Describe the feature

How can I plot the val_acc on the graph by the tools/analysis_tools/analyze_logs.py

Motivation

Hi, I am studying in this mmclassification, when I want to plot the curves and compare between acc and val_acc, I can not find some thing like val_acc to display.

Related resources

The tutorial just show the result of acc top 1, and top 5, currently I have not find out how to show the val_acc https://mmclassification.readthedocs.io/en/latest/tools/analysis.html

Additional context

As usual, by using Tensorflow or some libraries, we can show the graph to compare the result as the picture below:

Checklist

• I have searched related issues but cannot get the expected help.
• I have read related documents and don't know what to do.

Describe the question you meet

Hello, I trained Resnet-101 on my custom dataset. Then I tested the model with the official test tool on my test set and got an accuracy of 91.44%. However, when I try to inference the test set image one by one, I found the model can't classify images correctly. Almost all the images are misclassified. My data set has 3 labels '3h','5h' and '8h', The model will classify everything into ''8h'. But when I reload the model, it might predict everything into '5h'.

To control, I tried different models on different dataset and this issue still existed. Which result should I believe? Thank you!

Post related information

1. The output of pip list | grep "mmcv\|mmcls\|^torch"

mmcls 0.18.0 mmcv-full 1.4.0 torch 1.8.1 torchaudio 0.8.0a0+e4e171a torchvision 0.9.1

2. Your config file if you modified it or created a new one.

model = dict(
    type='ImageClassifier',
    backbone=dict(
        type='ResNet',
        depth=101,
        num_stages=4,
        out_indices=(3, ),
        style='pytorch'),
    neck=dict(type='GlobalAveragePooling'),
    head=dict(
        type='LinearClsHead',
        num_classes=3,
        in_channels=2048,
        loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
        topk=(1, 5)))
dataset_type = 'ImageNet'
img_norm_cfg = dict(
    mean=[420.57, 683.84, 1104.41], std=[100.3, 50.3, 150.6], to_rgb=True)
train_pipeline = [
    dict(type='LoadImageFromFile', color_type='unchanged', to_float32=True),
    dict(type='RandomResizedCrop', size=224),
    dict(type='RandomFlip', flip_prob=0.5, direction='horizontal'),
    dict(
        type='Normalize',
        mean=[420.57, 683.84, 1104.41],
        std=[100.3, 50.3, 150.6],
        to_rgb=True),
    dict(type='ImageToTensor', keys=['img']),
    dict(type='ToTensor', keys=['gt_label']),
    dict(type='Collect', keys=['img', 'gt_label'])
]
test_pipeline = [
    dict(type='LoadImageFromFile', color_type='unchanged', to_float32=True),
    dict(type='Resize', size=(256, -1)),
    dict(type='CenterCrop', crop_size=224),
    dict(
        type='Normalize',
        mean=[420.57, 683.84, 1104.41],
        std=[100.3, 50.3, 150.6],
        to_rgb=True),
    dict(type='ImageToTensor', keys=['img']),
    dict(type='Collect', keys=['img'])
]
data = dict(
    samples_per_gpu=32,
    workers_per_gpu=4,
    train=dict(
        type='ImageNet',
        data_prefix='/mnt/data/processed/microglia/ImageNet_merge/train',
        pipeline=[
            dict(
                type='LoadImageFromFile',
                color_type='unchanged',
                to_float32=True),
            dict(type='RandomResizedCrop', size=224),
            dict(type='RandomFlip', flip_prob=0.5, direction='horizontal'),
            dict(
                type='Normalize',
                mean=[420.57, 683.84, 1104.41],
                std=[100.3, 50.3, 150.6],
                to_rgb=True),
            dict(type='ImageToTensor', keys=['img']),
            dict(type='ToTensor', keys=['gt_label']),
            dict(type='Collect', keys=['img', 'gt_label'])
        ],
        classes='/mnt/data/processed/microglia/ImageNet_merge/classes.txt'),
    val=dict(
        type='ImageNet',
        data_prefix='/mnt/data/processed/microglia/ImageNet_merge/test',
        ann_file='/mnt/data/processed/microglia/ImageNet_merge/test/test.txt',
        pipeline=[
            dict(
                type='LoadImageFromFile',
                color_type='unchanged',
                to_float32=True),
            dict(type='Resize', size=(256, -1)),
            dict(type='CenterCrop', crop_size=224),
            dict(
                type='Normalize',
                mean=[420.57, 683.84, 1104.41],
                std=[100.3, 50.3, 150.6],
                to_rgb=True),
            dict(type='ImageToTensor', keys=['img']),
            dict(type='Collect', keys=['img'])
        ],
        classes='/mnt/data/processed/microglia/ImageNet_merge/classes.txt'),
    test=dict(
        type='ImageNet',
        data_prefix='/mnt/data/processed/microglia/ImageNet_merge/test',
        ann_file='/mnt/data/processed/microglia/ImageNet_merge/test/test.txt',
        pipeline=[
            dict(
                type='LoadImageFromFile',
                color_type='unchanged',
                to_float32=True),
            dict(type='Resize', size=(256, -1)),
            dict(type='CenterCrop', crop_size=224),
            dict(
                type='Normalize',
                mean=[420.57, 683.84, 1104.41],
                std=[100.3, 50.3, 150.6],
                to_rgb=True),
            dict(type='ImageToTensor', keys=['img']),
            dict(type='Collect', keys=['img'])
        ],
        classes='/mnt/data/processed/microglia/ImageNet_merge/classes.txt'))
evaluation = dict(interval=10, metric='accuracy', metric_options=dict(topk=1))
optimizer = dict(type='Adam', lr=0.0001, weight_decay=0.001, amsgrad=False)
optimizer_config = dict(grad_clip=None)
lr_config = dict(policy='step', step=[20, 50, 70], gamma=0.1)
runner = dict(type='EpochBasedRunner', max_epochs=100)
checkpoint_config = dict(interval=10)
log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
dist_params = dict(backend='nccl')
log_level = 'INFO'
load_from = None
resume_from = None
workflow = [('train', 1)]
work_dir = 'merge_resnet_100e_128'
gpu_ids = range(0, 1)

推荐使用英语模板 General question，以便你的问题帮助更多人。

首先确认以下内容

• 我已经查询了相关的 issue，但没有找到需要的帮助。
• 我已经阅读了相关文档，但仍不知道如何解决。

描述你遇到的问题

为什么同样的数据和模型，同样的超参数及优化器。模型精度比我在tf上的低了 96%降到92%

Hi, dear authors! Thank you for your amazing job!

I have a question about K-Fold config. In K-Fold, can the val and the test be the same? Like follow:

    train=dict(
        type='KFoldDataset',
        num_splits=5,
        dataset=dict(
            type='RepeatDataset',
            times=15,
            dataset=dict(
                type=dataset_type,
                data_prefix='data/my_data/train',
                pipeline=train_pipeline))),
    val=dict(
        type=dataset_type,
        data_prefix='data/my_data/val',
        ann_file='data/my_data/meta/val.txt',
        pipeline=test_pipeline),
    test=dict(
        # replace `data/val` with `data/test` for standard test
        type=dataset_type,
        data_prefix='data/my_data/val',
        ann_file='data/my_data/meta/val.txt',
        pipeline=test_pipeline))

Thank you!

您好： 非常感谢您的工作！ 我有一个关于 K-Fold 配置的问题。请问在 K-Fold 中，val 的配置和 test 的配置可以相同吗？（就是val 和 test公用一套图像） 就像下面这样:

    train=dict(
        type='KFoldDataset',
        num_splits=5,
        dataset=dict(
            type='RepeatDataset',
            times=15,
            dataset=dict(
                type=dataset_type,
                data_prefix='data/my_data/train',
                pipeline=train_pipeline))),
    val=dict(
        type=dataset_type,
        data_prefix='data/my_data/val',
        ann_file='data/my_data/meta/val.txt',
        pipeline=test_pipeline),
    test=dict(
        # replace `data/val` with `data/test` for standard test
        type=dataset_type,
        data_prefix='data/my_data/val',
        ann_file='data/my_data/meta/val.txt',
        pipeline=test_pipeline))

谢谢！

I have raised a PR in MMDetection contributing a dedicated MMDetWandbHook for MMDetection. I was playing around with MMClassification and realized that MMClassification can use a similar dedicated hook (with few minor modifications).

Motivation

The goal of this PR is to contribute a dedicated Weights and Biases hook for MMClassification.

Modification

The PR adds two new files:

• wandblogger_hook.py where all the Weights and Biases related logic lives and,
• eval_hooks.py was added so that the MMClsWandbHook can reuse the validation results.

The feature can easily be used this:

log_config = dict(
            interval=10,
            hooks=[
                dict(type='MMClsWandbHook',
                     wandb_init_kwargs={
                         'entity': WANDB_ENTITY,
                         'project': WANDB_PROJECT_NAME
                     },
                     log_checkpoint=True,
                     log_checkpoint_metadata=True,
                     num_eval_images=100)
            ])

Use cases (Optional)

Here are some of the use cases that this PR introduces and should be helpful to the community in general.

Metrics

• The WandbLogger will automatically log training and validation metrics.
• It will log system (CPU/GPU) metrics.

https://user-images.githubusercontent.com/31141479/161776566-57c59402-a16b-4940-a90a-76b64963a4d1.mov

Checkpointing with Metadata

If log_checkpoint is True, the checkpoint saved at every checkpoint interval will be saved as W&B Artifacts. On top of this, if log_checkpoint_metadata is True, every checkpoint artifact will have metadata associated with it as shown in the recording below.

https://user-images.githubusercontent.com/31141479/161777072-cd00f1ee-aa24-42a4-b47f-9c915600a2f4.mov

Log Model Prediction 🎉

If num_eval_images > 0, at every evaluation interval, theMMClsWandbHook logs the model prediction as interactive W&B Tables. To know more about W&B Tables, please refer to the docs here. The 'MMClsWandbHook' logs the predicted class labels along with ground truth labels.

https://user-images.githubusercontent.com/31141479/161784564-51943f43-53c8-41cf-a920-4d70e67bdc9b.mov

Checklist

Before PR:

• [x] Pre-commit or other linting tools are used to fix the potential lint issues.
• [ ] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [ ] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [ ] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [x] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [x] CLA has been signed and all committers have signed the CLA in this PR.

Motivation

DeiT III: Revenge of the ViT github

Checklist

Before PR:

• [x] Pre-commit or other linting tools are used to fix the potential lint issues.
• [x] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [x] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [x] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [x] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [x] CLA has been signed and all committers have signed the CLA in this PR.

Motivation

Support CUB dataset.

Augmentation and optimizer is based on A Novel Plug-in Module for Fine-Grained Visual Classification.

Checklist

Before PR:

• [x] Pre-commit or other linting tools are used to fix the potential lint issues.
• [x] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [x] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [x] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [x] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [x] CLA has been signed and all committers have signed the CLA in this PR.

推荐使用英语模板 General question，以便你的问题帮助更多人。

首先确认以下内容

• 我已经查询了相关的 issue，但没有找到需要的帮助。
• 我已经阅读了相关文档，但仍不知道如何解决。

描述你遇到的问题

您好！ 我需要在ImageNet1K上训练一个基于ResNet101V1c进行定制的backbone，即在每个bottleneck中的第二个卷积上增加一个经过卷积的残差输出，用作下游分割任务。有以下两个问题进行请教。

1. 我们实验室有8张GTX 3090，能否告知一下使用8张GPU需要训练多久。
2. 是应该完全从随机权重开始训练？还是应该对主支加载预训练权重，对每个bolck内新增残差模块随机初始化并在残差项最后使用0初始化的norm层来稳定训练？如果加载预训练权重，新增残差模块学习率是否应该与主支一样？ 万分感谢！

相关信息

1. pip list | grep "mmcv\|mmcls\|^torch" 命令的输出 [填写这里]
2. 如果你修改了，或者使用了新的配置文件，请在这里写明

[填写这里]

3. 如果你是在训练过程中遇到的问题，请填写完整的训练日志和报错信息 [填写这里]
4. 如果你对 mmcls 文件夹下的代码做了其他相关的修改，请在这里写明 [填写这里]

您好，作者！ 首先非常感谢您的工作！ 请问如何对两张输入的图像做相同的aug操作？ 具体来说，我将两张3通道的图像concat成了6通道图像(上头要求的)，之后我想对这种6通道图像进行aug操作，比如autoaug，albu等。但是这些aug只支持3通道图像，不支持6通道图像。 所以我想到先把这6通道图像拆成两张3通道图像，之后对这两张3通道图像进行相同的aug操作（mmcv.xxxx），之后再把它们重新合并成6通道图像。 可是我该如何对这两张3通道的图像做相同的aug操作？ 比如：

    def __call__(self, results):
        if np.random.rand() > self.prob:
            return results
        for key in results.get('img_fields', ['img']):
            img = results[key]
            img_contrasted = mmcv.auto_contrast(img)
            results[key] = img_contrasted.astype(img.dtype)
        return results

假设现在result[“img”]就是6通道图像，我现在把它拆成result[“img_1”]和result[“img_2”]，请问我该如何对result[“img_1”]和result[“img_2”]施加相同的mmcv.auto_contrast操作？ 如果我用到Albu，请问我该如何对result[“img_1”]和result[“img_2”]施加相同的Albu操作？ (因为autoaug和Albu中有一些操作具有随机性，比如RandomBrightnessContrast， ChannelShuffle，RadomCrop等，所以要确保result[“img_1”]和result[“img_2”]也要受到一模一样的随机操作。 如果我写成类似：

result[“img_1”] = Albu.RandomBrightnessContrast(result[“img_1”])
result[“img_2”] = Albu.RandomBrightnessContrast(result[“img_2"])

则不能保证result[“img_1”]和result[“img_2”]受到相同的RandomBrightnessContrast操作，因为RandomBrightnessContrast有随机性，result[“img_1”]和result[“img_2”]收到的Random是不一致的。 一模一样的意思是比如result[“img_1”]被RadomCrop为224，则result[“img_2”]也必须被crop成224，且其crop的位置必须和result[“img_1”]的一模一样) 谢谢您！期待您的回复！

when i infer my model, i meet this problem. Via the inferrence file of mmclassification, it turns out that the label of the image is 1 and the pred score is 0.8889318704605103. but when i use my own inferrence file, the output of fc is tensor([[ 0.5133, -0.5640]], device='cuda:0'). Do you have an ideas where i have operate incorrectly?

my own inferrence file is as below:

def forward_pytorch(weightfile, image):
    net=resnet.resnet18(num_classes=2)
    checkpoint = torch.load(weightfile)
    new_state_dict = OrderedDict()
    for k, v in checkpoint.items():
        if k == "state_dict":
            for layer_name, weight in v.items():
                layer_name1 = layer_name.replace("backbone.", "")
                layer_name1 = layer_name1.replace("head.", "")
                print(layer_name, layer_name1)
                new_state_dict[layer_name1] = weight
    net.load_state_dict(new_state_dict)
    net.cuda()
    net.eval()
    normalize = transforms.Normalize(mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375])
    transform1 = transforms.Compose([
        transforms.ToTensor(),
        normalize
        ]
    )
    image = transform1(image)  
    image = Variable(image.cuda())
    image = Variable(torch.unsqueeze(image, dim=0).float(), requires_grad=False)
    with torch.no_grad():
        blobs = net.forward(image)
    return blobs

Describe the bug

I was going through Visualization.

I try to ran

python tools/visualizations/vis_cam.py \
    demo/bird.JPEG \
    configs/resnet/resnet50_8xb32_in1k.py \
    https://download.openmmlab.com/mmclassification/v0/resnet/resnet50_batch256_imagenet_20200708-cfb998bf.pth \
    --method GradCAM
    # GradCAM++, XGradCAM, EigenCAM, EigenGradCAM, LayerCAM

It raised the following error:

Traceback (most recent call last):
  File "tools/visualizations/vis_cam.py", line 356, in <module>
    main()
  File "tools/visualizations/vis_cam.py", line 310, in main
    model = init_model(cfg, args.checkpoint, device=args.device)
  File "/public/liushuo/mmclassification-master/mmcls/apis/inference.py", line 34, in init_model
    config.model.pretrained = None
  File "/public/apps/anaconda3/envs/liushuo_UNet/lib/python3.8/site-packages/mmcv/utils/config.py", line 519, in __getattr__
    return getattr(self._cfg_dict, name)
  File "/public/apps/anaconda3/envs/liushuo_UNet/lib/python3.8/site-packages/mmcv/utils/config.py", line 50, in __getattr__
    raise ex
AttributeError: 'ConfigDict' object has no attribute 'model'

Post related information

1. The output of pip list | grep "mmcv\|mmcls\|^torch"

mmcls                          0.23.0                /public/liushuo/mmclassification-master
mmcv-full                      1.5.1
torch                          1.10.2
torchaudio                     0.10.2
torchsummary                   1.5.1
torchvision                    0.11.3

Please give me some advice to fix it. Thanks.

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

I am going to implement semi supervised learning, and reuse multibranch(https://mmdetection.readthedocs.io/en/3.x/user_guides/semi_det.html) from mmdetection. It would raise transform not found exceptions without add those prefix scope.

Also, it's necessary to add the scope since we may not use mmcls as our default scope.

Modification

Add mmcls scope to randaugment transforms

BC-breaking (Optional)

Does the modification introduce changes that break the backward compatibility of the downstream repositories? If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here and update the documentation.

Checklist

Before PR:

• [ ] Pre-commit or other linting tools are used to fix the potential lint issues.
• [ ] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [ ] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [ ] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [ ] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [ ] CLA has been signed and all committers have signed the CLA in this PR.

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Add deployment guide for mmclassifcation models with mmdeploy

Modification

Please briefly describe what modification is made in this PR.

BC-breaking (Optional)

Does the modification introduce changes that break the backward compatibility of the downstream repositories? If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here and update the documentation.

Checklist

Before PR:

• [ ] Pre-commit or other linting tools are used to fix the potential lint issues.
• [ ] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [ ] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [ ] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [ ] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [ ] CLA has been signed and all committers have signed the CLA in this PR.

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Add CSWin model.

Modification

Add CSWin model.

BC-breaking (Optional)

Does the modification introduce changes that break the backward compatibility of the downstream repositories? If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here and update the documentation.

Checklist

Before PR:

• [ ] Pre-commit or other linting tools are used to fix the potential lint issues.
• [ ] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests.
• [ ] The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
• [ ] The documentation has been modified accordingly, like docstring or example tutorials.

After PR:

• [ ] If the modification has potential influence on downstream or other related projects, this PR should be tested with those projects, like MMDet or MMSeg.
• [ ] CLA has been signed and all committers have signed the CLA in this PR.

Branch

master branch (0.24 or other 0.x version)

Describe the bug

After some issues with the v0.x.x I was recommended to move to v1.x.x . !1259

After finishing migrating all the configs I'm now hitting a wall with wanbd, I cannot figure out the equivalent to :

        dict(
            type="WandbLoggerHook",
            init_kwargs={
                "project": "mmcls",
                "name": "will-be-added-by-train-script",
            },
            interval=100,
            log_checkpoint=False,
            log_checkpoint_metadata=False,
            num_eval_images=100,
        ),

in the new version. Has WandbLoggerHook been removed?

The error I get is :

KeyError: 'WandbLoggerHook is not in the hook registry. Please check whether the value of `WandbLoggerHook` is correct or it was registered as expected. More details can be found at https://mmengine.readthedocs.io/en/latest/tutorials/config.html#import-custom-python-modules'

Any help would be much appreciated.

Environment

{'sys.platform': 'linux',
 'Python': '3.8.15 (default, Oct 12 2022, 19:15:16) [GCC 11.2.0]',
 'CUDA available': True,
 'numpy_random_seed': 2147483648,
 'GPU 0': 'NVIDIA GeForce RTX 3060 Laptop GPU',
 'CUDA_HOME': None,
 'GCC': 'x86_64-linux-gnu-gcc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0',
 'PyTorch': '1.11.0+cu113',
 'PyTorch compiling details': 'PyTorch built with:\n'
                              '  - GCC 7.3\n'
                              '  - C++ Version: 201402\n'
                              '  - Intel(R) Math Kernel Library Version '
                              '2020.0.0 Product Build 20191122 for Intel(R) 64 '
                              'architecture applications\n'
                              '  - Intel(R) MKL-DNN v2.5.2 (Git Hash '
                              'a9302535553c73243c632ad3c4c80beec3d19a1e)\n'
                              '  - OpenMP 201511 (a.k.a. OpenMP 4.5)\n'
                              '  - LAPACK is enabled (usually provided by '
                              'MKL)\n'
                              '  - NNPACK is enabled\n'
                              '  - CPU capability usage: AVX2\n'
                              '  - CUDA Runtime 11.3\n'
                              '  - NVCC architecture flags: '
                              '-gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86\n'
                              '  - CuDNN 8.2\n'
                              '  - Magma 2.5.2\n'
                              '  - Build settings: BLAS_INFO=mkl, '
                              'BUILD_TYPE=Release, CUDA_VERSION=11.3, '
                              'CUDNN_VERSION=8.2.0, '
                              'CXX_COMPILER=/opt/rh/devtoolset-7/root/usr/bin/c++, '
                              'CXX_FLAGS= -Wno-deprecated '
                              '-fvisibility-inlines-hidden -DUSE_PTHREADPOOL '
                              '-fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM '
                              '-DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK '
                              '-DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE '
                              '-DEDGE_PROFILER_USE_KINETO -O2 -fPIC '
                              '-Wno-narrowing -Wall -Wextra '
                              '-Werror=return-type '
                              '-Wno-missing-field-initializers '
                              '-Wno-type-limits -Wno-array-bounds '
                              '-Wno-unknown-pragmas -Wno-sign-compare '
                              '-Wno-unused-parameter -Wno-unused-function '
                              '-Wno-unused-result -Wno-unused-local-typedefs '
                              '-Wno-strict-overflow -Wno-strict-aliasing '
                              '-Wno-error=deprecated-declarations '
                              '-Wno-stringop-overflow -Wno-psabi '
                              '-Wno-error=pedantic -Wno-error=redundant-decls '
                              '-Wno-error=old-style-cast '
                              '-fdiagnostics-color=always -faligned-new '
                              '-Wno-unused-but-set-variable '
                              '-Wno-maybe-uninitialized -fno-math-errno '
                              '-fno-trapping-math -Werror=format '
                              '-Wno-stringop-overflow, LAPACK_INFO=mkl, '
                              'PERF_WITH_AVX=1, PERF_WITH_AVX2=1, '
                              'PERF_WITH_AVX512=1, TORCH_VERSION=1.11.0, '
                              'USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, '
                              'USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, '
                              'USE_MKLDNN=OFF, USE_MPI=OFF, USE_NCCL=ON, '
                              'USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, \n',
 'TorchVision': '0.12.0+cu113',
 'OpenCV': '4.6.0',
 'MMEngine': '0.3.2',
 'MMClassification': '1.0.0rc4+c9d87cb'}

Other information

No response

Branch

master branch (0.24 or other 0.x version)

Describe the feature

In the following modules of MixUp, CutMix, and ResizeMix, prob arg is not used to determine whether to apply the augmentation for each sample in the batch. https://github.com/open-mmlab/mmclassification/blob/master/mmcls/models/utils/augment/mixup.py#L31 https://github.com/open-mmlab/mmclassification/blob/master/mmcls/models/utils/augment/cutmix.py#L40 https://github.com/open-mmlab/mmclassification/blob/master/mmcls/models/utils/augment/resizemix.py#L68

This seems to mean that prob arg cannot be used to apply the augmentation (MixUp, CutMix, or ResizeMix) on only some samples in each batch keeping the other samples as is. If the augmentation is applied to the all the samples or none of the samples in each batch, the characteristics of each batch will be inconsistent over iterations and possibly cause the model training unstable.

I'd like to suggest mmclassification supports a feature to apply the mixing operation on only some samples in each batch.

Will you implement it?

• [ ] I would like to implement this feature and create a PR!

分支

master 分支 (0.24 或其他 0.x 版本)

描述该错误

In the Chinese tutorial document, it says evaluation = dict(interval=1, save_best=True, metric='accuracy', metric_options={'topk': (1, )}) , however, the keyword save_best should be str instead of bool , like "auto" or 'accuracy_top-1'

环境信息

python=3.6

其他信息

Thanks for your tutorial documentation! However, I was confused when I met the keyword error. It would be appreciated if you can upgrade the tutorial documentation.