO Test set: Loss: 0.7243, Accuracy: 49.94%) 2020-10-26 12:34:28.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:34:28.%f train.py:283[27649] INFO Time: 10677.06707572937 2020-10-26 12:34:28.%f train.py:267[27649] INFO Epoch: 14 2020-10-26 12:34:28.%f train.py:268[27649] INFO Train 2020-10-26 12:37:06.%f train.py:215[27649] INFO Train epoch: 14 [200/1000] Accuracy: 49.28% Loss: 0.722930 2020-10-26 12:39:44.%f train.py:215[27649] INFO Train epoch: 14 [400/1000] Accuracy: 50.58% Loss: 0.722355 2020-10-26 12:42:21.%f train.py:215[27649] INFO Train epoch: 14 [600/1000] Accuracy: 49.70% Loss: 0.723078 2020-10-26 12:45:04.%f train.py:215[27649] INFO Train epoch: 14 [800/1000] Accuracy: 50.50% Loss: 0.721714 2020-10-26 12:47:42.%f train.py:215[27649] INFO Train epoch: 14 [1000/1000] Accuracy: 49.30% Loss: 0.722824 2020-10-26 12:47:43.%f train.py:270[27649] INFO Time: 11471.877284765244 2020-10-26 12:47:43.%f train.py:271[27649] INFO Test 2020-10-26 12:48:30.%f train.py:254[27649] INFO Test set: Loss: 0.7246, Accuracy: 50.05%) 2020-10-26 12:48:30.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:48:30.%f train.py:283[27649] INFO Time: 11518.724534511566 2020-10-26 12:48:30.%f train.py:267[27649] INFO Epoch: 15 2020-10-26 12:48:30.%f train.py:268[27649] INFO Train 2020-10-26 12:51:09.%f train.py:215[27649] INFO Train epoch: 15 [200/1000] Accuracy: 50.44% Loss: 0.722518 2020-10-26 12:53:49.%f train.py:215[27649] INFO Train epoch: 15 [400/1000] Accuracy: 50.44% Loss: 0.722605 2020-10-26 12:56:33.%f train.py:215[27649] INFO Train epoch: 15 [600/1000] Accuracy: 50.66% Loss: 0.722649 2020-10-26 12:59:12.%f train.py:215[27649] INFO Train epoch: 15 [800/1000] Accuracy: 50.16% Loss: 0.722128 2020-10-26 13:01:51.%f train.py:215[27649] INFO Train epoch: 15 [1000/1000] Accuracy: 50.11% Loss: 0.722298 2020-10-26 13:01:52.%f train.py:270[27649] INFO Time: 12320.535963058472 2020-10-26 13:01:52.%f train.py:271[27649] INFO Test 2020-10-26 13:02:38.%f train.py:254[27649] INFO Test set: Loss: 0.7242, Accuracy: 50.08%) 2020-10-26 13:02:38.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 13:02:38.%f train.py:283[27649] INFO Time: 12367.37258219719 2020-10-26 13:02:39.%f train.py:267[27649] INFO Epoch: 16 2020-10-26 13:02:39.%f train.py:268[27649] INFO Train 2020-10-26 13:05:17.%f train.py:215[27649] INFO Train epoch: 16 [200/1000] Accuracy: 49.58% Loss: 0.722341 2020-10-26 13:08:00.%f train.py:215[27649] INFO Train epoch: 16 [400/1000] Accuracy: 49.52% Loss: 0.722842 2020-10-26 13:10:37.%f train.py:215[27649] INFO Train epoch: 16 [600/1000] Accuracy: 50.11% Loss: 0.722938 2020-10-26 13:13:15.%f train.py:215[27649] INFO Train epoch: 16 [800/1000] Accuracy: 49.05% Loss: 0.722603 2020-10-26 13:15:53.%f train.py:215[27649] INFO Train epoch: 16 [1000/1000] Accuracy: 50.19% Loss: 0.722058 2020-10-26 13:15:53.%f train.py:270[27649] INFO Time: 13162.007128238678 2020-10-26 13:15:53.%f train.py:271[27649] INFO Test 2020-10-26 13:16:40.%f train.py:254[27649] INFO Test set: Loss: 0.7241, Accuracy: 50.30%) 2020-10-26 13:16:40.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:16:40.%f train.py:283[27649] INFO Time: 13208.76679611206 2020-10-26 13:16:40.%f train.py:267[27649] INFO Epoch: 17 2020-10-26 13:16:40.%f train.py:268[27649] INFO Train 2020-10-26 13:19:18.%f train.py:215[27649] INFO Train epoch: 17 [200/1000] Accuracy: 49.69% Loss: 0.722293 2020-10-26 13:21:59.%f train.py:215[27649] INFO Train epoch: 17 [400/1000] Accuracy: 49.48% Loss: 0.722677 2020-10-26 13:24:36.%f train.py:215[27649] INFO Train epoch: 17 [600/1000] Accuracy: 49.44% Loss: 0.722369 2020-10-26 13:27:13.%f train.py:215[27649] INFO Train epoch: 17 [800/1000] Accuracy: 49.81% Loss: 0.722521 2020-10-26 13:29:54.%f train.py:215[27649] INFO Train epoch: 17 [1000/1000] Accuracy: 49.73% Loss: 0.722504 2020-10-26 13:29:55.%f train.py:270[27649] INFO Time: 14003.732480049133 2020-10-26 13:29:55.%f train.py:271[27649] INFO Test 2020-10-26 13:30:38.%f train.py:254[27649] INFO Test set: Loss: 0.7243, Accuracy: 50.11%) 2020-10-26 13:30:38.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:30:38.%f train.py:283[27649] INFO Time: 14046.820983886719 2020-10-26 13:30:38.%f train.py:267[27649] INFO Epoch: 18 2020-10-26 13:30:38.%f train.py:268[27649] INFO Train 2020-10-26 13:33:19.%f train.py:215[27649] INFO Train epoch: 18 [200/1000] Accuracy: 50.66% Loss: 0.722227 2020-10-26 13:35:56.%f train.py:215[27649] INFO Train epoch: 18 [400/1000] Accuracy: 49.22% Loss: 0.722706 2020-10-26 13:38:34.%f train.py:215[27649] INFO Train epoch: 18 [600/1000] Accuracy: 49.81% Loss: 0.722656 2020-10-26 13:41:17.%f train.py:215[27649] INFO Train epoch: 18 [800/1000] Accuracy: 49.55% Loss: 0.722005 2020-10-26 13:43:54.%f train.py:215[27649] INFO Train epoch: 18 [1000/1000] Accuracy: 50.89% Loss: 0.722071 2020-10-26 13:43:55.%f train.py:270[27649] INFO Time: 14843.993800878525 2020-10-26 13:43:55.%f train.py:271[27649] INFO Test 2020-10-26 13:44:42.%f train.py:254[27649] INFO Test set: Loss: 0.7247, Accuracy: 50.01%) 2020-10-26 13:44:42.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:44:42.%f train.py:283[27649] INFO Time: 14890.727468252182 2020-10-26 13:44:42.%f train.py:267[27649] INFO Epoch: 19 2020-10-26 13:44:42.%f train.py:268[27649] INFO Train 2020-10-26 13:47:19.%f train.py:215[27649] INFO Train epoch: 19 [200/1000] Accuracy: 50.53% Loss: 0.721319 2020-10-26 13:49:57.%f train.py:215[27649] INFO Train epoch: 19 [400/1000] Accuracy: 49.34% Loss: 0.722276 2020-10-26 13:52:38.%f train.py:215[27649] INFO Train epoch: 19 [600/1000] Accuracy: 49.38% Loss: 0.722035 2020-10-26 13:55:17.%f train.py:215[27649] INFO Train epoch: 19 [800/1000] Accuracy: 49.98% Loss: 0.721916 2020-10-26 13:57:57.%f train.py:215[27649] INFO Train epoch: 19 [1000/1000] Accuracy: 48.97% Loss: 0.722423 2020-10-26 13:57:57.%f train.py:270[27649] INFO Time: 15686.256188631058 2020-10-26 13:57:57.%f train.py:271[27649] INFO Test 2020-10-26 13:58:44.%f train.py:254[27649] INFO Test set: Loss: 0.7239, Accuracy: 50.05%) 2020-10-26 13:58:44.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:58:44.%f train.py:283[27649] INFO Time: 15732.983254909515 2020-10-26 13:58:44.%f train.py:267[27649] INFO Epoch: 20 2020-10-26 13:58:44.%f train.py:268[27649] INFO Train 2020-10-26 14:01:24.%f train.py:215[27649] INFO Train epoch: 20 [200/1000] Accuracy: 48.89% Loss: 0.722335 2020-10-26 14:04:04.%f train.py:215[27649] INFO Train epoch: 20 [400/1000] Accuracy: 50.56% Loss: 0.721885 2020-10-26 14:06:48.%f train.py:215[27649] INFO Train epoch: 20 [600/1000] Accuracy: 49.50% Loss: 0.722265 2020-10-26 14:09:28.%f train.py:215[27649] INFO Train epoch: 20 [800/1000] Accuracy: 49.81% Loss: 0.721991 2020-10-26 14:12:08.%f train.py:215[27649] INFO Train epoch: 20 [1000/1000] Accuracy: 49.33% Loss: 0.722232 2020-10-26 14:12:09.%f train.py:270[27649] INFO Time: 16537.672178030014 2020-10-26 14:12:09.%f train.py:271[27649] INFO Test 2020-10-26 14:12:56.%f train.py:254[27649] INFO Test set: Loss: 0.7236, Accuracy: 50.09%) 2020-10-26 14:12:56.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:12:56.%f train.py:283[27649] INFO Time: 16584.443323135376 2020-10-26 14:12:56.%f train.py:267[27649] INFO Epoch: 21 2020-10-26 14:12:56.%f train.py:268[27649] INFO Train 2020-10-26 14:15:36.%f train.py:215[27649] INFO Train epoch: 21 [200/1000] Accuracy: 49.75% Loss: 0.722232 2020-10-26 14:18:19.%f train.py:215[27649] INFO Train epoch: 21 [400/1000] Accuracy: 49.98% Loss: 0.722148 2020-10-26 14:20:58.%f train.py:215[27649] INFO Train epoch: 21 [600/1000] Accuracy: 49.28% Loss: 0.722148 2020-10-26 14:23:36.%f train.py:215[27649] INFO Train epoch: 21 [800/1000] Accuracy: 49.09% Loss: 0.721895 2020-10-26 14:26:17.%f train.py:215[27649] INFO Train epoch: 21 [1000/1000] Accuracy: 50.53% Loss: 0.721511 2020-10-26 14:26:18.%f train.py:270[27649] INFO Time: 17386.484574079514 2020-10-26 14:26:18.%f train.py:271[27649] INFO Test 2020-10-26 14:27:02.%f train.py:254[27649] INFO Test set: Loss: 0.7240, Accuracy: 50.08%) 2020-10-26 14:27:02.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:27:02.%f train.py:283[27649] INFO Time: 17430.920579195023 2020-10-26 14:27:02.%f train.py:267[27649] INFO Epoch: 22 2020-10-26 14:27:02.%f train.py:268[27649] INFO Train 2020-10-26 14:29:44.%f train.py:215[27649] INFO Train epoch: 22 [200/1000] Accuracy: 50.98% Loss: 0.721848 2020-10-26 14:32:22.%f train.py:215[27649] INFO Train epoch: 22 [400/1000] Accuracy: 49.48% Loss: 0.722381 2020-10-26 14:35:00.%f train.py:215[27649] INFO Train epoch: 22 [600/1000] Accuracy: 50.03% Loss: 0.721494 2020-10-26 14:37:39.%f train.py:215[27649] INFO Train epoch: 22 [800/1000] Accuracy: 50.00% Loss: 0.721463 2020-10-26 14:40:20.%f train.py:215[27649] INFO Train epoch: 22 [1000/1000] Accuracy: 49.16% Loss: 0.721714 2020-10-26 14:40:21.%f train.py:270[27649] INFO Time: 18229.89799261093 2020-10-26 14:40:21.%f train.py:271[27649] INFO Test 2020-10-26 14:41:07.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.14%) 2020-10-26 14:41:07.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:41:07.%f train.py:283[27649] INFO Time: 18275.499089717865 2020-10-26 14:41:07.%f train.py:267[27649] INFO Epoch: 23 2020-10-26 14:41:07.%f train.py:268[27649] INFO Train 2020-10-26 14:43:45.%f train.py:215[27649] INFO Train epoch: 23 [200/1000] Accuracy: 49.75% Loss: 0.722079 2020-10-26 14:46:24.%f train.py:215[27649] INFO Train epoch: 23 [400/1000] Accuracy: 49.42% Loss: 0.721907 2020-10-26 14:49:03.%f train.py:215[27649] INFO Train epoch: 23 [600/1000] Accuracy: 50.12% Loss: 0.721677 2020-10-26 14:51:46.%f train.py:215[27649] INFO Train epoch: 23 [800/1000] Accuracy: 50.41% Loss: 0.721803 2020-10-26 14:54:25.%f train.py:215[27649] INFO Train epoch: 23 [1000/1000] Accuracy: 49.38% Loss: 0.722154 2020-10-26 14:54:26.%f train.py:270[27649] INFO Time: 19074.978511571884 2020-10-26 14:54:26.%f train.py:271[27649] INFO Test 2020-10-26 14:55:13.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.04%) 2020-10-26 14:55:13.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:55:13.%f train.py:283[27649] INFO Time: 19121.877328634262 2020-10-26 14:55:13.%f train.py:267[27649] INFO Epoch: 24 2020-10-26 14:55:13.%f train.py:268[27649] INFO Train 2020-10-26 14:57:52.%f train.py:215[27649] INFO Train epoch: 24 [200/1000] Accuracy: 49.42% Loss: 0.721749 2020-10-26 15:00:30.%f train.py:215[27649] INFO Train epoch: 24 [400/1000] Accuracy: 50.80% Loss: 0.721292 2020-10-26 15:03:13.%f train.py:215[27649] INFO Train epoch: 24 [600/1000] Accuracy: 50.06% Loss: 0.722138 2020-10-26 15:05:52.%f train.py:215[27649] INFO Train epoch: 24 [800/1000] Accuracy: 50.14% Loss: 0.721673 2020-10-26 15:08:31.%f train.py:215[27649] INFO Train epoch: 24 [1000/1000] Accuracy: 48.86% Loss: 0.722384 2020-10-26 15:08:32.%f train.py:270[27649] INFO Time: 19920.678875923157 2020-10-26 15:08:32.%f train.py:271[27649] INFO Test 2020-10-26 15:09:18.%f train.py:254[27649] INFO Test set: Loss: 0.7233, Accuracy: 50.01%) 2020-10-26 15:09:18.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:09:18.%f train.py:283[27649] INFO Time: 19967.131550312042 2020-10-26 15:09:18.%f train.py:267[27649] INFO Epoch: 25 2020-10-26 15:09:18.%f train.py:268[27649] INFO Train 2020-10-26 15:11:28.%f train.py:215[27649] INFO Train epoch: 25 [200/1000] Accuracy: 49.50% Loss: 0.721968 2020-10-26 15:14:08.%f train.py:215[27649] INFO Train epoch: 25 [400/1000] Accuracy: 50.16% Loss: 0.721596 2020-10-26 15:16:52.%f train.py:215[27649] INFO Train epoch: 25 [600/1000] Accuracy: 49.67% Loss: 0.721803 2020-10-26 15:19:32.%f train.py:215[27649] INFO Train epoch: 25 [800/1000] Accuracy: 50.33% Loss: 0.721453 2020-10-26 15:22:11.%f train.py:215[27649] INFO Train epoch: 25 [1000/1000] Accuracy: 50.62% Loss: 0.721836 2020-10-26 15:22:12.%f train.py:270[27649] INFO Time: 20740.679056167603 2020-10-26 15:22:12.%f train.py:271[27649] INFO Test 2020-10-26 15:22:59.%f train.py:254[27649] INFO Test set: Loss: 0.7238, Accuracy: 50.18%) 2020-10-26 15:22:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:22:59.%f train.py:283[27649] INFO Time: 20787.505963802338 2020-10-26 15:22:59.%f train.py:267[27649] INFO Epoch: 26 2020-10-26 15:22:59.%f train.py:268[27649] INFO Train 2020-10-26 15:25:39.%f train.py:215[27649] INFO Train epoch: 26 [200/1000] Accuracy: 50.67% Loss: 0.722108 2020-10-26 15:28:22.%f train.py:215[27649] INFO Train epoch: 26 [400/1000] Accuracy: 49.27% Loss: 0.721857 2020-10-26 15:31:02.%f train.py:215[27649] INFO Train epoch: 26 [600/1000] Accuracy: 49.83% Loss: 0.722023 2020-10-26 15:33:41.%f train.py:215[27649] INFO Train epoch: 26 [800/1000] Accuracy: 49.09% Loss: 0.722354 2020-10-26 15:36:20.%f train.py:215[27649] INFO Train epoch: 26 [1000/1000] Accuracy: 49.58% Loss: 0.721721 2020-10-26 15:36:21.%f train.py:270[27649] INFO Time: 21589.455163240433 2020-10-26 15:36:21.%f train.py:271[27649] INFO Test 2020-10-26 15:37:07.%f train.py:254[27649] INFO Test set: Loss: 0.7234, Accuracy: 50.19%) 2020-10-26 15:37:07.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:37:07.%f train.py:283[27649] INFO Time: 21636.245799064636 2020-10-26 15:37:07.%f train.py:267[27649] INFO Epoch: 27 2020-10-26 15:37:07.%f train.py:268[27649] INFO Train 2020-10-26 15:39:46.%f train.py:215[27649] INFO Train epoch: 27 [200/1000] Accuracy: 49.50% Loss: 0.721894 2020-10-26 15:42:29.%f train.py:215[27649] INFO Train epoch: 27 [400/1000] Accuracy: 49.42% Loss: 0.721731 2020-10-26 15:45:09.%f train.py:215[27649] INFO Train epoch: 27 [600/1000] Accuracy: 50.00% Loss: 0.721781 2020-10-26 15:47:49.%f train.py:215[27649] INFO Train epoch: 27 [800/1000] Accuracy: 50.19% Loss: 0.721131 2020-10-26 15:50:34.%f train.py:215[27649] INFO Train epoch: 27 [1000/1000] Accuracy: 49.39% Loss: 0.722162 2020-10-26 15:50:35.%f train.py:270[27649] INFO Time: 22443.429463863373 2020-10-26 15:50:35.%f train.py:271[27649] INFO Test 2020-10-26 15:51:18.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.22%) 2020-10-26 15:51:18.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:51:18.%f train.py:283[27649] INFO Time: 22486.526964187622 2020-10-26 15:51:18.%f train.py:267[27649] INFO Epoch: 28 2020-10-26 15:51:18.%f train.py:268[27649] INFO Train 2020-10-26 15:54:03.%f train.py:215[27649] INFO Train epoch: 28 [200/1000] Accuracy: 50.61% Loss: 0.721195 2020-10-26 15:56:45.%f train.py:215[27649] INFO Train epoch: 28 [400/1000] Accuracy: 49.16% Loss: 0.722328 2020-10-26 15:59:28.%f train.py:215[27649] INFO Train epoch: 28 [600/1000] Accuracy: 49.47% Loss: 0.721517 2020-10-26 16:02:11.%f train.py:215[27649] INFO Train epoch: 28 [800/1000] Accuracy: 49.39% Loss: 0.721710 2020-10-26 16:04:56.%f train.py:215[27649] INFO Train epoch: 28 [1000/1000] Accuracy: 48.91% Loss: 0.721964 2020-10-26 16:04:57.%f train.py:270[27649] INFO Time: 23305.4538462162 2020-10-26 16:04:57.%f train.py:271[27649] INFO Test 2020-10-26 16:05:43.%f train.py:254[27649] INFO Test set: Loss: 0.7233, Accuracy: 50.02%) 2020-10-26 16:05:43.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:05:43.%f train.py:283[27649] INFO Time: 23351.919139146805 2020-10-26 16:05:43.%f train.py:267[27649] INFO Epoch: 29 2020-10-26 16:05:43.%f train.py:268[27649] INFO Train 2020-10-26 16:08:25.%f train.py:215[27649] INFO Train epoch: 29 [200/1000] Accuracy: 50.62% Loss: 0.721156 2020-10-26 16:11:07.%f train.py:215[27649] INFO Train epoch: 29 [400/1000] Accuracy: 49.31% Loss: 0.722137 2020-10-26 16:13:48.%f train.py:215[27649] INFO Train epoch: 29 [600/1000] Accuracy: 50.30% Loss: 0.721555 2020-10-26 16:16:33.%f train.py:215[27649] INFO Train epoch: 29 [800/1000] Accuracy: 49.88% Loss: 0.721814 2020-10-26 16:19:07.%f train.py:215[27649] INFO Train epoch: 29 [1000/1000] Accuracy: 49.12% Loss: 0.721810 2020-10-26 16:19:07.%f train.py:270[27649] INFO Time: 24156.372138738632 2020-10-26 16:19:07.%f train.py:271[27649] INFO Test 2020-10-26 16:19:30.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.22%) 2020-10-26 16:19:30.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:19:30.%f train.py:283[27649] INFO Time: 24179.02462387085 2020-10-26 16:19:30.%f train.py:267[27649] INFO Epoch: 30 2020-10-26 16:19:30.%f train.py:268[27649] INFO Train 2020-10-26 16:20:48.%f train.py:215[27649] INFO Train epoch: 30 [200/1000] Accuracy: 50.88% Loss: 0.721796 2020-10-26 16:22:06.%f train.py:215[27649] INFO Train epoch: 30 [400/1000] Accuracy: 49.59% Loss: 0.722117 2020-10-26 16:23:24.%f train.py:215[27649] INFO Train epoch: 30 [600/1000] Accuracy: 50.19% Loss: 0.721511 2020-10-26 16:24:42.%f train.py:215[27649] INFO Train epoch: 30 [800/1000] Accuracy: 49.66% Loss: 0.721838 2020-10-26 16:26:00.%f train.py:215[27649] INFO Train epoch: 30 [1000/1000] Accuracy: 49.92% Loss: 0.722011 2020-10-26 16:26:00.%f train.py:270[27649] INFO Time: 24569.025196790695 2020-10-26 16:26:00.%f train.py:271[27649] INFO Test 2020-10-26 16:26:23.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.05%) 2020-10-26 16:26:23.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:26:23.%f train.py:283[27649] INFO Time: 24591.9573366642 2020-10-26 16:26:23.%f train.py:267[27649] INFO Epoch: 31 2020-10-26 16:26:23.%f train.py:268[27649] INFO Train 2020-10-26 16:27:41.%f train.py:215[27649] INFO Train epoch: 31 [200/1000] Accuracy: 50.30% Loss: 0.721678 2020-10-26 16:28:58.%f train.py:215[27649] INFO Train epoch: 31 [400/1000] Accuracy: 51.16% Loss: 0.720728 2020-10-26 16:30:16.%f train.py:215[27649] INFO Train epoch: 31 [600/1000] Accuracy: 49.17% Loss: 0.722231 2020-10-26 16:31:34.%f train.py:215[27649] INFO Train epoch: 31 [800/1000] Accuracy: 49.19% Loss: 0.722184 2020-10-26 16:32:52.%f train.py:215[27649] INFO Train epoch: 31 [1000/1000] Accuracy: 49.48% Loss: 0.721670 2020-10-26 16:32:52.%f train.py:270[27649] INFO Time: 24980.771601200104 2020-10-26 16:32:52.%f train.py:271[27649] INFO Test 2020-10-26 16:33:14.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.19%) 2020-10-26 16:33:14.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:33:14.%f train.py:283[27649] INFO Time: 25003.343010663986 2020-10-26 16:33:14.%f train.py:267[27649] INFO Epoch: 32 2020-10-26 16:33:14.%f train.py:268[27649] INFO Train 2020-10-26 16:34:32.%f train.py:215[27649] INFO Train epoch: 32 [200/1000] Accuracy: 50.62% Loss: 0.720921 2020-10-26 16:35:50.%f train.py:215[27649] INFO Train epoch: 32 [400/1000] Accuracy: 49.50% Loss: 0.721514 2020-10-26 16:37:07.%f train.py:215[27649] INFO Train epoch: 32 [600/1000] Accuracy: 50.78% Loss: 0.721799 2020-10-26 16:38:25.%f train.py:215[27649] INFO Train epoch: 32 [800/1000] Accuracy: 50.00% Loss: 0.720876 2020-10-26 16:39:43.%f train.py:215[27649] INFO Train epoch: 32 [1000/1000] Accuracy: 49.89% Loss: 0.721589 2020-10-26 16:39:43.%f train.py:270[27649] INFO Time: 25391.875609874725 2020-10-26 16:39:43.%f train.py:271[27649] INFO Test 2020-10-26 16:40:06.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.11%) 2020-10-26 16:40:06.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:40:06.%f train.py:283[27649] INFO Time: 25414.722820043564 2020-10-26 16:40:06.%f train.py:267[27649] INFO Epoch: 33 2020-10-26 16:40:06.%f train.py:268[27649] INFO Train 2020-10-26 16:41:23.%f train.py:215[27649] INFO Train epoch: 33 [200/1000] Accuracy: 49.09% Loss: 0.721560 2020-10-26 16:42:39.%f train.py:215[27649] INFO Train epoch: 33 [400/1000] Accuracy: 48.88% Loss: 0.722098 2020-10-26 16:43:56.%f train.py:215[27649] INFO Train epoch: 33 [600/1000] Accuracy: 50.31% Loss: 0.721358 2020-10-26 16:45:13.%f train.py:215[27649] INFO Train epoch: 33 [800/1000] Accuracy: 50.06% Loss: 0.721508 2020-10-26 16:46:29.%f train.py:215[27649] INFO Train epoch: 33 [1000/1000] Accuracy: 49.72% Loss: 0.722063 2020-10-26 16:46:30.%f train.py:270[27649] INFO Time: 25798.640019655228 2020-10-26 16:46:30.%f train.py:271[27649] INFO Test 2020-10-26 16:46:52.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.14%) 2020-10-26 16:46:52.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:46:52.%f train.py:283[27649] INFO Time: 25821.180437088013 2020-10-26 16:46:52.%f train.py:267[27649] INFO Epoch: 34 2020-10-26 16:46:52.%f train.py:268[27649] INFO Train 2020-10-26 16:48:09.%f train.py:215[27649] INFO Train epoch: 34 [200/1000] Accuracy: 49.45% Loss: 0.721542 2020-10-26 16:49:26.%f train.py:215[27649] INFO Train epoch: 34 [400/1000] Accuracy: 50.08% Loss: 0.721261 2020-10-26 16:50:42.%f train.py:215[27649] INFO Train epoch: 34 [600/1000] Accuracy: 51.12% Loss: 0.721381 2020-10-26 16:51:59.%f train.py:215[27649] INFO Train epoch: 34 [800/1000] Accuracy: 50.38% Loss: 0.721490 2020-10-26 16:53:15.%f train.py:215[27649] INFO Train epoch: 34 [1000/1000] Accuracy: 50.19% Loss: 0.721831 2020-10-26 16:53:16.%f train.py:270[27649] INFO Time: 26204.604976654053 2020-10-26 16:53:16.%f train.py:271[27649] INFO Test 2020-10-26 16:53:38.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.06%) 2020-10-26 16:53:38.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:53:38.%f train.py:283[27649] INFO Time: 26227.149789333344 2020-10-26 16:53:38.%f train.py:267[27649] INFO Epoch: 35 2020-10-26 16:53:38.%f train.py:268[27649] INFO Train 2020-10-26 16:54:55.%f train.py:215[27649] INFO Train epoch: 35 [200/1000] Accuracy: 50.02% Loss: 0.721548 2020-10-26 16:56:11.%f train.py:215[27649] INFO Train epoch: 35 [400/1000] Accuracy: 49.72% Loss: 0.721850 2020-10-26 16:57:28.%f train.py:215[27649] INFO Train epoch: 35 [600/1000] Accuracy: 50.05% Loss: 0.721939 2020-10-26 16:58:45.%f train.py:215[27649] INFO Train epoch: 35 [800/1000] Accuracy: 50.11% Loss: 0.722074 2020-10-26 17:00:02.%f train.py:215[27649] INFO Train epoch: 35 [1000/1000] Accuracy: 49.80% Loss: 0.721860 2020-10-26 17:00:02.%f train.py:270[27649] INFO Time: 26611.17203426361 2020-10-26 17:00:02.%f train.py:271[27649] INFO Test 2020-10-26 17:00:25.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.18%) 2020-10-26 17:00:25.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:00:25.%f train.py:283[27649] INFO Time: 26633.60524392128 2020-10-26 17:00:25.%f train.py:267[27649] INFO Epoch: 36 2020-10-26 17:00:25.%f train.py:268[27649] INFO Train 2020-10-26 17:01:41.%f train.py:215[27649] INFO Train epoch: 36 [200/1000] Accuracy: 50.58% Loss: 0.721399 2020-10-26 17:02:58.%f train.py:215[27649] INFO Train epoch: 36 [400/1000] Accuracy: 50.42% Loss: 0.721195 2020-10-26 17:04:16.%f train.py:215[27649] INFO Train epoch: 36 [600/1000] Accuracy: 49.61% Loss: 0.721957 2020-10-26 17:05:32.%f train.py:215[27649] INFO Train epoch: 36 [800/1000] Accuracy: 50.20% Loss: 0.721555 2020-10-26 17:06:49.%f train.py:215[27649] INFO Train epoch: 36 [1000/1000] Accuracy: 49.50% Loss: 0.722120 2020-10-26 17:06:49.%f train.py:270[27649] INFO Time: 27018.353071689606 2020-10-26 17:06:49.%f train.py:271[27649] INFO Test 2020-10-26 17:07:12.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.09%) 2020-10-26 17:07:12.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:07:12.%f train.py:283[27649] INFO Time: 27041.12589907646 2020-10-26 17:07:12.%f train.py:267[27649] INFO Epoch: 37 2020-10-26 17:07:12.%f train.py:268[27649] INFO Train 2020-10-26 17:08:29.%f train.py:215[27649] INFO Train epoch: 37 [200/1000] Accuracy: 51.09% Loss: 0.720909 2020-10-26 17:09:46.%f train.py:215[27649] INFO Train epoch: 37 [400/1000] Accuracy: 50.59% Loss: 0.721484 2020-10-26 17:11:02.%f train.py:215[27649] INFO Train epoch: 37 [600/1000] Accuracy: 49.84% Loss: 0.721889 2020-10-26 17:12:19.%f train.py:215[27649] INFO Train epoch: 37 [800/1000] Accuracy: 50.52% Loss: 0.721360 2020-10-26 17:13:36.%f train.py:215[27649] INFO Train epoch: 37 [1000/1000] Accuracy: 49.75% Loss: 0.721795 2020-10-26 17:13:36.%f train.py:270[27649] INFO Time: 27425.170134544373 2020-10-26 17:13:36.%f train.py:271[27649] INFO Test 2020-10-26 17:13:59.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.12%) 2020-10-26 17:13:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:13:59.%f train.py:283[27649] INFO Time: 27447.889960050583 2020-10-26 17:13:59.%f train.py:267[27649] INFO Epoch: 38 2020-10-26 17:13:59.%f train.py:268[27649] INFO Train 2020-10-26 17:15:16.%f train.py:215[27649] INFO Train epoch: 38 [200/1000] Accuracy: 49.08% Loss: 0.721949 2020-10-26 17:16:33.%f train.py:215[27649] INFO Train epoch: 38 [400/1000] Accuracy: 49.73% Loss: 0.722057 2020-10-26 17:17:50.%f train.py:215[27649] INFO Train epoch: 38 [600/1000] Accuracy: 50.83% Loss: 0.720905 2020-10-26 17:19:06.%f train.py:215[27649] INFO Train epoch: 38 [800/1000] Accuracy: 50.88% Loss: 0.720798 2020-10-26 17:20:23.%f train.py:215[27649] INFO Train epoch: 38 [1000/1000] Accuracy: 49.64% Loss: 0.721864 2020-10-26 17:20:23.%f train.py:270[27649] INFO Time: 27832.214666843414 2020-10-26 17:20:23.%f train.py:271[27649] INFO Test 2020-10-26 17:20:46.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.00%) 2020-10-26 17:20:46.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:20:46.%f train.py:283[27649] INFO Time: 27854.788491249084 2020-10-26 17:20:46.%f train.py:267[27649] INFO Epoch: 39 2020-10-26 17:20:46.%f train.py:268[27649] INFO Train 2020-10-26 17:22:03.%f train.py:215[27649] INFO Train epoch: 39 [200/1000] Accuracy: 50.08% Loss: 0.721081 2020-10-26 17:23:20.%f train.py:215[27649] INFO Train epoch: 39 [400/1000] Accuracy: 50.06% Loss: 0.721671 2020-10-26 17:24:37.%f train.py:215[27649] INFO Train epoch: 39 [600/1000] Accuracy: 48.83% Loss: 0.722440 2020-10-26 17:25:54.%f train.py:215[27649] INFO Train epoch: 39 [800/1000] Accuracy: 49.69% Loss: 0.721849 2020-10-26 17:27:10.%f train.py:215[27649] INFO Train epoch: 39 [1000/1000] Accuracy: 50.22% Loss: 0.721833 2020-10-26 17:27:11.%f train.py:270[27649] INFO Time: 28239.423560619354 2020-10-26 17:27:11.%f train.py:271[27649] INFO Test 2020-10-26 17:27:33.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.14%) 2020-10-26 17:27:33.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:27:33.%f train.py:283[27649] INFO Time: 28261.885771036148 2020-10-26 17:27:33.%f train.py:267[27649] INFO Epoch: 40 2020-10-26 17:27:33.%f train.py:268[27649] INFO Train 2020-10-26 17:28:50.%f train.py:215[27649] INFO Train epoch: 40 [200/1000] Accuracy: 50.67% Loss: 0.721462 2020-10-26 17:30:07.%f train.py:215[27649] INFO Train epoch: 40 [400/1000] Accuracy: 50.20% Loss: 0.721098 2020-10-26 17:31:24.%f train.py:215[27649] INFO Train epoch: 40 [600/1000] Accuracy: 50.61% Loss: 0.721035 2020-10-26 17:32:41.%f train.py:215[27649] INFO Train epoch: 40 [800/1000] Accuracy: 49.78% Loss: 0.721563 2020-10-26 17:33:59.%f train.py:215[27649] INFO Train epoch: 40 [1000/1000] Accuracy: 49.39% Loss: 0.721895 2020-10-26 17:33:59.%f train.py:270[27649] INFO Time: 28647.86800456047 2020-10-26 17:33:59.%f train.py:271[27649] INFO Test 2020-10-26 17:34:22.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.05%) 2020-10-26 17:34:22.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:34:22.%f train.py:283[27649] INFO Time: 28670.392629146576 2020-10-26 17:34:22.%f train.py:267[27649] INFO Epoch: 41 2020-10-26 17:34:22.%f train.py:268[27649] INFO Train 2020-10-26 17:35:39.%f train.py:215[27649] INFO Train epoch: 41 [200/1000] Accuracy: 49.88% Loss: 0.721476 2020-10-26 17:36:56.%f train.py:215[27649] INFO Train epoch: 41 [400/1000] Accuracy: 50.09% Loss: 0.721099 2020-10-26 17:38:13.%f train.py:215[27649] INFO Train epoch: 41 [600/1000] Accuracy: 49.31% Loss: 0.721230 2020-10-26 17:39:31.%f train.py:215[27649] INFO Train epoch: 41 [800/1000] Accuracy: 50.80% Loss: 0.721088 2020-10-26 17:40:48.%f train.py:215[27649] INFO Train epoch: 41 [1000/1000] Accuracy: 50.23% Loss: 0.721994 2020-10-26 17:40:48.%f train.py:270[27649] INFO Time: 29057.317686080933 2020-10-26 17:40:48.%f train.py:271[27649] INFO Test 2020-10-26 17:41:11.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.08%) 2020-10-26 17:41:11.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:41:11.%f train.py:283[27649] INFO Time: 29079.925040006638 2020-10-26 17:41:11.%f train.py:267[27649] INFO Epoch: 42 2020-10-26 17:41:11.%f train.py:268[27649] INFO Train 2020-10-26 17:42:28.%f train.py:215[27649] INFO Train epoch: 42 [200/1000] Accuracy: 50.16% Loss: 0.721571 2020-10-26 17:43:45.%f train.py:215[27649] INFO Train epoch: 42 [400/1000] Accuracy: 49.53% Loss: 0.721621 2020-10-26 17:45:02.%f train.py:215[27649] INFO Train epoch: 42 [600/1000] Accuracy: 50.61% Loss: 0.721240 2020-10-26 17:46:19.%f train.py:215[27649] INFO Train epoch: 42 [800/1000] Accuracy: 49.98% Loss: 0.721345 2020-10-26 17:47:37.%f train.py:215[27649] INFO Train epoch: 42 [1000/1000] Accuracy: 49.59% Loss: 0.721730 2020-10-26 17:47:37.%f train.py:270[27649] INFO Time: 29465.678948640823 2020-10-26 17:47:37.%f train.py:271[27649] INFO Test 2020-10-26 17:47:59.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 49.94%) 2020-10-26 17:47:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:47:59.%f train.py:283[27649] INFO Time: 29488.179372787476 2020-10-26 17:47:59.%f train.py:267[27649] INFO Epoch: 43 2020-10-26 17:47:59.%f train.py:268[27649] INFO Train 2020-10-26 17:49:16.%f train.py:215[27649] INFO Train epoch: 43 [200/1000] Accuracy: 48.73% Loss: 0.721529 2020-10-26 17:50:33.%f train.py:215[27649] INFO Train epoch: 43 [400/1000] Accuracy: 49.44% Loss: 0.721806 2020-10-26 17:51:50.%f train.py:215[27649] INFO Train epoch: 43 [600/1000] Accuracy: 50.03% Loss: 0.721423 2020-10-26 17:53:07.%f train.py:215[27649] INFO Train epoch: 43 [800/1000] Accuracy: 50.31% Loss: 0.721637 2020-10-26 17:54:23.%f train.py:215[27649] INFO Train epoch: 43 [1000/1000] Accuracy: 49.55% Loss: 0.721455 2020-10-26 17:54:24.%f train.py:270[27649] INFO Time: 29872.40514731407 2020-10-26 17:54:24.%f train.py:271[27649] INFO Test 2020-10-26 17:54:46.%f train.py:254[27649] INFO Test set: Loss: 0.7224, Accuracy: 50.02%) 2020-10-26 17:54:46.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:54:46.%f train.py:283[27649] INFO Time: 29894.952974557877 2020-10-26 17:54:46.%f train.py:267[27649] INFO Epoch: 44 2020-10-26 17:54:46.%f train.py:268[27649] INFO Train 2020-10-26 17:56:02.%f train.py:215[27649] INFO Train epoch: 44 [200/1000] Accuracy: 49.70% Loss: 0.720927 2020-10-26 17:57:19.%f train.py:215[27649] INFO Train epoch: 44 [400/1000] Accuracy: 49.70% Loss: 0.721201 2020-10-26 17:58:36.%f train.py:215[27649] INFO Train epoch: 44 [600/1000] Accuracy: 50.69% Loss: 0.721111 2020-10-26 17:59:52.%f train.py:215[27649] INFO Train epoch: 44 [800/1000] Accuracy: 50.77% Loss: 0.720785 2020-10-26 18:01:09.%f train.py:215[27649] INFO Train epoch: 44 [1000/1000] Accuracy: 50.03% Loss: 0.721340 2020-10-26 18:01:09.%f train.py:270[27649] INFO Time: 30277.745666980743 2020-10-26 18:01:09.%f train.py:271[27649] INFO Test 2020-10-26 18:01:31.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 50.11%) 2020-10-26 18:01:31.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:01:31.%f train.py:283[27649] INFO Time: 30300.194601535797 2020-10-26 18:01:31.%f train.py:267[27649] INFO Epoch: 45 2020-10-26 18:01:31.%f train.py:268[27649] INFO Train 2020-10-26 18:02:48.%f train.py:215[27649] INFO Train epoch: 45 [200/1000] Accuracy: 49.02% Loss: 0.721996 2020-10-26 18:04:04.%f train.py:215[27649] INFO Train epoch: 45 [400/1000] Accuracy: 50.08% Loss: 0.721470 2020-10-26 18:05:21.%f train.py:215[27649] INFO Train epoch: 45 [600/1000] Accuracy: 50.27% Loss: 0.721050 2020-10-26 18:06:37.%f train.py:215[27649] INFO Train epoch: 45 [800/1000] Accuracy: 50.50% Loss: 0.721582 2020-10-26 18:07:54.%f train.py:215[27649] INFO Train epoch: 45 [1000/1000] Accuracy: 49.67% Loss: 0.721660 2020-10-26 18:07:54.%f train.py:270[27649] INFO Time: 30683.1965944767 2020-10-26 18:07:54.%f train.py:271[27649] INFO Test 2020-10-26 18:08:17.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.00%) 2020-10-26 18:08:17.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:08:17.%f train.py:283[27649] INFO Time: 30705.85431098938 2020-10-26 18:08:17.%f train.py:267[27649] INFO Epoch: 46 2020-10-26 18:08:17.%f train.py:268[27649] INFO Train 2020-10-26 18:09:34.%f train.py:215[27649] INFO Train epoch: 46 [200/1000] Accuracy: 50.27% Loss: 0.721508 2020-10-26 18:10:50.%f train.py:215[27649] INFO Train epoch: 46 [400/1000] Accuracy: 50.97% Loss: 0.721218 2020-10-26 18:12:07.%f train.py:215[27649] INFO Train epoch: 46 [600/1000] Accuracy: 50.03% Loss: 0.721920 2020-10-26 18:13:24.%f train.py:215[27649] INFO Train epoch: 46 [800/1000] Accuracy: 50.38% Loss: 0.720731 2020-10-26 18:14:40.%f train.py:215[27649] INFO Train epoch: 46 [1000/1000] Accuracy: 50.38% Loss: 0.721281 2020-10-26 18:14:41.%f train.py:270[27649] INFO Time: 31089.40986776352 2020-10-26 18:14:41.%f train.py:271[27649] INFO Test 2020-10-26 18:15:03.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 49.96%) 2020-10-26 18:15:03.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:15:03.%f train.py:283[27649] INFO Time: 31111.84361052513 2020-10-26 18:15:03.%f train.py:267[27649] INFO Epoch: 47 2020-10-26 18:15:03.%f train.py:268[27649] INFO Train 2020-10-26 18:16:20.%f train.py:215[27649] INFO Train epoch: 47 [200/1000] Accuracy: 49.14% Loss: 0.721514 2020-10-26 18:17:36.%f train.py:215[27649] INFO Train epoch: 47 [400/1000] Accuracy: 49.91% Loss: 0.721505 2020-10-26 18:18:53.%f train.py:215[27649] INFO Train epoch: 47 [600/1000] Accuracy: 49.98% Loss: 0.721421 2020-10-26 18:20:10.%f train.py:215[27649] INFO Train epoch: 47 [800/1000] Accuracy: 49.73% Loss: 0.721589 2020-10-26 18:21:26.%f train.py:215[27649] INFO Train epoch: 47 [1000/1000] Accuracy: 49.84% Loss: 0.721595 2020-10-26 18:21:27.%f train.py:270[27649] INFO Time: 31495.542852640152 2020-10-26 18:21:27.%f train.py:271[27649] INFO Test 2020-10-26 18:21:49.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.20%) 2020-10-26 18:21:49.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:21:49.%f train.py:283[27649] INFO Time: 31518.005761146545 2020-10-26 18:21:49.%f train.py:267[27649] INFO Epoch: 48 2020-10-26 18:21:49.%f train.py:268[27649] INFO Train 2020-10-26 18:23:06.%f train.py:215[27649] INFO Train epoch: 48 [200/1000] Accuracy: 49.94% Loss: 0.721164 2020-10-26 18:24:22.%f train.py:215[27649] INFO Train epoch: 48 [400/1000] Accuracy: 51.23% Loss: 0.721384 2020-10-26 18:25:39.%f train.py:215[27649] INFO Train epoch: 48 [600/1000] Accuracy: 49.59% Loss: 0.721872 2020-10-26 18:26:55.%f train.py:215[27649] INFO Train epoch: 48 [800/1000] Accuracy: 49.88% Loss: 0.721526 2020-10-26 18:28:12.%f train.py:215[27649] INFO Train epoch: 48 [1000/1000] Accuracy: 49.00% Loss: 0.721775 2020-10-26 18:28:12.%f train.py:270[27649] INFO Time: 31901.144391536713 2020-10-26 18:28:12.%f train.py:271[27649] INFO Test 2020-10-26 18:28:35.%f train.py:254[27649] INFO Test set: Loss: 0.7226, Accuracy: 50.11%) 2020-10-26 18:28:35.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:28:35.%f train.py:283[27649] INFO Time: 31923.629400491714 2020-10-26 18:28:35.%f train.py:267[27649] INFO Epoch: 49 2020-10-26 18:28:35.%f train.py:268[27649] INFO Train 2020-10-26 18:29:51.%f train.py:215[27649] INFO Train epoch: 49 [200/1000] Accuracy: 50.20% Loss: 0.721193 2020-10-26 18:31:08.%f train.py:215[27649] INFO Train epoch: 49 [400/1000] Accuracy: 48.86% Loss: 0.721818 2020-10-26 18:32:25.%f train.py:215[27649] INFO Train epoch: 49 [600/1000] Accuracy: 49.16% Loss: 0.721519 2020-10-26 18:33:41.%f train.py:215[27649] INFO Train epoch: 49 [800/1000] Accuracy: 49.73% Loss: 0.721708 2020-10-26 18:34:58.%f train.py:215[27649] INFO Train epoch: 49 [1000/1000] Accuracy: 49.69% Loss: 0.721645 2020-10-26 18:34:58.%f train.py:270[27649] INFO Time: 32306.78603386879 2020-10-26 18:34:58.%f train.py:271[27649] INFO Test 2020-10-26 18:35:20.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.19%) 2020-10-26 18:35:20.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:35:20.%f train.py:283[27649] INFO Time: 32329.21870279312 2020-10-26 18:35:20.%f train.py:267[27649] INFO Epoch: 50 2020-10-26 18:35:20.%f train.py:268[27649] INFO Train 2020-10-26 18:36:37.%f train.py:215[27649] INFO Train epoch: 50 [200/1000] Accuracy: 50.05% Loss: 0.721832 2020-10-26 18:37:53.%f train.py:215[27649] INFO Train epoch: 50 [400/1000] Accuracy: 49.45% Loss: 0.721136 2020-10-26 18:39:10.%f train.py:215[27649] INFO Train epoch: 50 [600/1000] Accuracy: 49.75% Loss: 0.721008 2020-10-26 18:40:26.%f train.py:215[27649] INFO Train epoch: 50 [800/1000] Accuracy: 49.98% Loss: 0.721070 2020-10-26 18:41:43.%f train.py:215[27649] INFO Train epoch: 50 [1000/1000] Accuracy: 49.12% Loss: 0.721500 2020-10-26 18:41:44.%f train.py:270[27649] INFO Time: 32712.40008211136 2020-10-26 18:41:44.%f train.py:271[27649] INFO Test 2020-10-26 18:42:06.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.18%) 2020-10-26 18:42:06.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:42:06.%f train.py:283[27649] INFO Time: 32734.86405801773 2020-10-26 18:42:06.%f train.py:267[27649] INFO Epoch: 51 2020-10-26 18:42:06.%f train.py:268[27649] INFO Train 2020-10-26 18:43:24.%f train.py:215[27649] INFO Train epoch: 51 [200/1000] Accuracy: 51.00% Loss: 0.721218 2020-10-26 18:44:42.%f train.py:215[27649] INFO Train epoch: 51 [400/1000] Accuracy: 50.23% Loss: 0.720999 2020-10-26 18:45:59.%f train.py:215[27649] INFO Train epoch: 51 [600/1000] Accuracy: 49.58% Loss: 0.722122 2020-10-26 18:47:16.%f train.py:215[27649] INFO Train epoch: 51 [800/1000] Accuracy: 50.22% Loss: 0.721728 2020-10-26 18:48:34.%f train.py:215[27649] INFO Train epoch: 51 [1000/1000] Accuracy: 49.73% Loss: 0.721392 2020-10-26 18:48:34.%f train.py:270[27649] INFO Time: 33122.7360265255 2020-10-26 18:48:34.%f train.py:271[27649] INFO Test 2020-10-26 18:48:56.%f train.py:254[27649] INFO Test set: Loss: 0.7222, Accuracy: 50.06%) 2020-10-26 18:48:56.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:48:56.%f train.py:283[27649] INFO Time: 33145.32122063637 2020-10-26 18:48:56.%f train.py:267[27649] INFO Epoch: 52 2020-10-26 18:48:56.%f train.py:268[27649] INFO Train 2020-10-26 18:50:13.%f train.py:215[27649] INFO Train epoch: 52 [200/1000] Accuracy: 50.22% Loss: 0.721378 2020-10-26 18:51:30.%f train.py:215[27649] INFO Train epoch: 52 [400/1000] Accuracy: 49.62% Loss: 0.721496 2020-10-26 18:52:48.%f train.py:215[27649] INFO Train epoch: 52 [600/1000] Accuracy: 50.80% Loss: 0.721010 2020-10-26 18:54:05.%f train.py:215[27649] INFO Train epoch: 52 [800/1000] Accuracy: 50.55% Loss: 0.721563 2020-10-26 18:55:25.%f train.py:215[27649] INFO Train epoch: 52 [1000/1000] Accuracy: 50.56% Loss: 0.721192 2020-10-26 18:55:25.%f train.py:270[27649] INFO Time: 33534.01327371597 2020-10-26 18:55:25.%f train.py:271[27649] INFO Test 2020-10-26 18:55:50.%f train.py:254[27649] INFO Test set: Loss: 0.7222, Accuracy: 50.06%) 2020-10-26 18:55:50.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:55:50.%f train.py:283[27649] INFO Time: 33558.541908741 2020-10-26 18:55:50.%f train.py:267[27649] INFO Epoch: 53 2020-10-26 18:55:50.%f train.py:268[27649] INFO Train 2020-10-26 18:57:10.%f train.py:215[27649] INFO Train epoch: 53 [200/1000] Accuracy: 51.30% Loss: 0.720911 2020-10-26 18:58:29.%f train.py:215[27649] INFO Train epoch: 53 [400/1000] Accuracy: 48.81% Loss: 0.721708 2020-10-26 18:59:48.%f train.py:215[27649] INFO Train epoch: 53 [600/1000] Accuracy: 49.72% Loss: 0.721488 2020-10-26 19:01:05.%f train.py:215[27649] INFO Train epoch: 53 [800/1000] Accuracy: 49.03% Loss: 0.721965 2020-10-26 19:02:22.%f train.py:215[27649] INFO Train epoch: 53 [1000/1000] Accuracy: 49.03% Loss: 0.721585 2020-10-26 19:02:23.%f train.py:270[27649] INFO Time: 33951.554489851 2020-10-26 19:02:23.%f train.py:271[27649] INFO Test 2020-10-26 19:02:45.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.04%) 2020-10-26 19:02:45.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:02:45.%f train.py:283[27649] INFO Time: 33974.32499027252 2020-10-26 19:02:45.%f train.py:267[27649] INFO Epoch: 54 2020-10-26 19:02:45.%f train.py:268[27649] INFO Train 2020-10-26 19:04:02.%f train.py:215[27649] INFO Train epoch: 54 [200/1000] Accuracy: 50.77% Loss: 0.721066 2020-10-26 19:05:20.%f train.py:215[27649] INFO Train epoch: 54 [400/1000] Accuracy: 50.33% Loss: 0.721573 2020-10-26 19:06:38.%f train.py:215[27649] INFO Train epoch: 54 [600/1000] Accuracy: 49.39% Loss: 0.721073 2020-10-26 19:07:55.%f train.py:215[27649] INFO Train epoch: 54 [800/1000] Accuracy: 48.78% Loss: 0.721267 2020-10-26 19:09:12.%f train.py:215[27649] INFO Train epoch: 54 [1000/1000] Accuracy: 49.88% Loss: 0.721365 2020-10-26 19:09:12.%f train.py:270[27649] INFO Time: 34361.14559984207 2020-10-26 19:09:12.%f train.py:271[27649] INFO Test 2020-10-26 19:09:35.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.29%) 2020-10-26 19:09:35.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:09:35.%f train.py:283[27649] INFO Time: 34384.130383491516 2020-10-26 19:09:35.%f train.py:267[27649] INFO Epoch: 55 2020-10-26 19:09:35.%f train.py:268[27649] INFO Train 2020-10-26 19:10:52.%f train.py:215[27649] INFO Train epoch: 55 [200/1000] Accuracy: 49.23% Loss: 0.721556 2020-10-26 19:12:09.%f train.py:215[27649] INFO Train epoch: 55 [400/1000] Accuracy: 50.66% Loss: 0.720750 2020-10-26 19:13:26.%f train.py:215[27649] INFO Train epoch: 55 [600/1000] Accuracy: 50.25% Loss: 0.720918 2020-10-26 19:14:43.%f train.py:215[27649] INFO Train epoch: 55 [800/1000] Accuracy: 49.09% Loss: 0.721661 2020-10-26 19:16:00.%f train.py:215[27649] INFO Train epoch: 55 [1000/1000] Accuracy: 50.88% Loss: 0.720696 2020-10-26 19:16:00.%f train.py:270[27649] INFO Time: 34769.300970077515 2020-10-26 19:16:00.%f train.py:271[27649] INFO Test 2020-10-26 19:16:23.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.30%) 2020-10-26 19:16:23.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:16:23.%f train.py:283[27649] INFO Time: 34792.19325590134 2020-10-26 19:16:23.%f train.py:267[27649] INFO Epoch: 56 2020-10-26 19:16:23.%f train.py:268[27649] INFO Train 2020-10-26 19:17:40.%f train.py:215[27649] INFO Train epoch: 56 [200/1000] Accuracy: 49.16% Loss: 0.721391 2020-10-26 19:18:57.%f train.py:215[27649] INFO Train epoch: 56 [400/1000] Accuracy: 49.30% Loss: 0.721471 2020-10-26 19:20:14.%f train.py:215[27649] INFO Train epoch: 56 [600/1000] Accuracy: 50.09% Loss: 0.720977 2020-10-26 19:21:31.%f train.py:215[27649] INFO Train epoch: 56 [800/1000] Accuracy: 50.50% Loss: 0.721714 2020-10-26 19:22:49.%f train.py:215[27649] INFO Train epoch: 56 [1000/1000] Accuracy: 49.75% Loss: 0.721326 2020-10-26 19:22:49.%f train.py:270[27649] INFO Time: 35177.6729080677 2020-10-26 19:22:49.%f train.py:271[27649] INFO Test 2020-10-26 19:23:12.%f train.py:254[27649] INFO Test set: Loss: 0.7224, Accuracy: 50.38%) 2020-10-26 19:23:12.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:23:12.%f train.py:283[27649] INFO Time: 35200.54278087616 2020-10-26 19:23:12.%f train.py:267[27649] INFO Epoch: 57 2020-10-26 19:23:12.%f train.py:268[27649] INFO Train 2020-10-26 19:24:29.%f train.py:215[27649] INFO Train epoch: 57 [200/1000] Accuracy: 49.30% Loss: 0.721213 2020-10-26 19:25:46.%f train.py:215[27649] INFO Train epoch: 57 [400/1000] Accuracy: 50.39% Loss: 0.721261 2020-10-26 19:27:03.%f train.py:215[27649] INFO Train epoch: 57 [600/1000] Accuracy: 50.08% Loss: 0.720789 2020-10-26 19:28:20.%f train.py:215[27649] INFO Train epoch: 57 [800/1000] Accuracy: 49.72% Loss: 0.721149 2020-10-26 19:29:37.%f train.py:215[27649] INFO Train epoch: 57 [1000/1000] Accuracy: 50.05% Loss: 0.721388 2020-10-26 19:29:37.%f train.py:270[27649] INFO Time: 35586.03649306297 2020-10-26 19:29:37.%f train.py:271[27649] INFO Test 2020-10-26 19:30:00.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.09%) 2020-10-26 19:30:00.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:30:00.%f train.py:283[27649] INFO Time: 35608.92472934723 2020-10-26 19:30:00.%f train.py:267[27649] INFO Epoch: 58 2020-10-26 19:30:00.%f train.py:268[27649] INFO Train 2020-10-26 19:31:17.%f train.py:215[27649] INFO Train epoch: 58 [200/1000] Accuracy: 50.05% Loss: 0.721109 2020-10-26 19:32:34.%f train.py:215[27649] INFO Train epoch: 58 [400/1000] Accuracy: 49.72% Loss: 0.721442 2020-10-26 19:33:51.%f train.py:215[27649] INFO Train epoch: 58 [600/1000] Accuracy: 49.86% Loss: 0.721890 2020-10-26 19:35:08.%f train.py:215[27649] INFO Train epoch: 58 [800/1000] Accuracy: 51.59% Loss: 0.720619 2020-10-26 19:36:25.%f train.py:215[27649] INFO Train epoch: 58 [1000/1000] Accuracy: 50.92% Loss: 0.720742 2020-10-26 19:36:26.%f train.py:270[27649] INFO Time: 35994.47965550423 2020-10-26 19:36:26.%f train.py:271[27649] INFO Test 2020-10-26 19:36:48.%f train.py:254[27649] INFO Test set: Loss: 0.7221, Accuracy: 50.05%) 2020-10-26 19:36:48.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:36:48.%f train.py:283[27649] INFO Time: 36017.25994181633 2020-10-26 19:36:48.%f train.py:267[27649] INFO Epoch: 59 2020-10-26 19:36:48.%f train.py:268[27649] INFO Train 2020-10-26 19:38:05.%f train.py:215[27649] INFO Train epoch: 59 [200/1000] Accuracy: 49.64% Loss: 0.720858 2020-10-26 19:39:22.%f train.py:215[27649] INFO Train epoch: 59 [400/1000] Accuracy: 49.27% Loss: 0.721158 2020-10-26 19:40:39.%f train.py:215[27649] INFO Train epoch: 59 [600/1000] Accuracy: 50.83% Loss: 0.721004 2020-10-26 19:41:56.%f train.py:215[27649] INFO Train epoch: 59 [800/1000] Accuracy: 49.95% Loss: 0.721280 2020-10-26 19:43:13.%f train.py:215[27649] INFO Train epoch: 59 [1000/1000] Accuracy: 49.91% Loss: 0.721499 2020-10-26 19:43:14.%f train.py:270[27649] INFO Time: 36402.47468829155 2020-10-26 19:43:14.%f train.py:271[27649] INFO Test 2020-10-26 19:43:36.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.00%) 2020-10-26 19:43:36.%f
可能说2080ti更适合打游戏太武断啦!
不过之前有小伙伴也反馈过类似的问题,复现的结果是确实2080ti不行,猜测是浮点计算精度可能不够? 因为该任务可能对精度有很高的要求,后续尝试过两种解决方案都可以: (1)用一个预训练模型带一下,就可以收敛了 (2)换不是2080ti的机器
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月27日(星期二) 下午4:01 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
非常感谢您的回复!!! 我使用的GPU型号为 GeForece RTX 2080Ti,应该不存在使用游戏显卡进行训练的问题。
以下具体描述一下本人的实验的相关配置,以及目前的实验结果。
实验配置 系统:ubuntu 18.04 CUDA version 11.0 python 3.7.9 torch 1.4.0 torchaudio 0.4.0 torchvision 0.4.0 使用的代码为您提供的SiaStegNet 按照您的README中提供的启动方式进行启动,使用的除了train以及valid的cover和stego目录(4个目录)参数进行了替换,model使用kenet
实验数据集构建 本人构建数据集按照fridrish论文SRNet中的实验数据集进行了构造, 使用了BOSSbase以及BOWS一共20000张图片,选取BOSSbase中的4000张作为valid cover,剩下的16000张作为 train cover。利用fridrish官网提供的 S-UNIWARD算法,以0.4的payload生成了以上(16000:4000)对应的stego用于训练。
附件是训练了200多个epoch的结果,以下是其中的一部分信息: 2020-10-27 15:46:52.%f train.py:267[27649] INFO Epoch: 239 2020-10-27 15:46:52.%f train.py:268[27649] INFO Train 2020-10-27 15:48:09.%f train.py:215[27649] INFO Train epoch: 239 [200/1000] Accuracy: 50.59% Loss: 0.720610 2020-10-27 15:49:27.%f train.py:215[27649] INFO Train epoch: 239 [400/1000] Accuracy: 50.73% Loss: 0.720714 2020-10-27 15:50:45.%f train.py:215[27649] INFO Train epoch: 239 [600/1000] Accuracy: 49.02% Loss: 0.721326 2020-10-27 15:52:02.%f train.py:215[27649] INFO Train epoch: 239 [800/1000] Accuracy: 49.77% Loss: 0.720646 2020-10-27 15:53:19.%f train.py:215[27649] INFO Train epoch: 239 [1000/1000] Accuracy: 50.11% Loss: 0.720887 2020-10-27 15:53:19.%f train.py:270[27649] INFO Time: 109008.01739406586 2020-10-27 15:53:19.%f train.py:271[27649] INFO Test 2020-10-27 15:53:42.%f train.py:254[27649] INFO Test set: Loss: 0.7220, Accuracy: 50.04%) 2020-10-27 15:53:42.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-27 15:53:42.%f train.py:283[27649] INFO Time: 109030.6237590313
如果您有时间非常感谢指导!!!
无论如何非常感谢您的回复,祝您工作顺利,完事如意!
SiaStg <notifications@github.com> 于2020年10月27日周二 下午2:27写道:
> 您好!感谢您对我的研究成果感兴趣。 > > > 根据以往的经验和其他使用者的反馈,一般使用已设定好的默认的参数,构建正常的训练数据集、验证数据集,就可以取得很好的效果。 > 唯一需要注意的是:不能用不适用计算的显卡(比如Nvidia 2080等只适合打游戏的显卡),是否您用了这类计算资源呢? > > > 期待您更多的反馈信息 > > > 祝好 > > > > > ------------------ 原始邮件 ------------------ > 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; > 发送时间: 2020年10月26日(星期一) 晚上8:10 > 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; > 抄送: "Subscribed"<subscribed@noreply.github.com>; > 主题: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) > > > > > > > 作者你好,我想问一下关于这篇论文如何具体构造数据集,如何进行训练。 > 鉴于本人水平有限,使用该repo的代码进行实验的时候不能取得很好的效果。 > > 关于如何设置学习率、如何构建训练数据集、验证数据集,具体如何训练的问题希望能够稍微指导一下。 > > 感谢 > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub, or unsubscribe. > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > <https://github.com/SiaStg/SiaStegNet/issues/1#issuecomment-717016277>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ALXJEEQM3DXN26KTCXJUTF3SMZR67ANCNFSM4S7JJCIA> > . >
2020-10-26 09:36:28.%f env.py:27[27649] INFO Using a generated random seed 28362477 2020-10-26 09:36:28.%f train.py:85[27649] INFO Command Line Arguments: Namespace(alpha=0.1, batch_size=32, ckpt_dir='./kenet_result', cuda=True, epoch=500, eps=1e-08, finetune=None, gpu_id=0, log_interval=200, lr=0.001, lr_str=2, margin=1.0, model='kenet', num_workers=0, random_crop=False, random_crop_train=False, seed=-1, train_cover_dir='/mnt/sda4/datasets/netdata/train/cover', train_stego_dir='/mnt/sda4/datasets/netdata/train/stego', val_cover_dir='/mnt/sda4/datasets/netdata/valid/cover', val_stego_dir='/mnt/sda4/datasets/netdata/valid/stego', wd=0.0001) 2020-10-26 09:36:28.%f train.py:91[27649] INFO Building data loader 2020-10-26 09:36:29.%f dataloader.py:105[27649] INFO Training set length is 32000 2020-10-26 09:36:29.%f dataloader.py:106[27649] INFO Training epoch length is 1000 2020-10-26 09:36:29.%f dataloader.py:152[27649] INFO Testing set length is 8000 2020-10-26 09:36:29.%f train.py:109[27649] INFO Building model 2020-10-26 09:36:31.%f train.py:267[27649] INFO Epoch: 1 2020-10-26 09:36:31.%f train.py:268[27649] INFO Train 2020-10-26 09:37:43.%f train.py:215[27649] INFO Train epoch: 1 [200/1000] Accuracy: 50.25% Loss: 0.734632 2020-10-26 09:38:57.%f train.py:215[27649] INFO Train epoch: 1 [400/1000] Accuracy: 49.83% Loss: 0.724080 2020-10-26 09:41:11.%f train.py:215[27649] INFO Train epoch: 1 [600/1000] Accuracy: 49.33% Loss: 0.724804 2020-10-26 09:43:47.%f train.py:215[27649] INFO Train epoch: 1 [800/1000] Accuracy: 50.06% Loss: 0.723864 2020-10-26 09:46:29.%f train.py:215[27649] INFO Train epoch: 1 [1000/1000] Accuracy: 50.41% Loss: 0.723972 2020-10-26 09:46:29.%f train.py:270[27649] INFO Time: 598.1813097000122 2020-10-26 09:46:29.%f train.py:271[27649] INFO Test 2020-10-26 09:47:13.%f train.py:254[27649] INFO Test set: Loss: 0.7251, Accuracy: 50.12%) 2020-10-26 09:47:13.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 09:47:13.%f train.py:283[27649] INFO Time: 641.4804406166077 2020-10-26 09:47:13.%f train.py:267[27649] INFO Epoch: 2 2020-10-26 09:47:13.%f train.py:268[27649] INFO Train 2020-10-26 09:49:56.%f train.py:215[27649] INFO Train epoch: 2 [200/1000] Accuracy: 50.08% Loss: 0.724571 2020-10-26 09:52:45.%f train.py:215[27649] INFO Train epoch: 2 [400/1000] Accuracy: 51.30% Loss: 0.723082 2020-10-26 09:55:23.%f train.py:215[27649] INFO Train epoch: 2 [600/1000] Accuracy: 48.69% Loss: 0.724915 2020-10-26 09:58:04.%f train.py:215[27649] INFO Train epoch: 2 [800/1000] Accuracy: 50.47% Loss: 0.722644 2020-10-26 10:00:44.%f train.py:215[27649] INFO Train epoch: 2 [1000/1000] Accuracy: 50.42% Loss: 0.722950 2020-10-26 10:00:45.%f train.py:270[27649] INFO Time: 1453.9544608592987 2020-10-26 10:00:45.%f train.py:271[27649] INFO Test 2020-10-26 10:01:32.%f train.py:254[27649] INFO Test set: Loss: 0.7250, Accuracy: 50.05%) 2020-10-26 10:01:32.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:01:32.%f train.py:283[27649] INFO Time: 1500.8485856056213 2020-10-26 10:01:32.%f train.py:267[27649] INFO Epoch: 3 2020-10-26 10:01:32.%f train.py:268[27649] INFO Train 2020-10-26 10:04:10.%f train.py:215[27649] INFO Train epoch: 3 [200/1000] Accuracy: 49.70% Loss: 0.724378 2020-10-26 10:06:42.%f train.py:215[27649] INFO Train epoch: 3 [400/1000] Accuracy: 49.97% Loss: 0.724390 2020-10-26 10:09:20.%f train.py:215[27649] INFO Train epoch: 3 [600/1000] Accuracy: 49.53% Loss: 0.724867 2020-10-26 10:12:00.%f train.py:215[27649] INFO Train epoch: 3 [800/1000] Accuracy: 49.56% Loss: 0.723496 2020-10-26 10:14:39.%f train.py:215[27649] INFO Train epoch: 3 [1000/1000] Accuracy: 50.38% Loss: 0.722862 2020-10-26 10:14:40.%f train.py:270[27649] INFO Time: 2288.688045501709 2020-10-26 10:14:40.%f train.py:271[27649] INFO Test 2020-10-26 10:15:27.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.06%) 2020-10-26 10:15:27.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:15:27.%f train.py:283[27649] INFO Time: 2336.0412130355835 2020-10-26 10:15:27.%f train.py:267[27649] INFO Epoch: 4 2020-10-26 10:15:27.%f train.py:268[27649] INFO Train 2020-10-26 10:18:08.%f train.py:215[27649] INFO Train epoch: 4 [200/1000] Accuracy: 50.41% Loss: 0.723008 2020-10-26 10:20:53.%f train.py:215[27649] INFO Train epoch: 4 [400/1000] Accuracy: 50.16% Loss: 0.724553 2020-10-26 10:23:33.%f train.py:215[27649] INFO Train epoch: 4 [600/1000] Accuracy: 51.03% Loss: 0.723544 2020-10-26 10:26:14.%f train.py:215[27649] INFO Train epoch: 4 [800/1000] Accuracy: 50.28% Loss: 0.723467 2020-10-26 10:29:00.%f train.py:215[27649] INFO Train epoch: 4 [1000/1000] Accuracy: 48.58% Loss: 0.724647 2020-10-26 10:29:01.%f train.py:270[27649] INFO Time: 3149.5196056365967 2020-10-26 10:29:01.%f train.py:271[27649] INFO Test 2020-10-26 10:29:44.%f train.py:254[27649] INFO Test set: Loss: 0.7248, Accuracy: 50.11%) 2020-10-26 10:29:44.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:29:44.%f train.py:283[27649] INFO Time: 3192.6355996131897 2020-10-26 10:29:44.%f train.py:267[27649] INFO Epoch: 5 2020-10-26 10:29:44.%f train.py:268[27649] INFO Train 2020-10-26 10:32:30.%f train.py:215[27649] INFO Train epoch: 5 [200/1000] Accuracy: 50.56% Loss: 0.722997 2020-10-26 10:35:11.%f train.py:215[27649] INFO Train epoch: 5 [400/1000] Accuracy: 50.25% Loss: 0.723275 2020-10-26 10:37:51.%f train.py:215[27649] INFO Train epoch: 5 [600/1000] Accuracy: 49.91% Loss: 0.723445 2020-10-26 10:40:34.%f train.py:215[27649] INFO Train epoch: 5 [800/1000] Accuracy: 50.12% Loss: 0.723526 2020-10-26 10:42:59.%f train.py:215[27649] INFO Train epoch: 5 [1000/1000] Accuracy: 51.03% Loss: 0.723624 2020-10-26 10:42:59.%f train.py:270[27649] INFO Time: 3987.87735247612 2020-10-26 10:42:59.%f train.py:271[27649] INFO Test 2020-10-26 10:43:31.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.12%) 2020-10-26 10:43:31.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:43:31.%f train.py:283[27649] INFO Time: 4019.8337297439575 2020-10-26 10:43:31.%f train.py:267[27649] INFO Epoch: 6 2020-10-26 10:43:31.%f train.py:268[27649] INFO Train 2020-10-26 10:46:15.%f train.py:215[27649] INFO Train epoch: 6 [200/1000] Accuracy: 50.52% Loss: 0.722977 2020-10-26 10:48:55.%f train.py:215[27649] INFO Train epoch: 6 [400/1000] Accuracy: 50.16% Loss: 0.723495 2020-10-26 10:51:35.%f train.py:215[27649] INFO Train epoch: 6 [600/1000] Accuracy: 49.84% Loss: 0.723730 2020-10-26 10:54:18.%f train.py:215[27649] INFO Train epoch: 6 [800/1000] Accuracy: 49.83% Loss: 0.723103 2020-10-26 10:56:58.%f train.py:215[27649] INFO Train epoch: 6 [1000/1000] Accuracy: 49.44% Loss: 0.724185 2020-10-26 10:56:58.%f train.py:270[27649] INFO Time: 4827.371083021164 2020-10-26 10:56:58.%f train.py:271[27649] INFO Test 2020-10-26 10:57:46.%f train.py:254[27649] INFO Test set: Loss: 0.7251, Accuracy: 50.02%) 2020-10-26 10:57:46.%f train.py:282[27649] INFO Best accuracy: 0.50125 2020-10-26 10:57:46.%f train.py:283[27649] INFO Time: 4874.3904139995575 2020-10-26 10:57:46.%f train.py:267[27649] INFO Epoch: 7 2020-10-26 10:57:46.%f train.py:268[27649] INFO Train 2020-10-26 11:00:26.%f train.py:215[27649] INFO Train epoch: 7 [200/1000] Accuracy: 50.39% Loss: 0.723030 2020-10-26 11:03:04.%f train.py:215[27649] INFO Train epoch: 7 [400/1000] Accuracy: 49.62% Loss: 0.723639 2020-10-26 11:04:22.%f train.py:215[27649] INFO Train epoch: 7 [600/1000] Accuracy: 50.14% Loss: 0.723232 2020-10-26 11:06:20.%f train.py:215[27649] INFO Train epoch: 7 [800/1000] Accuracy: 50.88% Loss: 0.723464 2020-10-26 11:09:00.%f train.py:215[27649] INFO Train epoch: 7 [1000/1000] Accuracy: 50.33% Loss: 0.723632 2020-10-26 11:09:00.%f train.py:270[27649] INFO Time: 5549.245084285736 2020-10-26 11:09:00.%f train.py:271[27649] INFO Test 2020-10-26 11:09:47.%f train.py:254[27649] INFO Test set: Loss: 0.7249, Accuracy: 50.14%) 2020-10-26 11:09:47.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:09:47.%f train.py:283[27649] INFO Time: 5596.314600229263 2020-10-26 11:09:47.%f train.py:267[27649] INFO Epoch: 8 2020-10-26 11:09:47.%f train.py:268[27649] INFO Train 2020-10-26 11:12:28.%f train.py:215[27649] INFO Train epoch: 8 [200/1000] Accuracy: 50.12% Loss: 0.722704 2020-10-26 11:15:11.%f train.py:215[27649] INFO Train epoch: 8 [400/1000] Accuracy: 49.34% Loss: 0.723899 2020-10-26 11:17:51.%f train.py:215[27649] INFO Train epoch: 8 [600/1000] Accuracy: 49.66% Loss: 0.723716 2020-10-26 11:20:31.%f train.py:215[27649] INFO Train epoch: 8 [800/1000] Accuracy: 49.61% Loss: 0.723263 2020-10-26 11:23:15.%f train.py:215[27649] INFO Train epoch: 8 [1000/1000] Accuracy: 50.39% Loss: 0.722728 2020-10-26 11:23:15.%f train.py:270[27649] INFO Time: 6404.300854444504 2020-10-26 11:23:15.%f train.py:271[27649] INFO Test 2020-10-26 11:23:59.%f train.py:254[27649] INFO Test set: Loss: 0.7250, Accuracy: 50.06%) 2020-10-26 11:23:59.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:23:59.%f train.py:283[27649] INFO Time: 6447.4487590789795 2020-10-26 11:23:59.%f train.py:267[27649] INFO Epoch: 9 2020-10-26 11:23:59.%f train.py:268[27649] INFO Train 2020-10-26 11:26:44.%f train.py:215[27649] INFO Train epoch: 9 [200/1000] Accuracy: 50.09% Loss: 0.723151 2020-10-26 11:29:24.%f train.py:215[27649] INFO Train epoch: 9 [400/1000] Accuracy: 50.05% Loss: 0.723475 2020-10-26 11:32:04.%f train.py:215[27649] INFO Train epoch: 9 [600/1000] Accuracy: 50.33% Loss: 0.723219 2020-10-26 11:34:47.%f train.py:215[27649] INFO Train epoch: 9 [800/1000] Accuracy: 49.06% Loss: 0.723540 2020-10-26 11:37:25.%f train.py:215[27649] INFO Train epoch: 9 [1000/1000] Accuracy: 50.19% Loss: 0.723136 2020-10-26 11:37:26.%f train.py:270[27649] INFO Time: 7254.803579568863 2020-10-26 11:37:26.%f train.py:271[27649] INFO Test 2020-10-26 11:38:13.%f train.py:254[27649] INFO Test set: Loss: 0.7253, Accuracy: 50.00%) 2020-10-26 11:38:13.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:38:13.%f train.py:283[27649] INFO Time: 7301.76851773262 2020-10-26 11:38:13.%f train.py:267[27649] INFO Epoch: 10 2020-10-26 11:38:13.%f train.py:268[27649] INFO Train 2020-10-26 11:40:53.%f train.py:215[27649] INFO Train epoch: 10 [200/1000] Accuracy: 50.08% Loss: 0.722869 2020-10-26 11:43:33.%f train.py:215[27649] INFO Train epoch: 10 [400/1000] Accuracy: 49.64% Loss: 0.723510 2020-10-26 11:46:17.%f train.py:215[27649] INFO Train epoch: 10 [600/1000] Accuracy: 50.70% Loss: 0.722605 2020-10-26 11:48:57.%f train.py:215[27649] INFO Train epoch: 10 [800/1000] Accuracy: 49.70% Loss: 0.722730 2020-10-26 11:51:37.%f train.py:215[27649] INFO Train epoch: 10 [1000/1000] Accuracy: 49.05% Loss: 0.723217 2020-10-26 11:51:38.%f train.py:270[27649] INFO Time: 8106.941040277481 2020-10-26 11:51:38.%f train.py:271[27649] INFO Test 2020-10-26 11:52:25.%f train.py:254[27649] INFO Test set: Loss: 0.7246, Accuracy: 49.99%) 2020-10-26 11:52:25.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 11:52:25.%f train.py:283[27649] INFO Time: 8153.689094305038 2020-10-26 11:52:25.%f train.py:267[27649] INFO Epoch: 11 2020-10-26 11:52:25.%f train.py:268[27649] INFO Train 2020-10-26 11:55:04.%f train.py:215[27649] INFO Train epoch: 11 [200/1000] Accuracy: 48.84% Loss: 0.722997 2020-10-26 11:57:43.%f train.py:215[27649] INFO Train epoch: 11 [400/1000] Accuracy: 49.75% Loss: 0.722570 2020-10-26 12:00:25.%f train.py:215[27649] INFO Train epoch: 11 [600/1000] Accuracy: 50.05% Loss: 0.722979 2020-10-26 12:03:03.%f train.py:215[27649] INFO Train epoch: 11 [800/1000] Accuracy: 50.08% Loss: 0.722205 2020-10-26 12:05:40.%f train.py:215[27649] INFO Train epoch: 11 [1000/1000] Accuracy: 50.39% Loss: 0.722835 2020-10-26 12:05:41.%f train.py:270[27649] INFO Time: 8949.71140408516 2020-10-26 12:05:41.%f train.py:271[27649] INFO Test 2020-10-26 12:06:27.%f train.py:254[27649] INFO Test set: Loss: 0.7241, Accuracy: 50.09%) 2020-10-26 12:06:27.%f train.py:282[27649] INFO Best accuracy: 0.501375 2020-10-26 12:06:27.%f train.py:283[27649] INFO Time: 8996.346478939056 2020-10-26 12:06:27.%f train.py:267[27649] INFO Epoch: 12 2020-10-26 12:06:27.%f train.py:268[27649] INFO Train 2020-10-26 12:09:04.%f train.py:215[27649] INFO Train epoch: 12 [200/1000] Accuracy: 50.06% Loss: 0.722660 2020-10-26 12:11:44.%f train.py:215[27649] INFO Train epoch: 12 [400/1000] Accuracy: 49.77% Loss: 0.723204 2020-10-26 12:14:22.%f train.py:215[27649] INFO Train epoch: 12 [600/1000] Accuracy: 49.12% Loss: 0.723967 2020-10-26 12:17:00.%f train.py:215[27649] INFO Train epoch: 12 [800/1000] Accuracy: 51.12% Loss: 0.722024 2020-10-26 12:19:41.%f train.py:215[27649] INFO Train epoch: 12 [1000/1000] Accuracy: 50.50% Loss: 0.722844 2020-10-26 12:19:41.%f train.py:270[27649] INFO Time: 9790.267852544785 2020-10-26 12:19:41.%f train.py:271[27649] INFO Test 2020-10-26 12:20:24.%f train.py:254[27649] INFO Test set: Loss: 0.7245, Accuracy: 50.18%) 2020-10-26 12:20:24.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:20:24.%f train.py:283[27649] INFO Time: 9833.335562229156 2020-10-26 12:20:24.%f train.py:267[27649] INFO Epoch: 13 2020-10-26 12:20:24.%f train.py:268[27649] INFO Train 2020-10-26 12:23:05.%f train.py:215[27649] INFO Train epoch: 13 [200/1000] Accuracy: 50.50% Loss: 0.722402 2020-10-26 12:25:43.%f train.py:215[27649] INFO Train epoch: 13 [400/1000] Accuracy: 49.67% Loss: 0.722641 2020-10-26 12:28:21.%f train.py:215[27649] INFO Train epoch: 13 [600/1000] Accuracy: 50.34% Loss: 0.722604 2020-10-26 12:31:00.%f train.py:215[27649] INFO Train epoch: 13 [800/1000] Accuracy: 49.55% Loss: 0.723070 2020-10-26 12:33:41.%f train.py:215[27649] INFO Train epoch: 13 [1000/1000] Accuracy: 50.64% Loss: 0.721970 2020-10-26 12:33:41.%f train.py:270[27649] INFO Time: 10630.292489290237 2020-10-26 12:33:41.%f train.py:271[27649] INFO Test 2020-10-26 12:34:28.%f train.py:254[27649] INFO Test set: Loss: 0.7243, Accuracy: 49.94%) 2020-10-26 12:34:28.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:34:28.%f train.py:283[27649] INFO Time: 10677.06707572937 2020-10-26 12:34:28.%f train.py:267[27649] INFO Epoch: 14 2020-10-26 12:34:28.%f train.py:268[27649] INFO Train 2020-10-26 12:37:06.%f train.py:215[27649] INFO Train epoch: 14 [200/1000] Accuracy: 49.28% Loss: 0.722930 2020-10-26 12:39:44.%f train.py:215[27649] INFO Train epoch: 14 [400/1000] Accuracy: 50.58% Loss: 0.722355 2020-10-26 12:42:21.%f train.py:215[27649] INFO Train epoch: 14 [600/1000] Accuracy: 49.70% Loss: 0.723078 2020-10-26 12:45:04.%f train.py:215[27649] INFO Train epoch: 14 [800/1000] Accuracy: 50.50% Loss: 0.721714 2020-10-26 12:47:42.%f train.py:215[27649] INFO Train epoch: 14 [1000/1000] Accuracy: 49.30% Loss: 0.722824 2020-10-26 12:47:43.%f train.py:270[27649] INFO Time: 11471.877284765244 2020-10-26 12:47:43.%f train.py:271[27649] INFO Test 2020-10-26 12:48:30.%f train.py:254[27649] INFO Test set: Loss: 0.7246, Accuracy: 50.05%) 2020-10-26 12:48:30.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 12:48:30.%f train.py:283[27649] INFO Time: 11518.724534511566 2020-10-26 12:48:30.%f train.py:267[27649] INFO Epoch: 15 2020-10-26 12:48:30.%f train.py:268[27649] INFO Train 2020-10-26 12:51:09.%f train.py:215[27649] INFO Train epoch: 15 [200/1000] Accuracy: 50.44% Loss: 0.722518 2020-10-26 12:53:49.%f train.py:215[27649] INFO Train epoch: 15 [400/1000] Accuracy: 50.44% Loss: 0.722605 2020-10-26 12:56:33.%f train.py:215[27649] INFO Train epoch: 15 [600/1000] Accuracy: 50.66% Loss: 0.722649 2020-10-26 12:59:12.%f train.py:215[27649] INFO Train epoch: 15 [800/1000] Accuracy: 50.16% Loss: 0.722128 2020-10-26 13:01:51.%f train.py:215[27649] INFO Train epoch: 15 [1000/1000] Accuracy: 50.11% Loss: 0.722298 2020-10-26 13:01:52.%f train.py:270[27649] INFO Time: 12320.535963058472 2020-10-26 13:01:52.%f train.py:271[27649] INFO Test 2020-10-26 13:02:38.%f train.py:254[27649] INFO Test set: Loss: 0.7242, Accuracy: 50.08%) 2020-10-26 13:02:38.%f train.py:282[27649] INFO Best accuracy: 0.50175 2020-10-26 13:02:38.%f train.py:283[27649] INFO Time: 12367.37258219719 2020-10-26 13:02:39.%f train.py:267[27649] INFO Epoch: 16 2020-10-26 13:02:39.%f train.py:268[27649] INFO Train 2020-10-26 13:05:17.%f train.py:215[27649] INFO Train epoch: 16 [200/1000] Accuracy: 49.58% Loss: 0.722341 2020-10-26 13:08:00.%f train.py:215[27649] INFO Train epoch: 16 [400/1000] Accuracy: 49.52% Loss: 0.722842 2020-10-26 13:10:37.%f train.py:215[27649] INFO Train epoch: 16 [600/1000] Accuracy: 50.11% Loss: 0.722938 2020-10-26 13:13:15.%f train.py:215[27649] INFO Train epoch: 16 [800/1000] Accuracy: 49.05% Loss: 0.722603 2020-10-26 13:15:53.%f train.py:215[27649] INFO Train epoch: 16 [1000/1000] Accuracy: 50.19% Loss: 0.722058 2020-10-26 13:15:53.%f train.py:270[27649] INFO Time: 13162.007128238678 2020-10-26 13:15:53.%f train.py:271[27649] INFO Test 2020-10-26 13:16:40.%f train.py:254[27649] INFO Test set: Loss: 0.7241, Accuracy: 50.30%) 2020-10-26 13:16:40.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:16:40.%f train.py:283[27649] INFO Time: 13208.76679611206 2020-10-26 13:16:40.%f train.py:267[27649] INFO Epoch: 17 2020-10-26 13:16:40.%f train.py:268[27649] INFO Train 2020-10-26 13:19:18.%f train.py:215[27649] INFO Train epoch: 17 [200/1000] Accuracy: 49.69% Loss: 0.722293 2020-10-26 13:21:59.%f train.py:215[27649] INFO Train epoch: 17 [400/1000] Accuracy: 49.48% Loss: 0.722677 2020-10-26 13:24:36.%f train.py:215[27649] INFO Train epoch: 17 [600/1000] Accuracy: 49.44% Loss: 0.722369 2020-10-26 13:27:13.%f train.py:215[27649] INFO Train epoch: 17 [800/1000] Accuracy: 49.81% Loss: 0.722521 2020-10-26 13:29:54.%f train.py:215[27649] INFO Train epoch: 17 [1000/1000] Accuracy: 49.73% Loss: 0.722504 2020-10-26 13:29:55.%f train.py:270[27649] INFO Time: 14003.732480049133 2020-10-26 13:29:55.%f train.py:271[27649] INFO Test 2020-10-26 13:30:38.%f train.py:254[27649] INFO Test set: Loss: 0.7243, Accuracy: 50.11%) 2020-10-26 13:30:38.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:30:38.%f train.py:283[27649] INFO Time: 14046.820983886719 2020-10-26 13:30:38.%f train.py:267[27649] INFO Epoch: 18 2020-10-26 13:30:38.%f train.py:268[27649] INFO Train 2020-10-26 13:33:19.%f train.py:215[27649] INFO Train epoch: 18 [200/1000] Accuracy: 50.66% Loss: 0.722227 2020-10-26 13:35:56.%f train.py:215[27649] INFO Train epoch: 18 [400/1000] Accuracy: 49.22% Loss: 0.722706 2020-10-26 13:38:34.%f train.py:215[27649] INFO Train epoch: 18 [600/1000] Accuracy: 49.81% Loss: 0.722656 2020-10-26 13:41:17.%f train.py:215[27649] INFO Train epoch: 18 [800/1000] Accuracy: 49.55% Loss: 0.722005 2020-10-26 13:43:54.%f train.py:215[27649] INFO Train epoch: 18 [1000/1000] Accuracy: 50.89% Loss: 0.722071 2020-10-26 13:43:55.%f train.py:270[27649] INFO Time: 14843.993800878525 2020-10-26 13:43:55.%f train.py:271[27649] INFO Test 2020-10-26 13:44:42.%f train.py:254[27649] INFO Test set: Loss: 0.7247, Accuracy: 50.01%) 2020-10-26 13:44:42.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:44:42.%f train.py:283[27649] INFO Time: 14890.727468252182 2020-10-26 13:44:42.%f train.py:267[27649] INFO Epoch: 19 2020-10-26 13:44:42.%f train.py:268[27649] INFO Train 2020-10-26 13:47:19.%f train.py:215[27649] INFO Train epoch: 19 [200/1000] Accuracy: 50.53% Loss: 0.721319 2020-10-26 13:49:57.%f train.py:215[27649] INFO Train epoch: 19 [400/1000] Accuracy: 49.34% Loss: 0.722276 2020-10-26 13:52:38.%f train.py:215[27649] INFO Train epoch: 19 [600/1000] Accuracy: 49.38% Loss: 0.722035 2020-10-26 13:55:17.%f train.py:215[27649] INFO Train epoch: 19 [800/1000] Accuracy: 49.98% Loss: 0.721916 2020-10-26 13:57:57.%f train.py:215[27649] INFO Train epoch: 19 [1000/1000] Accuracy: 48.97% Loss: 0.722423 2020-10-26 13:57:57.%f train.py:270[27649] INFO Time: 15686.256188631058 2020-10-26 13:57:57.%f train.py:271[27649] INFO Test 2020-10-26 13:58:44.%f train.py:254[27649] INFO Test set: Loss: 0.7239, Accuracy: 50.05%) 2020-10-26 13:58:44.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 13:58:44.%f train.py:283[27649] INFO Time: 15732.983254909515 2020-10-26 13:58:44.%f train.py:267[27649] INFO Epoch: 20 2020-10-26 13:58:44.%f train.py:268[27649] INFO Train 2020-10-26 14:01:24.%f train.py:215[27649] INFO Train epoch: 20 [200/1000] Accuracy: 48.89% Loss: 0.722335 2020-10-26 14:04:04.%f train.py:215[27649] INFO Train epoch: 20 [400/1000] Accuracy: 50.56% Loss: 0.721885 2020-10-26 14:06:48.%f train.py:215[27649] INFO Train epoch: 20 [600/1000] Accuracy: 49.50% Loss: 0.722265 2020-10-26 14:09:28.%f train.py:215[27649] INFO Train epoch: 20 [800/1000] Accuracy: 49.81% Loss: 0.721991 2020-10-26 14:12:08.%f train.py:215[27649] INFO Train epoch: 20 [1000/1000] Accuracy: 49.33% Loss: 0.722232 2020-10-26 14:12:09.%f train.py:270[27649] INFO Time: 16537.672178030014 2020-10-26 14:12:09.%f train.py:271[27649] INFO Test 2020-10-26 14:12:56.%f train.py:254[27649] INFO Test set: Loss: 0.7236, Accuracy: 50.09%) 2020-10-26 14:12:56.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:12:56.%f train.py:283[27649] INFO Time: 16584.443323135376 2020-10-26 14:12:56.%f train.py:267[27649] INFO Epoch: 21 2020-10-26 14:12:56.%f train.py:268[27649] INFO Train 2020-10-26 14:15:36.%f train.py:215[27649] INFO Train epoch: 21 [200/1000] Accuracy: 49.75% Loss: 0.722232 2020-10-26 14:18:19.%f train.py:215[27649] INFO Train epoch: 21 [400/1000] Accuracy: 49.98% Loss: 0.722148 2020-10-26 14:20:58.%f train.py:215[27649] INFO Train epoch: 21 [600/1000] Accuracy: 49.28% Loss: 0.722148 2020-10-26 14:23:36.%f train.py:215[27649] INFO Train epoch: 21 [800/1000] Accuracy: 49.09% Loss: 0.721895 2020-10-26 14:26:17.%f train.py:215[27649] INFO Train epoch: 21 [1000/1000] Accuracy: 50.53% Loss: 0.721511 2020-10-26 14:26:18.%f train.py:270[27649] INFO Time: 17386.484574079514 2020-10-26 14:26:18.%f train.py:271[27649] INFO Test 2020-10-26 14:27:02.%f train.py:254[27649] INFO Test set: Loss: 0.7240, Accuracy: 50.08%) 2020-10-26 14:27:02.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:27:02.%f train.py:283[27649] INFO Time: 17430.920579195023 2020-10-26 14:27:02.%f train.py:267[27649] INFO Epoch: 22 2020-10-26 14:27:02.%f train.py:268[27649] INFO Train 2020-10-26 14:29:44.%f train.py:215[27649] INFO Train epoch: 22 [200/1000] Accuracy: 50.98% Loss: 0.721848 2020-10-26 14:32:22.%f train.py:215[27649] INFO Train epoch: 22 [400/1000] Accuracy: 49.48% Loss: 0.722381 2020-10-26 14:35:00.%f train.py:215[27649] INFO Train epoch: 22 [600/1000] Accuracy: 50.03% Loss: 0.721494 2020-10-26 14:37:39.%f train.py:215[27649] INFO Train epoch: 22 [800/1000] Accuracy: 50.00% Loss: 0.721463 2020-10-26 14:40:20.%f train.py:215[27649] INFO Train epoch: 22 [1000/1000] Accuracy: 49.16% Loss: 0.721714 2020-10-26 14:40:21.%f train.py:270[27649] INFO Time: 18229.89799261093 2020-10-26 14:40:21.%f train.py:271[27649] INFO Test 2020-10-26 14:41:07.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.14%) 2020-10-26 14:41:07.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:41:07.%f train.py:283[27649] INFO Time: 18275.499089717865 2020-10-26 14:41:07.%f train.py:267[27649] INFO Epoch: 23 2020-10-26 14:41:07.%f train.py:268[27649] INFO Train 2020-10-26 14:43:45.%f train.py:215[27649] INFO Train epoch: 23 [200/1000] Accuracy: 49.75% Loss: 0.722079 2020-10-26 14:46:24.%f train.py:215[27649] INFO Train epoch: 23 [400/1000] Accuracy: 49.42% Loss: 0.721907 2020-10-26 14:49:03.%f train.py:215[27649] INFO Train epoch: 23 [600/1000] Accuracy: 50.12% Loss: 0.721677 2020-10-26 14:51:46.%f train.py:215[27649] INFO Train epoch: 23 [800/1000] Accuracy: 50.41% Loss: 0.721803 2020-10-26 14:54:25.%f train.py:215[27649] INFO Train epoch: 23 [1000/1000] Accuracy: 49.38% Loss: 0.722154 2020-10-26 14:54:26.%f train.py:270[27649] INFO Time: 19074.978511571884 2020-10-26 14:54:26.%f train.py:271[27649] INFO Test 2020-10-26 14:55:13.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.04%) 2020-10-26 14:55:13.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 14:55:13.%f train.py:283[27649] INFO Time: 19121.877328634262 2020-10-26 14:55:13.%f train.py:267[27649] INFO Epoch: 24 2020-10-26 14:55:13.%f train.py:268[27649] INFO Train 2020-10-26 14:57:52.%f train.py:215[27649] INFO Train epoch: 24 [200/1000] Accuracy: 49.42% Loss: 0.721749 2020-10-26 15:00:30.%f train.py:215[27649] INFO Train epoch: 24 [400/1000] Accuracy: 50.80% Loss: 0.721292 2020-10-26 15:03:13.%f train.py:215[27649] INFO Train epoch: 24 [600/1000] Accuracy: 50.06% Loss: 0.722138 2020-10-26 15:05:52.%f train.py:215[27649] INFO Train epoch: 24 [800/1000] Accuracy: 50.14% Loss: 0.721673 2020-10-26 15:08:31.%f train.py:215[27649] INFO Train epoch: 24 [1000/1000] Accuracy: 48.86% Loss: 0.722384 2020-10-26 15:08:32.%f train.py:270[27649] INFO Time: 19920.678875923157 2020-10-26 15:08:32.%f train.py:271[27649] INFO Test 2020-10-26 15:09:18.%f train.py:254[27649] INFO Test set: Loss: 0.7233, Accuracy: 50.01%) 2020-10-26 15:09:18.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:09:18.%f train.py:283[27649] INFO Time: 19967.131550312042 2020-10-26 15:09:18.%f train.py:267[27649] INFO Epoch: 25 2020-10-26 15:09:18.%f train.py:268[27649] INFO Train 2020-10-26 15:11:28.%f train.py:215[27649] INFO Train epoch: 25 [200/1000] Accuracy: 49.50% Loss: 0.721968 2020-10-26 15:14:08.%f train.py:215[27649] INFO Train epoch: 25 [400/1000] Accuracy: 50.16% Loss: 0.721596 2020-10-26 15:16:52.%f train.py:215[27649] INFO Train epoch: 25 [600/1000] Accuracy: 49.67% Loss: 0.721803 2020-10-26 15:19:32.%f train.py:215[27649] INFO Train epoch: 25 [800/1000] Accuracy: 50.33% Loss: 0.721453 2020-10-26 15:22:11.%f train.py:215[27649] INFO Train epoch: 25 [1000/1000] Accuracy: 50.62% Loss: 0.721836 2020-10-26 15:22:12.%f train.py:270[27649] INFO Time: 20740.679056167603 2020-10-26 15:22:12.%f train.py:271[27649] INFO Test 2020-10-26 15:22:59.%f train.py:254[27649] INFO Test set: Loss: 0.7238, Accuracy: 50.18%) 2020-10-26 15:22:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:22:59.%f train.py:283[27649] INFO Time: 20787.505963802338 2020-10-26 15:22:59.%f train.py:267[27649] INFO Epoch: 26 2020-10-26 15:22:59.%f train.py:268[27649] INFO Train 2020-10-26 15:25:39.%f train.py:215[27649] INFO Train epoch: 26 [200/1000] Accuracy: 50.67% Loss: 0.722108 2020-10-26 15:28:22.%f train.py:215[27649] INFO Train epoch: 26 [400/1000] Accuracy: 49.27% Loss: 0.721857 2020-10-26 15:31:02.%f train.py:215[27649] INFO Train epoch: 26 [600/1000] Accuracy: 49.83% Loss: 0.722023 2020-10-26 15:33:41.%f train.py:215[27649] INFO Train epoch: 26 [800/1000] Accuracy: 49.09% Loss: 0.722354 2020-10-26 15:36:20.%f train.py:215[27649] INFO Train epoch: 26 [1000/1000] Accuracy: 49.58% Loss: 0.721721 2020-10-26 15:36:21.%f train.py:270[27649] INFO Time: 21589.455163240433 2020-10-26 15:36:21.%f train.py:271[27649] INFO Test 2020-10-26 15:37:07.%f train.py:254[27649] INFO Test set: Loss: 0.7234, Accuracy: 50.19%) 2020-10-26 15:37:07.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:37:07.%f train.py:283[27649] INFO Time: 21636.245799064636 2020-10-26 15:37:07.%f train.py:267[27649] INFO Epoch: 27 2020-10-26 15:37:07.%f train.py:268[27649] INFO Train 2020-10-26 15:39:46.%f train.py:215[27649] INFO Train epoch: 27 [200/1000] Accuracy: 49.50% Loss: 0.721894 2020-10-26 15:42:29.%f train.py:215[27649] INFO Train epoch: 27 [400/1000] Accuracy: 49.42% Loss: 0.721731 2020-10-26 15:45:09.%f train.py:215[27649] INFO Train epoch: 27 [600/1000] Accuracy: 50.00% Loss: 0.721781 2020-10-26 15:47:49.%f train.py:215[27649] INFO Train epoch: 27 [800/1000] Accuracy: 50.19% Loss: 0.721131 2020-10-26 15:50:34.%f train.py:215[27649] INFO Train epoch: 27 [1000/1000] Accuracy: 49.39% Loss: 0.722162 2020-10-26 15:50:35.%f train.py:270[27649] INFO Time: 22443.429463863373 2020-10-26 15:50:35.%f train.py:271[27649] INFO Test 2020-10-26 15:51:18.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.22%) 2020-10-26 15:51:18.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 15:51:18.%f train.py:283[27649] INFO Time: 22486.526964187622 2020-10-26 15:51:18.%f train.py:267[27649] INFO Epoch: 28 2020-10-26 15:51:18.%f train.py:268[27649] INFO Train 2020-10-26 15:54:03.%f train.py:215[27649] INFO Train epoch: 28 [200/1000] Accuracy: 50.61% Loss: 0.721195 2020-10-26 15:56:45.%f train.py:215[27649] INFO Train epoch: 28 [400/1000] Accuracy: 49.16% Loss: 0.722328 2020-10-26 15:59:28.%f train.py:215[27649] INFO Train epoch: 28 [600/1000] Accuracy: 49.47% Loss: 0.721517 2020-10-26 16:02:11.%f train.py:215[27649] INFO Train epoch: 28 [800/1000] Accuracy: 49.39% Loss: 0.721710 2020-10-26 16:04:56.%f train.py:215[27649] INFO Train epoch: 28 [1000/1000] Accuracy: 48.91% Loss: 0.721964 2020-10-26 16:04:57.%f train.py:270[27649] INFO Time: 23305.4538462162 2020-10-26 16:04:57.%f train.py:271[27649] INFO Test 2020-10-26 16:05:43.%f train.py:254[27649] INFO Test set: Loss: 0.7233, Accuracy: 50.02%) 2020-10-26 16:05:43.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:05:43.%f train.py:283[27649] INFO Time: 23351.919139146805 2020-10-26 16:05:43.%f train.py:267[27649] INFO Epoch: 29 2020-10-26 16:05:43.%f train.py:268[27649] INFO Train 2020-10-26 16:08:25.%f train.py:215[27649] INFO Train epoch: 29 [200/1000] Accuracy: 50.62% Loss: 0.721156 2020-10-26 16:11:07.%f train.py:215[27649] INFO Train epoch: 29 [400/1000] Accuracy: 49.31% Loss: 0.722137 2020-10-26 16:13:48.%f train.py:215[27649] INFO Train epoch: 29 [600/1000] Accuracy: 50.30% Loss: 0.721555 2020-10-26 16:16:33.%f train.py:215[27649] INFO Train epoch: 29 [800/1000] Accuracy: 49.88% Loss: 0.721814 2020-10-26 16:19:07.%f train.py:215[27649] INFO Train epoch: 29 [1000/1000] Accuracy: 49.12% Loss: 0.721810 2020-10-26 16:19:07.%f train.py:270[27649] INFO Time: 24156.372138738632 2020-10-26 16:19:07.%f train.py:271[27649] INFO Test 2020-10-26 16:19:30.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.22%) 2020-10-26 16:19:30.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:19:30.%f train.py:283[27649] INFO Time: 24179.02462387085 2020-10-26 16:19:30.%f train.py:267[27649] INFO Epoch: 30 2020-10-26 16:19:30.%f train.py:268[27649] INFO Train 2020-10-26 16:20:48.%f train.py:215[27649] INFO Train epoch: 30 [200/1000] Accuracy: 50.88% Loss: 0.721796 2020-10-26 16:22:06.%f train.py:215[27649] INFO Train epoch: 30 [400/1000] Accuracy: 49.59% Loss: 0.722117 2020-10-26 16:23:24.%f train.py:215[27649] INFO Train epoch: 30 [600/1000] Accuracy: 50.19% Loss: 0.721511 2020-10-26 16:24:42.%f train.py:215[27649] INFO Train epoch: 30 [800/1000] Accuracy: 49.66% Loss: 0.721838 2020-10-26 16:26:00.%f train.py:215[27649] INFO Train epoch: 30 [1000/1000] Accuracy: 49.92% Loss: 0.722011 2020-10-26 16:26:00.%f train.py:270[27649] INFO Time: 24569.025196790695 2020-10-26 16:26:00.%f train.py:271[27649] INFO Test 2020-10-26 16:26:23.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.05%) 2020-10-26 16:26:23.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:26:23.%f train.py:283[27649] INFO Time: 24591.9573366642 2020-10-26 16:26:23.%f train.py:267[27649] INFO Epoch: 31 2020-10-26 16:26:23.%f train.py:268[27649] INFO Train 2020-10-26 16:27:41.%f train.py:215[27649] INFO Train epoch: 31 [200/1000] Accuracy: 50.30% Loss: 0.721678 2020-10-26 16:28:58.%f train.py:215[27649] INFO Train epoch: 31 [400/1000] Accuracy: 51.16% Loss: 0.720728 2020-10-26 16:30:16.%f train.py:215[27649] INFO Train epoch: 31 [600/1000] Accuracy: 49.17% Loss: 0.722231 2020-10-26 16:31:34.%f train.py:215[27649] INFO Train epoch: 31 [800/1000] Accuracy: 49.19% Loss: 0.722184 2020-10-26 16:32:52.%f train.py:215[27649] INFO Train epoch: 31 [1000/1000] Accuracy: 49.48% Loss: 0.721670 2020-10-26 16:32:52.%f train.py:270[27649] INFO Time: 24980.771601200104 2020-10-26 16:32:52.%f train.py:271[27649] INFO Test 2020-10-26 16:33:14.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.19%) 2020-10-26 16:33:14.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:33:14.%f train.py:283[27649] INFO Time: 25003.343010663986 2020-10-26 16:33:14.%f train.py:267[27649] INFO Epoch: 32 2020-10-26 16:33:14.%f train.py:268[27649] INFO Train 2020-10-26 16:34:32.%f train.py:215[27649] INFO Train epoch: 32 [200/1000] Accuracy: 50.62% Loss: 0.720921 2020-10-26 16:35:50.%f train.py:215[27649] INFO Train epoch: 32 [400/1000] Accuracy: 49.50% Loss: 0.721514 2020-10-26 16:37:07.%f train.py:215[27649] INFO Train epoch: 32 [600/1000] Accuracy: 50.78% Loss: 0.721799 2020-10-26 16:38:25.%f train.py:215[27649] INFO Train epoch: 32 [800/1000] Accuracy: 50.00% Loss: 0.720876 2020-10-26 16:39:43.%f train.py:215[27649] INFO Train epoch: 32 [1000/1000] Accuracy: 49.89% Loss: 0.721589 2020-10-26 16:39:43.%f train.py:270[27649] INFO Time: 25391.875609874725 2020-10-26 16:39:43.%f train.py:271[27649] INFO Test 2020-10-26 16:40:06.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.11%) 2020-10-26 16:40:06.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:40:06.%f train.py:283[27649] INFO Time: 25414.722820043564 2020-10-26 16:40:06.%f train.py:267[27649] INFO Epoch: 33 2020-10-26 16:40:06.%f train.py:268[27649] INFO Train 2020-10-26 16:41:23.%f train.py:215[27649] INFO Train epoch: 33 [200/1000] Accuracy: 49.09% Loss: 0.721560 2020-10-26 16:42:39.%f train.py:215[27649] INFO Train epoch: 33 [400/1000] Accuracy: 48.88% Loss: 0.722098 2020-10-26 16:43:56.%f train.py:215[27649] INFO Train epoch: 33 [600/1000] Accuracy: 50.31% Loss: 0.721358 2020-10-26 16:45:13.%f train.py:215[27649] INFO Train epoch: 33 [800/1000] Accuracy: 50.06% Loss: 0.721508 2020-10-26 16:46:29.%f train.py:215[27649] INFO Train epoch: 33 [1000/1000] Accuracy: 49.72% Loss: 0.722063 2020-10-26 16:46:30.%f train.py:270[27649] INFO Time: 25798.640019655228 2020-10-26 16:46:30.%f train.py:271[27649] INFO Test 2020-10-26 16:46:52.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.14%) 2020-10-26 16:46:52.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:46:52.%f train.py:283[27649] INFO Time: 25821.180437088013 2020-10-26 16:46:52.%f train.py:267[27649] INFO Epoch: 34 2020-10-26 16:46:52.%f train.py:268[27649] INFO Train 2020-10-26 16:48:09.%f train.py:215[27649] INFO Train epoch: 34 [200/1000] Accuracy: 49.45% Loss: 0.721542 2020-10-26 16:49:26.%f train.py:215[27649] INFO Train epoch: 34 [400/1000] Accuracy: 50.08% Loss: 0.721261 2020-10-26 16:50:42.%f train.py:215[27649] INFO Train epoch: 34 [600/1000] Accuracy: 51.12% Loss: 0.721381 2020-10-26 16:51:59.%f train.py:215[27649] INFO Train epoch: 34 [800/1000] Accuracy: 50.38% Loss: 0.721490 2020-10-26 16:53:15.%f train.py:215[27649] INFO Train epoch: 34 [1000/1000] Accuracy: 50.19% Loss: 0.721831 2020-10-26 16:53:16.%f train.py:270[27649] INFO Time: 26204.604976654053 2020-10-26 16:53:16.%f train.py:271[27649] INFO Test 2020-10-26 16:53:38.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.06%) 2020-10-26 16:53:38.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 16:53:38.%f train.py:283[27649] INFO Time: 26227.149789333344 2020-10-26 16:53:38.%f train.py:267[27649] INFO Epoch: 35 2020-10-26 16:53:38.%f train.py:268[27649] INFO Train 2020-10-26 16:54:55.%f train.py:215[27649] INFO Train epoch: 35 [200/1000] Accuracy: 50.02% Loss: 0.721548 2020-10-26 16:56:11.%f train.py:215[27649] INFO Train epoch: 35 [400/1000] Accuracy: 49.72% Loss: 0.721850 2020-10-26 16:57:28.%f train.py:215[27649] INFO Train epoch: 35 [600/1000] Accuracy: 50.05% Loss: 0.721939 2020-10-26 16:58:45.%f train.py:215[27649] INFO Train epoch: 35 [800/1000] Accuracy: 50.11% Loss: 0.722074 2020-10-26 17:00:02.%f train.py:215[27649] INFO Train epoch: 35 [1000/1000] Accuracy: 49.80% Loss: 0.721860 2020-10-26 17:00:02.%f train.py:270[27649] INFO Time: 26611.17203426361 2020-10-26 17:00:02.%f train.py:271[27649] INFO Test 2020-10-26 17:00:25.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.18%) 2020-10-26 17:00:25.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:00:25.%f train.py:283[27649] INFO Time: 26633.60524392128 2020-10-26 17:00:25.%f train.py:267[27649] INFO Epoch: 36 2020-10-26 17:00:25.%f train.py:268[27649] INFO Train 2020-10-26 17:01:41.%f train.py:215[27649] INFO Train epoch: 36 [200/1000] Accuracy: 50.58% Loss: 0.721399 2020-10-26 17:02:58.%f train.py:215[27649] INFO Train epoch: 36 [400/1000] Accuracy: 50.42% Loss: 0.721195 2020-10-26 17:04:16.%f train.py:215[27649] INFO Train epoch: 36 [600/1000] Accuracy: 49.61% Loss: 0.721957 2020-10-26 17:05:32.%f train.py:215[27649] INFO Train epoch: 36 [800/1000] Accuracy: 50.20% Loss: 0.721555 2020-10-26 17:06:49.%f train.py:215[27649] INFO Train epoch: 36 [1000/1000] Accuracy: 49.50% Loss: 0.722120 2020-10-26 17:06:49.%f train.py:270[27649] INFO Time: 27018.353071689606 2020-10-26 17:06:49.%f train.py:271[27649] INFO Test 2020-10-26 17:07:12.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.09%) 2020-10-26 17:07:12.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:07:12.%f train.py:283[27649] INFO Time: 27041.12589907646 2020-10-26 17:07:12.%f train.py:267[27649] INFO Epoch: 37 2020-10-26 17:07:12.%f train.py:268[27649] INFO Train 2020-10-26 17:08:29.%f train.py:215[27649] INFO Train epoch: 37 [200/1000] Accuracy: 51.09% Loss: 0.720909 2020-10-26 17:09:46.%f train.py:215[27649] INFO Train epoch: 37 [400/1000] Accuracy: 50.59% Loss: 0.721484 2020-10-26 17:11:02.%f train.py:215[27649] INFO Train epoch: 37 [600/1000] Accuracy: 49.84% Loss: 0.721889 2020-10-26 17:12:19.%f train.py:215[27649] INFO Train epoch: 37 [800/1000] Accuracy: 50.52% Loss: 0.721360 2020-10-26 17:13:36.%f train.py:215[27649] INFO Train epoch: 37 [1000/1000] Accuracy: 49.75% Loss: 0.721795 2020-10-26 17:13:36.%f train.py:270[27649] INFO Time: 27425.170134544373 2020-10-26 17:13:36.%f train.py:271[27649] INFO Test 2020-10-26 17:13:59.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.12%) 2020-10-26 17:13:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:13:59.%f train.py:283[27649] INFO Time: 27447.889960050583 2020-10-26 17:13:59.%f train.py:267[27649] INFO Epoch: 38 2020-10-26 17:13:59.%f train.py:268[27649] INFO Train 2020-10-26 17:15:16.%f train.py:215[27649] INFO Train epoch: 38 [200/1000] Accuracy: 49.08% Loss: 0.721949 2020-10-26 17:16:33.%f train.py:215[27649] INFO Train epoch: 38 [400/1000] Accuracy: 49.73% Loss: 0.722057 2020-10-26 17:17:50.%f train.py:215[27649] INFO Train epoch: 38 [600/1000] Accuracy: 50.83% Loss: 0.720905 2020-10-26 17:19:06.%f train.py:215[27649] INFO Train epoch: 38 [800/1000] Accuracy: 50.88% Loss: 0.720798 2020-10-26 17:20:23.%f train.py:215[27649] INFO Train epoch: 38 [1000/1000] Accuracy: 49.64% Loss: 0.721864 2020-10-26 17:20:23.%f train.py:270[27649] INFO Time: 27832.214666843414 2020-10-26 17:20:23.%f train.py:271[27649] INFO Test 2020-10-26 17:20:46.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.00%) 2020-10-26 17:20:46.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:20:46.%f train.py:283[27649] INFO Time: 27854.788491249084 2020-10-26 17:20:46.%f train.py:267[27649] INFO Epoch: 39 2020-10-26 17:20:46.%f train.py:268[27649] INFO Train 2020-10-26 17:22:03.%f train.py:215[27649] INFO Train epoch: 39 [200/1000] Accuracy: 50.08% Loss: 0.721081 2020-10-26 17:23:20.%f train.py:215[27649] INFO Train epoch: 39 [400/1000] Accuracy: 50.06% Loss: 0.721671 2020-10-26 17:24:37.%f train.py:215[27649] INFO Train epoch: 39 [600/1000] Accuracy: 48.83% Loss: 0.722440 2020-10-26 17:25:54.%f train.py:215[27649] INFO Train epoch: 39 [800/1000] Accuracy: 49.69% Loss: 0.721849 2020-10-26 17:27:10.%f train.py:215[27649] INFO Train epoch: 39 [1000/1000] Accuracy: 50.22% Loss: 0.721833 2020-10-26 17:27:11.%f train.py:270[27649] INFO Time: 28239.423560619354 2020-10-26 17:27:11.%f train.py:271[27649] INFO Test 2020-10-26 17:27:33.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.14%) 2020-10-26 17:27:33.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:27:33.%f train.py:283[27649] INFO Time: 28261.885771036148 2020-10-26 17:27:33.%f train.py:267[27649] INFO Epoch: 40 2020-10-26 17:27:33.%f train.py:268[27649] INFO Train 2020-10-26 17:28:50.%f train.py:215[27649] INFO Train epoch: 40 [200/1000] Accuracy: 50.67% Loss: 0.721462 2020-10-26 17:30:07.%f train.py:215[27649] INFO Train epoch: 40 [400/1000] Accuracy: 50.20% Loss: 0.721098 2020-10-26 17:31:24.%f train.py:215[27649] INFO Train epoch: 40 [600/1000] Accuracy: 50.61% Loss: 0.721035 2020-10-26 17:32:41.%f train.py:215[27649] INFO Train epoch: 40 [800/1000] Accuracy: 49.78% Loss: 0.721563 2020-10-26 17:33:59.%f train.py:215[27649] INFO Train epoch: 40 [1000/1000] Accuracy: 49.39% Loss: 0.721895 2020-10-26 17:33:59.%f train.py:270[27649] INFO Time: 28647.86800456047 2020-10-26 17:33:59.%f train.py:271[27649] INFO Test 2020-10-26 17:34:22.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.05%) 2020-10-26 17:34:22.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:34:22.%f train.py:283[27649] INFO Time: 28670.392629146576 2020-10-26 17:34:22.%f train.py:267[27649] INFO Epoch: 41 2020-10-26 17:34:22.%f train.py:268[27649] INFO Train 2020-10-26 17:35:39.%f train.py:215[27649] INFO Train epoch: 41 [200/1000] Accuracy: 49.88% Loss: 0.721476 2020-10-26 17:36:56.%f train.py:215[27649] INFO Train epoch: 41 [400/1000] Accuracy: 50.09% Loss: 0.721099 2020-10-26 17:38:13.%f train.py:215[27649] INFO Train epoch: 41 [600/1000] Accuracy: 49.31% Loss: 0.721230 2020-10-26 17:39:31.%f train.py:215[27649] INFO Train epoch: 41 [800/1000] Accuracy: 50.80% Loss: 0.721088 2020-10-26 17:40:48.%f train.py:215[27649] INFO Train epoch: 41 [1000/1000] Accuracy: 50.23% Loss: 0.721994 2020-10-26 17:40:48.%f train.py:270[27649] INFO Time: 29057.317686080933 2020-10-26 17:40:48.%f train.py:271[27649] INFO Test 2020-10-26 17:41:11.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.08%) 2020-10-26 17:41:11.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:41:11.%f train.py:283[27649] INFO Time: 29079.925040006638 2020-10-26 17:41:11.%f train.py:267[27649] INFO Epoch: 42 2020-10-26 17:41:11.%f train.py:268[27649] INFO Train 2020-10-26 17:42:28.%f train.py:215[27649] INFO Train epoch: 42 [200/1000] Accuracy: 50.16% Loss: 0.721571 2020-10-26 17:43:45.%f train.py:215[27649] INFO Train epoch: 42 [400/1000] Accuracy: 49.53% Loss: 0.721621 2020-10-26 17:45:02.%f train.py:215[27649] INFO Train epoch: 42 [600/1000] Accuracy: 50.61% Loss: 0.721240 2020-10-26 17:46:19.%f train.py:215[27649] INFO Train epoch: 42 [800/1000] Accuracy: 49.98% Loss: 0.721345 2020-10-26 17:47:37.%f train.py:215[27649] INFO Train epoch: 42 [1000/1000] Accuracy: 49.59% Loss: 0.721730 2020-10-26 17:47:37.%f train.py:270[27649] INFO Time: 29465.678948640823 2020-10-26 17:47:37.%f train.py:271[27649] INFO Test 2020-10-26 17:47:59.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 49.94%) 2020-10-26 17:47:59.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:47:59.%f train.py:283[27649] INFO Time: 29488.179372787476 2020-10-26 17:47:59.%f train.py:267[27649] INFO Epoch: 43 2020-10-26 17:47:59.%f train.py:268[27649] INFO Train 2020-10-26 17:49:16.%f train.py:215[27649] INFO Train epoch: 43 [200/1000] Accuracy: 48.73% Loss: 0.721529 2020-10-26 17:50:33.%f train.py:215[27649] INFO Train epoch: 43 [400/1000] Accuracy: 49.44% Loss: 0.721806 2020-10-26 17:51:50.%f train.py:215[27649] INFO Train epoch: 43 [600/1000] Accuracy: 50.03% Loss: 0.721423 2020-10-26 17:53:07.%f train.py:215[27649] INFO Train epoch: 43 [800/1000] Accuracy: 50.31% Loss: 0.721637 2020-10-26 17:54:23.%f train.py:215[27649] INFO Train epoch: 43 [1000/1000] Accuracy: 49.55% Loss: 0.721455 2020-10-26 17:54:24.%f train.py:270[27649] INFO Time: 29872.40514731407 2020-10-26 17:54:24.%f train.py:271[27649] INFO Test 2020-10-26 17:54:46.%f train.py:254[27649] INFO Test set: Loss: 0.7224, Accuracy: 50.02%) 2020-10-26 17:54:46.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 17:54:46.%f train.py:283[27649] INFO Time: 29894.952974557877 2020-10-26 17:54:46.%f train.py:267[27649] INFO Epoch: 44 2020-10-26 17:54:46.%f train.py:268[27649] INFO Train 2020-10-26 17:56:02.%f train.py:215[27649] INFO Train epoch: 44 [200/1000] Accuracy: 49.70% Loss: 0.720927 2020-10-26 17:57:19.%f train.py:215[27649] INFO Train epoch: 44 [400/1000] Accuracy: 49.70% Loss: 0.721201 2020-10-26 17:58:36.%f train.py:215[27649] INFO Train epoch: 44 [600/1000] Accuracy: 50.69% Loss: 0.721111 2020-10-26 17:59:52.%f train.py:215[27649] INFO Train epoch: 44 [800/1000] Accuracy: 50.77% Loss: 0.720785 2020-10-26 18:01:09.%f train.py:215[27649] INFO Train epoch: 44 [1000/1000] Accuracy: 50.03% Loss: 0.721340 2020-10-26 18:01:09.%f train.py:270[27649] INFO Time: 30277.745666980743 2020-10-26 18:01:09.%f train.py:271[27649] INFO Test 2020-10-26 18:01:31.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 50.11%) 2020-10-26 18:01:31.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:01:31.%f train.py:283[27649] INFO Time: 30300.194601535797 2020-10-26 18:01:31.%f train.py:267[27649] INFO Epoch: 45 2020-10-26 18:01:31.%f train.py:268[27649] INFO Train 2020-10-26 18:02:48.%f train.py:215[27649] INFO Train epoch: 45 [200/1000] Accuracy: 49.02% Loss: 0.721996 2020-10-26 18:04:04.%f train.py:215[27649] INFO Train epoch: 45 [400/1000] Accuracy: 50.08% Loss: 0.721470 2020-10-26 18:05:21.%f train.py:215[27649] INFO Train epoch: 45 [600/1000] Accuracy: 50.27% Loss: 0.721050 2020-10-26 18:06:37.%f train.py:215[27649] INFO Train epoch: 45 [800/1000] Accuracy: 50.50% Loss: 0.721582 2020-10-26 18:07:54.%f train.py:215[27649] INFO Train epoch: 45 [1000/1000] Accuracy: 49.67% Loss: 0.721660 2020-10-26 18:07:54.%f train.py:270[27649] INFO Time: 30683.1965944767 2020-10-26 18:07:54.%f train.py:271[27649] INFO Test 2020-10-26 18:08:17.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.00%) 2020-10-26 18:08:17.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:08:17.%f train.py:283[27649] INFO Time: 30705.85431098938 2020-10-26 18:08:17.%f train.py:267[27649] INFO Epoch: 46 2020-10-26 18:08:17.%f train.py:268[27649] INFO Train 2020-10-26 18:09:34.%f train.py:215[27649] INFO Train epoch: 46 [200/1000] Accuracy: 50.27% Loss: 0.721508 2020-10-26 18:10:50.%f train.py:215[27649] INFO Train epoch: 46 [400/1000] Accuracy: 50.97% Loss: 0.721218 2020-10-26 18:12:07.%f train.py:215[27649] INFO Train epoch: 46 [600/1000] Accuracy: 50.03% Loss: 0.721920 2020-10-26 18:13:24.%f train.py:215[27649] INFO Train epoch: 46 [800/1000] Accuracy: 50.38% Loss: 0.720731 2020-10-26 18:14:40.%f train.py:215[27649] INFO Train epoch: 46 [1000/1000] Accuracy: 50.38% Loss: 0.721281 2020-10-26 18:14:41.%f train.py:270[27649] INFO Time: 31089.40986776352 2020-10-26 18:14:41.%f train.py:271[27649] INFO Test 2020-10-26 18:15:03.%f train.py:254[27649] INFO Test set: Loss: 0.7227, Accuracy: 49.96%) 2020-10-26 18:15:03.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:15:03.%f train.py:283[27649] INFO Time: 31111.84361052513 2020-10-26 18:15:03.%f train.py:267[27649] INFO Epoch: 47 2020-10-26 18:15:03.%f train.py:268[27649] INFO Train 2020-10-26 18:16:20.%f train.py:215[27649] INFO Train epoch: 47 [200/1000] Accuracy: 49.14% Loss: 0.721514 2020-10-26 18:17:36.%f train.py:215[27649] INFO Train epoch: 47 [400/1000] Accuracy: 49.91% Loss: 0.721505 2020-10-26 18:18:53.%f train.py:215[27649] INFO Train epoch: 47 [600/1000] Accuracy: 49.98% Loss: 0.721421 2020-10-26 18:20:10.%f train.py:215[27649] INFO Train epoch: 47 [800/1000] Accuracy: 49.73% Loss: 0.721589 2020-10-26 18:21:26.%f train.py:215[27649] INFO Train epoch: 47 [1000/1000] Accuracy: 49.84% Loss: 0.721595 2020-10-26 18:21:27.%f train.py:270[27649] INFO Time: 31495.542852640152 2020-10-26 18:21:27.%f train.py:271[27649] INFO Test 2020-10-26 18:21:49.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.20%) 2020-10-26 18:21:49.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:21:49.%f train.py:283[27649] INFO Time: 31518.005761146545 2020-10-26 18:21:49.%f train.py:267[27649] INFO Epoch: 48 2020-10-26 18:21:49.%f train.py:268[27649] INFO Train 2020-10-26 18:23:06.%f train.py:215[27649] INFO Train epoch: 48 [200/1000] Accuracy: 49.94% Loss: 0.721164 2020-10-26 18:24:22.%f train.py:215[27649] INFO Train epoch: 48 [400/1000] Accuracy: 51.23% Loss: 0.721384 2020-10-26 18:25:39.%f train.py:215[27649] INFO Train epoch: 48 [600/1000] Accuracy: 49.59% Loss: 0.721872 2020-10-26 18:26:55.%f train.py:215[27649] INFO Train epoch: 48 [800/1000] Accuracy: 49.88% Loss: 0.721526 2020-10-26 18:28:12.%f train.py:215[27649] INFO Train epoch: 48 [1000/1000] Accuracy: 49.00% Loss: 0.721775 2020-10-26 18:28:12.%f train.py:270[27649] INFO Time: 31901.144391536713 2020-10-26 18:28:12.%f train.py:271[27649] INFO Test 2020-10-26 18:28:35.%f train.py:254[27649] INFO Test set: Loss: 0.7226, Accuracy: 50.11%) 2020-10-26 18:28:35.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:28:35.%f train.py:283[27649] INFO Time: 31923.629400491714 2020-10-26 18:28:35.%f train.py:267[27649] INFO Epoch: 49 2020-10-26 18:28:35.%f train.py:268[27649] INFO Train 2020-10-26 18:29:51.%f train.py:215[27649] INFO Train epoch: 49 [200/1000] Accuracy: 50.20% Loss: 0.721193 2020-10-26 18:31:08.%f train.py:215[27649] INFO Train epoch: 49 [400/1000] Accuracy: 48.86% Loss: 0.721818 2020-10-26 18:32:25.%f train.py:215[27649] INFO Train epoch: 49 [600/1000] Accuracy: 49.16% Loss: 0.721519 2020-10-26 18:33:41.%f train.py:215[27649] INFO Train epoch: 49 [800/1000] Accuracy: 49.73% Loss: 0.721708 2020-10-26 18:34:58.%f train.py:215[27649] INFO Train epoch: 49 [1000/1000] Accuracy: 49.69% Loss: 0.721645 2020-10-26 18:34:58.%f train.py:270[27649] INFO Time: 32306.78603386879 2020-10-26 18:34:58.%f train.py:271[27649] INFO Test 2020-10-26 18:35:20.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.19%) 2020-10-26 18:35:20.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:35:20.%f train.py:283[27649] INFO Time: 32329.21870279312 2020-10-26 18:35:20.%f train.py:267[27649] INFO Epoch: 50 2020-10-26 18:35:20.%f train.py:268[27649] INFO Train 2020-10-26 18:36:37.%f train.py:215[27649] INFO Train epoch: 50 [200/1000] Accuracy: 50.05% Loss: 0.721832 2020-10-26 18:37:53.%f train.py:215[27649] INFO Train epoch: 50 [400/1000] Accuracy: 49.45% Loss: 0.721136 2020-10-26 18:39:10.%f train.py:215[27649] INFO Train epoch: 50 [600/1000] Accuracy: 49.75% Loss: 0.721008 2020-10-26 18:40:26.%f train.py:215[27649] INFO Train epoch: 50 [800/1000] Accuracy: 49.98% Loss: 0.721070 2020-10-26 18:41:43.%f train.py:215[27649] INFO Train epoch: 50 [1000/1000] Accuracy: 49.12% Loss: 0.721500 2020-10-26 18:41:44.%f train.py:270[27649] INFO Time: 32712.40008211136 2020-10-26 18:41:44.%f train.py:271[27649] INFO Test 2020-10-26 18:42:06.%f train.py:254[27649] INFO Test set: Loss: 0.7229, Accuracy: 50.18%) 2020-10-26 18:42:06.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:42:06.%f train.py:283[27649] INFO Time: 32734.86405801773 2020-10-26 18:42:06.%f train.py:267[27649] INFO Epoch: 51 2020-10-26 18:42:06.%f train.py:268[27649] INFO Train 2020-10-26 18:43:24.%f train.py:215[27649] INFO Train epoch: 51 [200/1000] Accuracy: 51.00% Loss: 0.721218 2020-10-26 18:44:42.%f train.py:215[27649] INFO Train epoch: 51 [400/1000] Accuracy: 50.23% Loss: 0.720999 2020-10-26 18:45:59.%f train.py:215[27649] INFO Train epoch: 51 [600/1000] Accuracy: 49.58% Loss: 0.722122 2020-10-26 18:47:16.%f train.py:215[27649] INFO Train epoch: 51 [800/1000] Accuracy: 50.22% Loss: 0.721728 2020-10-26 18:48:34.%f train.py:215[27649] INFO Train epoch: 51 [1000/1000] Accuracy: 49.73% Loss: 0.721392 2020-10-26 18:48:34.%f train.py:270[27649] INFO Time: 33122.7360265255 2020-10-26 18:48:34.%f train.py:271[27649] INFO Test 2020-10-26 18:48:56.%f train.py:254[27649] INFO Test set: Loss: 0.7222, Accuracy: 50.06%) 2020-10-26 18:48:56.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:48:56.%f train.py:283[27649] INFO Time: 33145.32122063637 2020-10-26 18:48:56.%f train.py:267[27649] INFO Epoch: 52 2020-10-26 18:48:56.%f train.py:268[27649] INFO Train 2020-10-26 18:50:13.%f train.py:215[27649] INFO Train epoch: 52 [200/1000] Accuracy: 50.22% Loss: 0.721378 2020-10-26 18:51:30.%f train.py:215[27649] INFO Train epoch: 52 [400/1000] Accuracy: 49.62% Loss: 0.721496 2020-10-26 18:52:48.%f train.py:215[27649] INFO Train epoch: 52 [600/1000] Accuracy: 50.80% Loss: 0.721010 2020-10-26 18:54:05.%f train.py:215[27649] INFO Train epoch: 52 [800/1000] Accuracy: 50.55% Loss: 0.721563 2020-10-26 18:55:25.%f train.py:215[27649] INFO Train epoch: 52 [1000/1000] Accuracy: 50.56% Loss: 0.721192 2020-10-26 18:55:25.%f train.py:270[27649] INFO Time: 33534.01327371597 2020-10-26 18:55:25.%f train.py:271[27649] INFO Test 2020-10-26 18:55:50.%f train.py:254[27649] INFO Test set: Loss: 0.7222, Accuracy: 50.06%) 2020-10-26 18:55:50.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 18:55:50.%f train.py:283[27649] INFO Time: 33558.541908741 2020-10-26 18:55:50.%f train.py:267[27649] INFO Epoch: 53 2020-10-26 18:55:50.%f train.py:268[27649] INFO Train 2020-10-26 18:57:10.%f train.py:215[27649] INFO Train epoch: 53 [200/1000] Accuracy: 51.30% Loss: 0.720911 2020-10-26 18:58:29.%f train.py:215[27649] INFO Train epoch: 53 [400/1000] Accuracy: 48.81% Loss: 0.721708 2020-10-26 18:59:48.%f train.py:215[27649] INFO Train epoch: 53 [600/1000] Accuracy: 49.72% Loss: 0.721488 2020-10-26 19:01:05.%f train.py:215[27649] INFO Train epoch: 53 [800/1000] Accuracy: 49.03% Loss: 0.721965 2020-10-26 19:02:22.%f train.py:215[27649] INFO Train epoch: 53 [1000/1000] Accuracy: 49.03% Loss: 0.721585 2020-10-26 19:02:23.%f train.py:270[27649] INFO Time: 33951.554489851 2020-10-26 19:02:23.%f train.py:271[27649] INFO Test 2020-10-26 19:02:45.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.04%) 2020-10-26 19:02:45.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:02:45.%f train.py:283[27649] INFO Time: 33974.32499027252 2020-10-26 19:02:45.%f train.py:267[27649] INFO Epoch: 54 2020-10-26 19:02:45.%f train.py:268[27649] INFO Train 2020-10-26 19:04:02.%f train.py:215[27649] INFO Train epoch: 54 [200/1000] Accuracy: 50.77% Loss: 0.721066 2020-10-26 19:05:20.%f train.py:215[27649] INFO Train epoch: 54 [400/1000] Accuracy: 50.33% Loss: 0.721573 2020-10-26 19:06:38.%f train.py:215[27649] INFO Train epoch: 54 [600/1000] Accuracy: 49.39% Loss: 0.721073 2020-10-26 19:07:55.%f train.py:215[27649] INFO Train epoch: 54 [800/1000] Accuracy: 48.78% Loss: 0.721267 2020-10-26 19:09:12.%f train.py:215[27649] INFO Train epoch: 54 [1000/1000] Accuracy: 49.88% Loss: 0.721365 2020-10-26 19:09:12.%f train.py:270[27649] INFO Time: 34361.14559984207 2020-10-26 19:09:12.%f train.py:271[27649] INFO Test 2020-10-26 19:09:35.%f train.py:254[27649] INFO Test set: Loss: 0.7231, Accuracy: 50.29%) 2020-10-26 19:09:35.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:09:35.%f train.py:283[27649] INFO Time: 34384.130383491516 2020-10-26 19:09:35.%f train.py:267[27649] INFO Epoch: 55 2020-10-26 19:09:35.%f train.py:268[27649] INFO Train 2020-10-26 19:10:52.%f train.py:215[27649] INFO Train epoch: 55 [200/1000] Accuracy: 49.23% Loss: 0.721556 2020-10-26 19:12:09.%f train.py:215[27649] INFO Train epoch: 55 [400/1000] Accuracy: 50.66% Loss: 0.720750 2020-10-26 19:13:26.%f train.py:215[27649] INFO Train epoch: 55 [600/1000] Accuracy: 50.25% Loss: 0.720918 2020-10-26 19:14:43.%f train.py:215[27649] INFO Train epoch: 55 [800/1000] Accuracy: 49.09% Loss: 0.721661 2020-10-26 19:16:00.%f train.py:215[27649] INFO Train epoch: 55 [1000/1000] Accuracy: 50.88% Loss: 0.720696 2020-10-26 19:16:00.%f train.py:270[27649] INFO Time: 34769.300970077515 2020-10-26 19:16:00.%f train.py:271[27649] INFO Test 2020-10-26 19:16:23.%f train.py:254[27649] INFO Test set: Loss: 0.7225, Accuracy: 50.30%) 2020-10-26 19:16:23.%f train.py:282[27649] INFO Best accuracy: 0.503 2020-10-26 19:16:23.%f train.py:283[27649] INFO Time: 34792.19325590134 2020-10-26 19:16:23.%f train.py:267[27649] INFO Epoch: 56 2020-10-26 19:16:23.%f train.py:268[27649] INFO Train 2020-10-26 19:17:40.%f train.py:215[27649] INFO Train epoch: 56 [200/1000] Accuracy: 49.16% Loss: 0.721391 2020-10-26 19:18:57.%f train.py:215[27649] INFO Train epoch: 56 [400/1000] Accuracy: 49.30% Loss: 0.721471 2020-10-26 19:20:14.%f train.py:215[27649] INFO Train epoch: 56 [600/1000] Accuracy: 50.09% Loss: 0.720977 2020-10-26 19:21:31.%f train.py:215[27649] INFO Train epoch: 56 [800/1000] Accuracy: 50.50% Loss: 0.721714 2020-10-26 19:22:49.%f train.py:215[27649] INFO Train epoch: 56 [1000/1000] Accuracy: 49.75% Loss: 0.721326 2020-10-26 19:22:49.%f train.py:270[27649] INFO Time: 35177.6729080677 2020-10-26 19:22:49.%f train.py:271[27649] INFO Test 2020-10-26 19:23:12.%f train.py:254[27649] INFO Test set: Loss: 0.7224, Accuracy: 50.38%) 2020-10-26 19:23:12.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:23:12.%f train.py:283[27649] INFO Time: 35200.54278087616 2020-10-26 19:23:12.%f train.py:267[27649] INFO Epoch: 57 2020-10-26 19:23:12.%f train.py:268[27649] INFO Train 2020-10-26 19:24:29.%f train.py:215[27649] INFO Train epoch: 57 [200/1000] Accuracy: 49.30% Loss: 0.721213 2020-10-26 19:25:46.%f train.py:215[27649] INFO Train epoch: 57 [400/1000] Accuracy: 50.39% Loss: 0.721261 2020-10-26 19:27:03.%f train.py:215[27649] INFO Train epoch: 57 [600/1000] Accuracy: 50.08% Loss: 0.720789 2020-10-26 19:28:20.%f train.py:215[27649] INFO Train epoch: 57 [800/1000] Accuracy: 49.72% Loss: 0.721149 2020-10-26 19:29:37.%f train.py:215[27649] INFO Train epoch: 57 [1000/1000] Accuracy: 50.05% Loss: 0.721388 2020-10-26 19:29:37.%f train.py:270[27649] INFO Time: 35586.03649306297 2020-10-26 19:29:37.%f train.py:271[27649] INFO Test 2020-10-26 19:30:00.%f train.py:254[27649] INFO Test set: Loss: 0.7228, Accuracy: 50.09%) 2020-10-26 19:30:00.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:30:00.%f train.py:283[27649] INFO Time: 35608.92472934723 2020-10-26 19:30:00.%f train.py:267[27649] INFO Epoch: 58 2020-10-26 19:30:00.%f train.py:268[27649] INFO Train 2020-10-26 19:31:17.%f train.py:215[27649] INFO Train epoch: 58 [200/1000] Accuracy: 50.05% Loss: 0.721109 2020-10-26 19:32:34.%f train.py:215[27649] INFO Train epoch: 58 [400/1000] Accuracy: 49.72% Loss: 0.721442 2020-10-26 19:33:51.%f train.py:215[27649] INFO Train epoch: 58 [600/1000] Accuracy: 49.86% Loss: 0.721890 2020-10-26 19:35:08.%f train.py:215[27649] INFO Train epoch: 58 [800/1000] Accuracy: 51.59% Loss: 0.720619 2020-10-26 19:36:25.%f train.py:215[27649] INFO Train epoch: 58 [1000/1000] Accuracy: 50.92% Loss: 0.720742 2020-10-26 19:36:26.%f train.py:270[27649] INFO Time: 35994.47965550423 2020-10-26 19:36:26.%f train.py:271[27649] INFO Test 2020-10-26 19:36:48.%f train.py:254[27649] INFO Test set: Loss: 0.7221, Accuracy: 50.05%) 2020-10-26 19:36:48.%f train.py:282[27649] INFO Best accuracy: 0.50375 2020-10-26 19:36:48.%f train.py:283[27649] INFO Time: 36017.25994181633 2020-10-26 19:36:48.%f train.py:267[27649] INFO Epoch: 59 2020-10-26 19:36:48.%f train.py:268[27649] INFO Train 2020-10-26 19:38:05.%f train.py:215[27649] INFO Train epoch: 59 [200/1000] Accuracy: 49.64% Loss: 0.720858 2020-10-26 19:39:22.%f train.py:215[27649] INFO Train epoch: 59 [400/1000] Accuracy: 49.27% Loss: 0.721158 2020-10-26 19:40:39.%f train.py:215[27649] INFO Train epoch: 59 [600/1000] Accuracy: 50.83% Loss: 0.721004 2020-10-26 19:41:56.%f train.py:215[27649] INFO Train epoch: 59 [800/1000] Accuracy: 49.95% Loss: 0.721280 2020-10-26 19:43:13.%f train.py:215[27649] INFO Train epoch: 59 [1000/1000] Accuracy: 49.91% Loss: 0.721499 2020-10-26 19:43:14.%f train.py:270[27649] INFO Time: 36402.47468829155 2020-10-26 19:43:14.%f train.py:271[27649] INFO Test 2020-10-26 19:43:36.%f train.py:254[27649] INFO Test set: Loss: 0.7230, Accuracy: 50.00%) 2020-10
非常感谢您的回复。 我再使用1080Ti和Tesla V100试一试。如果有结果就联系您。
万分感谢!!!
好的!推荐使用算力更高的V100 祝顺利!:)
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月27日(星期二) 下午4:21 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
非常感谢您的回复。 我再使用1080Ti和Tesla V100试一试。如果有结果就联系您。
万分感谢!!!
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
It is a wonderful wok~surprising!
作者您好:
您说的两种方法我都有尝试: 我尝试使用了 【(1)用一个预训练模型带一下,就可以收敛了 】发现模型的确是收敛了,但是最终收敛的效果可能还有一定的差距。目前训练集的结果大概80%多,测试集合也80%左右(s-uniward 0.4) 【(2)换不是2080ti的机器】,我使用了tesla v100进行训练,但是训练的结果还是50%不收敛。
对于目前的结果,有一些希望作者您提供的帮助!!! 1.非常希望作者您能提供一下您数据集构造的方法(尤其是stego数据集),以及使用的机器型号,我将在学校尝试寻找一样的机器进行训练。 2.对于您提到的使用预训练的模型带一下,非常希望您能在为我指定数据集构造方法后提供一个您曾经训练好的模型,我来尝试带动训练。
感谢作者一直以来的回复。万分感谢!
可能说2080ti更适合打游戏太武断啦! 不过之前有小伙伴也反馈过类似的问题,复现的结果是确实2080ti不行,猜测是浮点计算精度可能不够? 因为该任务可能对精度有很高的要求,后续尝试过两种解决方案都可以: (1)用一个预训练模型带一下,就可以收敛了 (2)换不是2080ti的机器 祝好!
您好! 我重新阅读了您的实验配置: 1.我们对stego的制作方式相同,都使用了Fridrich老师的代码,机器型号应该无关 2.当前github展示代码中的默认超参数都是针对训练集为BOSSbase的;对于混合BOWS的训练集,可能超参数要进行微调。 您能否方便Train中先只使用BOSSbase库图片,“【(2)换不是2080ti的机器】,使用了tesla v100进行训练”,看看效果呢?这样可能方便排查一些问题。
之前训练好的模型将随后上传
非常抱歉给您的研究工作带来困扰!我也感到很奇怪,s-uniward 0.4较易检测,复现了很多次,按理来说不应该出问题。
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月30日(星期五) 下午2:54 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
作者您好:
您说的两种方法我都有尝试: 我尝试使用了 【(1)用一个预训练模型带一下,就可以收敛了 】发现模型的确是收敛了,但是最终收敛的效果可能还有一定的差距。目前训练集的结果大概80%多,测试集合也80%左右(s-uniward 0.4) 【(2)换不是2080ti的机器】,我使用了tesla v100进行训练,但是训练的结果还是50%不收敛。
对于目前的结果,有一些希望作者您提供的帮助!!! 1.非常希望作者您能提供一下您数据集构造的方法(尤其是stego数据集),以及使用的机器型号,我将在学校尝试寻找一样的机器进行训练。 2.对于您提到的使用预训练的模型带一下,非常希望您能在为我指定数据集构造方法后提供一个您曾经训练好的模型,我来尝试带动训练。
感谢作者一直以来的回复。万分感谢!
可能说2080ti更适合打游戏太武断啦! 不过之前有小伙伴也反馈过类似的问题,复现的结果是确实2080ti不行,猜测是浮点计算精度可能不够? 因为该任务可能对精度有很高的要求,后续尝试过两种解决方案都可以: (1)用一个预训练模型带一下,就可以收敛了 (2)换不是2080ti的机器 祝好!
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好! 我重新阅读了您的实验配置: 1.我们对stego的制作方式相同,都使用了Fridrich老师的代码,机器型号应该无关 2.当前github展示代码中的默认超参数都是针对训练集为BOSSbase的;对于混合BOWS的训练集,可能超参数要进行微调。 您能否方便Train中先只使用BOSSbase库图片,“【(2)换不是2080ti的机器】,使用了tesla v100进行训练”,看看效果呢?这样可能方便排查一些问题。 之前训练好的模型将随后上传 非常抱歉给您的研究工作带来困扰!我也感到很奇怪,s-uniward 0.4较易检测,复现了很多次,按理来说不应该出问题。 祝好! … ------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年10月30日(星期五) 下午2:54 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) 作者您好: 您说的两种方法我都有尝试: 我尝试使用了 【(1)用一个预训练模型带一下,就可以收敛了 】发现模型的确是收敛了,但是最终收敛的效果可能还有一定的差距。目前训练集的结果大概80%多,测试集合也80%左右(s-uniward 0.4) 【(2)换不是2080ti的机器】,我使用了tesla v100进行训练,但是训练的结果还是50%不收敛。 对于目前的结果,有一些希望作者您提供的帮助!!! 1.非常希望作者您能提供一下您数据集构造的方法(尤其是stego数据集),以及使用的机器型号,我将在学校尝试寻找一样的机器进行训练。 2.对于您提到的使用预训练的模型带一下,非常希望您能在为我指定数据集构造方法后提供一个您曾经训练好的模型,我来尝试带动训练。 感谢作者一直以来的回复。万分感谢! 可能说2080ti更适合打游戏太武断啦! 不过之前有小伙伴也反馈过类似的问题,复现的结果是确实2080ti不行,猜测是浮点计算精度可能不够? 因为该任务可能对精度有很高的要求,后续尝试过两种解决方案都可以: (1)用一个预训练模型带一下,就可以收敛了 (2)换不是2080ti的机器 祝好! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
非常感谢您的回复!!!我再按照您的指导进行一些尝试。感谢一直以来的支持
您好,按照您的指导 我是用BOSSbase的前6000张图片作为训练cover,生成对应的s-uniward 0.4 stego 用6000-9000张图像作为valid cover,生成对应的stego
使用 tesla V100 训练,学习率策略使用您默认的配置 第二种学习率设置。不使用反转,切割图像等预处理,不生成增强数据集。
实验的结果依旧是无法有效收敛,可能预训练模型对于网络收敛具有比较大的指导作用。 如果您的预训练模型测试好了,能不能给我使用一下,非常感谢。
以下是训练第500epoch的结果
您好!确实没有遇到这种奇怪的情况
您有试过这些样本跑其他的模型,比如SRNet,可以正常收敛对吗?
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月2日(星期一) 下午3:24 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
您好,按照您的指导 我是用BOSSbase的前6000张图片作为训练cover,生成对应的s-uniward 0.4 stego 用6000-9000张图像作为cover,生成对应的stego
使用 tesla V100 训练,学习率策略使用您默认的配置 第二种学习率设置。不使用反转,切割图像等预处理,不生成增强数据集。
实验的结果依旧是无法有效收敛,可能预训练模型对于网络收敛具有比较大的指导作用。 如果您的预训练模型测试好了,能不能给我使用一下,非常感谢。
以下是训练第500epoch的结果
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好,使用这些数据集合构造的样本。 关于这些数据集应用于SRNet的问题(我也经过多次实验,但是好像效果也有一些问题)
训练结果如下: 由上图可见,在train set的时候,loss和acc是正常的 但是: 在验证集上,检测器的检测出现了一些问题。目前也没有解决。如果您有好的SRNet的实验复现方法,能不能也简单指导一下,非常感激!!!
Em。。。 我觉得如果是这样的话! 我们首先要理清一下样本是不是有什么问题,我这边直接跑SRNet,不进行数据增强,训练和验证也都是没问题的。(也或许我的样本和SRNet原作者出现了相同特征?
制备样本是使用的matlab代码吗?还是C++?
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 上午9:21 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
您好,使用这些数据集合构造的样本。 关于这些数据集应用于SRNet的问题(我也经过多次实验,但是好像效果也有一些问题)
使用SRNet网络(tensorflow版本,由fridrich官网提供),默认配置,使用SRNet_Example.py(启动网络训练过程),使用BOWS 10000张,BOSSbase 6000张,作为train cover, 生成对应 train stego。 使用BOSSbase 剩下4000张作为 valid cover,生成对应stego作为 valid stego。
训练结果如下:
由上图可见,在train set的时候,loss和acc是正常的 但是:
在验证集上,检测器的检测出现了一些问题。目前也没有解决。
使用您的网络train part(略微修改,数据的preprocess函数),使用SRNet网络来自(https://github.com/brijeshiitg/Pytorch-implementation-of-SRNet)。**使用数据集同上**,训练结果为**_不收敛_**,可以参见issue(https://github.com/brijeshiitg/Pytorch-implementation-of-SRNet/issues/5, 好像有很多人提出了这个问题,原作者说需要使用数据增强,不增强没有用,最近准备尝试一下)
关于本人实验数据集的问题,我是用可视化观察嵌入率改变的cover与stego的残差,编写脚本随机选择一个图像进行测试。 以下显示几个测试结果(本人实验感觉嵌入应该是没有问题,数据集构造应该是正常的) 由左到右依次为 0.1 .2 .3 .4 .5 嵌入率的残差,s,m分别表示标准差和均值,图像上面的标题表示了使用的算法
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好! 我使用的是C++的代码进行样本的制作。
也或许我的样本和SRNet原作者出现了相同特征
有一些帮助,如果您愿意提供将不胜感激
关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式
如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激!
这就对了! 不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文
我建议先使用binghamton实验室的matlab的代码重新生成实验图片重新尝试。(注意,是子函数也是matlab代码的matlab生成方式
我这边能分享的实验数据很多,不过最好是能够直接在您那边跑通呀!
祝好
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:28 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
您好! 我使用的是C++的代码进行样本的制作。
也或许我的样本和SRNet原作者出现了相同特征
有一些帮助,如果您愿意提供将不胜感激
您好能不能把当时的SRNet实验数据,给我copy一份?
或者您是如何生成您的数据的(如果有数据构建脚本能够分享,感谢)
关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式
如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激!
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
好的 感谢您的指导
客气!!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:43 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
好的 感谢您的指导
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好,我的问题得到了解决,解决过程中您 提供的帮助本人表示非常感谢!
关于最后训练的情况: 另外 最后一次我用 2080Ti 训练您的SiaStegNet,一次成功! SRNet官网的tf网络在tesla上也取得与原论文相近的结果!
不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文
关于您上述,本人还有一个疑问,我想知道这篇论文的名称,去学习一下为什么会出现这种情况!感谢
太棒啦! 客气
论文的名字是 Deep learning is a good steganalysis tool when embedding key is reused for different images, even if there is a cover sourcemismatch
和 Steganalysis via a Convolutional Neural Network using Large Convolution Filters for Embedding Process with Same Stego Key (不过论文里似乎没有明确指出是那份C++代码的问题)
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月4日(星期三) 上午9:24 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
您好,我的问题得到了解决,解决过程中您 提供的帮助本人表示非常感谢!
另外 最后一次我用 2080Ti 训练,一次成功! SRNet官网的tf网络在tesla上也取得与原论文相近的结果!
不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文 关于您上述,本人还有一个疑问,我想知道这篇论文的名称,去学习一下为什么会出现这种情况!感谢
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好,我的问题得到了解决,解决过程中您 提供的帮助本人表示非常感谢!
关于最后训练的情况: 另外 最后一次我用 2080Ti 训练您的SiaStegNet,一次成功! SRNet官网的tf网络在tesla上也取得与原论文相近的结果!
不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文
关于您上述,本人还有一个疑问,我想知道这篇论文的名称,去学习一下为什么会出现这种情况!感谢
你好,我也遇到了和你相似的情况。所以训练不收敛以及SRNet tensorflow版本验证集很抖动的原因都是因为使用了Jessica网站提供的C++Win版本的S-UNIWARD.zip来生成stego图像吗,用matlab版本生成stego图像就可以解决了吗?另外,请问一下SRNet的pytorch版本您可以成功收敛了吗?
很困扰,期待回复,祝好!
你好! 很高兴您对我的研究成果感兴趣!
1.训练不收敛以及SRNet tensorflow版本验证集很抖动的原因都是因为使用了Jessica网站提供的C++Win版本的S-UNIWARD.zip来生成stego图像吗,用matlab版本生成stego图像就可以解决了吗? 有很大的可能,但是也不一定!如果用matlab版本生成stego图像还不可以解决,您可以新开一个issue我跟您一起排查问题。
2.SRNet的pytorch版本您可以成功收敛了吗? 在进行对照实验时,我为了保证和原作者效果相同,使用了tensorflow的原版代码。或许有效的SRNet的pytorch版本可以在https://github.com/brijeshiitg/Pytorch-implementation-of-SRNet获取。
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月13日(星期五) 下午4:36 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
您好,我的问题得到了解决,解决过程中您 提供的帮助本人表示非常感谢!
关于最后训练的情况: 另外 最后一次我用 2080Ti 训练您的SiaStegNet,一次成功! SRNet官网的tf网络在tesla上也取得与原论文相近的结果!
不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文
关于您上述,本人还有一个疑问,我想知道这篇论文的名称,去学习一下为什么会出现这种情况!感谢
你好,我也遇到了和你相似的情况。所以训练不收敛以及SRNet tensorflow版本验证集很抖动的原因都是因为使用了Jessica网站提供的C++Win版本的S-UNIWARD.zip来生成stego图像吗,用matlab版本生成stego图像就可以解决了吗?另外,请问一下SRNet的pytorch版本您可以成功收敛了吗?
很困扰,期待回复,祝好!
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
感谢您的回复!目前打算试一下用matlab版本生成stego图像试试。
在SRNet论文中在检测S-UNIWARD0.4bpp的时候,他的准确率是0.8977,在您的论文中复现结果为92.22吗?我跑SRNet tensorflow版本的时候准确率只有0.74啊 T.T (训练集:0.9280 验证集:0.8050 测试集:Accuracy: 0.7435003 | Loss: 30.671312)。
我目前就是根据brijeshiitg这个代码在训练SRNet pytorch版本。但由于他没有提供训练代码,我的训练效果S-UNIWARD0.4bpp只能达到0.84,而且在前100epoch左右loss一直为0.69,后面才下降。
我是用的数据集均为BOSSBASE(训练:5000对,验证1000对,测试4000对)
brijeshiitg好像说想要训练代码可以给他发邮件,您可以试试哈哈。 论文中复现的结果都是复现了好多次啦,应该木有什么问题
祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月13日(星期五) 下午5:38 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
感谢您的回复!目前打算试一下用matlab版本生成stego图像试试。
在SRNet论文中在检测S-UNIWARD0.4bpp的时候,他的准确率是0.8977,在您的论文中复现结果为92.22吗?我跑SRNet tensorflow版本的时候准确率只有0.74啊 T.T (训练集:0.9280 验证集:0.8050 测试集:Accuracy: 0.7435003 | Loss: 30.671312)。
我目前就是根据brijeshiitg这个代码在训练SRNet pytorch版本。但由于他没有提供训练代码,我的训练效果S-UNIWARD0.4bpp只能达到0.84,而且在前100epoch左右loss一直为0.69,后面才下降。
我是用的数据集均为BOSSBASE(训练:5000对,验证1000对,测试4000对)
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
感谢帮助!我去试试,祝好!
作者您好,请问HILL的代码是在哪里有呢?请问您知道除了Jessica的网站还有其他学习资料库嘛?
感谢!
这就对了! 不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文 我建议先使用binghamton实验室的matlab的代码重新生成实验图片重新尝试。(注意,是子函数也是matlab代码的matlab生成方式 我这边能分享的实验数据很多,不过最好是能够直接在您那边跑通呀! 祝好 … ------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:28 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) 您好! 我使用的是C++的代码进行样本的制作。 也或许我的样本和SRNet原作者出现了相同特征 有一些帮助,如果您愿意提供将不胜感激 您好能不能把当时的SRNet实验数据,给我copy一份? 或者您是如何生成您的数据的(如果有数据构建脚本能够分享,感谢) 关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式 如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
这就对了! 不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文 我建议先使用binghamton实验室的matlab的代码重新生成实验图片重新尝试。(注意,是子函数也是matlab代码的matlab生成方式 我这边能分享的实验数据很多,不过最好是能够直接在您那边跑通呀! 祝好 … ------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:28 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) 您好! 我使用的是C++的代码进行样本的制作。 也或许我的样本和SRNet原作者出现了相同特征 有一些帮助,如果您愿意提供将不胜感激 您好能不能把当时的SRNet实验数据,给我copy一份? 或者您是如何生成您的数据的(如果有数据构建脚本能够分享,感谢) 关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式 如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好,我重新用matlab生成了stego数据集用来训练SRNet。结果十分奇怪:第一个epoch就直接val_acc=0.991,loss=0.024。拿这个模型去测试Acc也直接高达0.99。怀疑自己。。。
哈哈哈哈哈这……应该是有问题哦 再检查一下图片数据 祝好!
------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月20日(星期五) 上午9:51 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1)
这就对了! 不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文 我建议先使用binghamton实验室的matlab的代码重新生成实验图片重新尝试。(注意,是子函数也是matlab代码的matlab生成方式 我这边能分享的实验数据很多,不过最好是能够直接在您那边跑通呀! 祝好 … ------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:28 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) 您好! 我使用的是C++的代码进行样本的制作。 也或许我的样本和SRNet原作者出现了相同特征 有一些帮助,如果您愿意提供将不胜感激 您好能不能把当时的SRNet实验数据,给我copy一份? 或者您是如何生成您的数据的(如果有数据构建脚本能够分享,感谢) 关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式 如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
这就对了! 不要用那份binghamton实验室的C++的代码 有点问题!(该问题之前被写成过论文 我建议先使用binghamton实验室的matlab的代码重新生成实验图片重新尝试。(注意,是子函数也是matlab代码的matlab生成方式 我这边能分享的实验数据很多,不过最好是能够直接在您那边跑通呀! 祝好 … ------------------ 原始邮件 ------------------ 发件人: "SiaStg/SiaStegNet" <notifications@github.com>; 发送时间: 2020年11月3日(星期二) 下午4:28 收件人: "SiaStg/SiaStegNet"<SiaStegNet@noreply.github.com>; 抄送: "可可"<2622679282@qq.com>;"Comment"<comment@noreply.github.com>; 主题: Re: [SiaStg/SiaStegNet] 关于具体如何训练 (#1) 您好! 我使用的是C++的代码进行样本的制作。 也或许我的样本和SRNet原作者出现了相同特征 有一些帮助,如果您愿意提供将不胜感激 您好能不能把当时的SRNet实验数据,给我copy一份? 或者您是如何生成您的数据的(如果有数据构建脚本能够分享,感谢) 关于如何传递数据集、或文件 1.本人可以提供一个公网IP,您可以通过一个临时账户使用scp等命令进行上传,如果需要ftp服务,我也可以帮忙开启 2.可以通过添加QQ好友等进行数据传递 3.通过线上网盘进行数据分享 4.其他您愿意的方式 如果您愿意提供帮助,请告诉我您愿意怎样提供数据,不胜感激! — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
您好,我重新用matlab生成了stego数据集。结果十分奇怪:第一个epoch就直接val_acc=0.991,loss=0.024。拿这个模型去测试Acc也直接高达0.99。怀疑自己。。。
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.
作者你好,我想问一下关于这篇论文如何具体构造数据集,如何进行训练。 鉴于本人水平有限,使用该repo的代码进行实验的时候不能取得很好的效果。
关于如何设置学习率、如何构建训练数据集、验证数据集,具体如何训练的问题希望能够稍微指导一下。
感谢