ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' #14

Lhj-cd · 2021-11-03T02:45:07Z

您好，非常感谢您开源您的代码。
我在配置完环境之后发现会报错:ImportError: cannot import name 'Runner_kd' from 'mmcv.runner'。我看到您说从source编译mmcv即可解决这个问题，请问什么是从source编译mmcv？不是指直接运行 python setup.py develop吗？还望指教，谢谢！

The text was updated successfully, but these errors were encountered:

fishlovingcat · 2021-11-08T06:31:42Z

我也遇到了这个问题，请问你解决了吗

Lhj-cd · 2021-11-08T06:36:04Z

没有，作者也还没回复我，期待作者回复

ruiningTang · 2021-11-08T06:44:52Z

安装mmcv的时候不要用mmlab官方的mmcv，用作者自带的mmcv进行安装（clone之后进入mmcv目录编译就行），安装完之后pip list 可以看到mmcv库的location是xxxx/DeFeat.pytorch/mmcv，保证你当前环境运行的mmcv是作者带的，应该就没问题了

fishlovingcat · 2021-11-08T06:46:08Z

感谢你的回复~

…

------------------ 原始邮件 ------------------ 发件人: "ggjy/DeFeat.pytorch" ***@***.***>; 发送时间: 2021年11月8日(星期一) 下午2:45 ***@***.***>; ***@***.******@***.***>; 主题: Re: [ggjy/DeFeat.pytorch] ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' (Issue #14) 您好，非常感谢您开源您的代码。我在配置完环境之后发现会报错:ImportError: cannot import name 'Runner_kd' from 'mmcv.runner'。我看到您说从source编译mmcv即可解决这个问题，请问什么是从source编译mmcv？不是指直接运行 python setup.py develop吗？还望指教，谢谢！安装mmcv的时候不要用mmlab官方的mmcv，用作者自带的mmcv进行安装（clone之后进入mmcv目录编译就行），安装完之后pip list 可以看到mmcv库的location是xxxx/DeFeat.pytorch/mmcv，保证你当前环境运行的mmcv是作者带的，应该就没问题了 — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

ggjy · 2021-11-08T06:47:49Z

感谢@ruiningTang 的回复，因为我在mmcv/runner 下面加了个runner_kd.py 所以直接pip install mmcv是没有我自己加的那个.py的所以会import error。是我没写清楚“从source编译的含义”，就是clone下来之后把runner_kd.py放进去一起编译就行了。

fishlovingcat · 2021-11-08T07:03:38Z

使用作者的代码，解决了刚刚那个问题后我发现又有个“TypeError:conv2d(): argument input must be Tensor, not DataContainer"，你解决import的问题后能跑起来吗 ------------------ 原始邮件 ------------------ 发件人: "ggjy/DeFeat.pytorch" ***@***.***>; 发送时间: 2021年11月8日(星期一) 下午2:47 ***@***.***>; ***@***.******@***.***>; 主题: Re: [ggjy/DeFeat.pytorch] ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' (Issue #14) ***@***.*** 的回复，因为我在mmcv/runner 下面加了个runner_kd.py 所以直接pip install mmcv是没有我自己加的那个.py的所以会import error。是我没写清楚“从source编译的含义”，就是clone下来之后把runner_kd.py放进去一起编译就行了。 — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

ruiningTang · 2021-11-08T07:44:54Z

open-mmlab/mmdetection#2782 (comment)

ggjy · 2021-11-08T13:35:51Z

用python -m torch.distributed.launch --nproc_per_node=8 tools/train.py这个代码训练也会报这个错吗

Lhj-cd · 2021-11-09T01:05:50Z

请问进入mmcv目录编译具体是怎么个操作，作者自带的mmcv没有setup.py文件啊？...我第一次用mmdetection，不是很明白。或者能不能来个傻瓜版的步骤，谢谢。

Lhj-cd · 2021-11-09T01:12:21Z

也是会的

ruiningTang · 2021-11-09T02:38:10Z

我仔细看了一下我的，我是找了mmdetv2.0版本对应的mmcv（我用的0.5.6亲测可行），然后把作者的runner_kd.py复制到mmcv/runner下，复制好之后删除了作者自带的mmcv，再按mmlab官方的从源编译python setup.py develop，亲测可行。还有你那个“TypeError:conv2d()的问题，你是几卡运行？

Lhj-cd · 2021-11-09T08:30:43Z

非常感谢你的答复，按照你的步骤，我已经可以跑通了，再次感谢。我没有遇到TypeError:conv2d()这个问题。
目前的话就是在mmdetection官网下载的resnet50-19c8e357和faster_rcnn_x101_32x4d_fpn权重分别作为student和teacher权重，但是会提示基本上所有的权重都和cfg文件定义的网络结构不匹配..目前不知道怎么解决...

ruiningTang · 2021-11-09T08:40:28Z

R50下不下载都行，我是没有下载直接用的torchvision的，把学生config改一下就行。pretrained='torchvision://resnet50'，加载预训练的时候可能会提示多了unexpected key in source state_dict: fc.weight, fc.bias,这个可以不用管，因为预训练自带的fc在检测里面不会用到。
作者使用的教师模型是R101的，你下x101肯定不匹配，想用x101的话把教师config修改成x101就行，按照mmdet官方的x101配置改就行

Lhj-cd · 2021-11-09T09:02:34Z

明白了，我没注意这个R和X前缀，X是代表ResNeXt吧，多谢大佬多谢大佬~~~

fishlovingcat · 2021-11-10T07:32:29Z

运行环境：

Python: 3.7.11 (default, Jul 27 2021, 14:32:16) [GCC 7.5.0]
CUDA available: True
CUDA_HOME: /usr/local/cuda
NVCC: Build cuda_11.1.TC455_06.29069683_0
GPU 0: NVIDIA GeForce RTX 3060
GCC: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
PyTorch: 1.8.0
PyTorch compiling details: PyTorch built with:

GCC 7.3
C++ Version: 201402
Intel(R) oneAPI Math Kernel Library Version 2021.4-Product Build 20210904 for Intel(R) 64 architecture applications
Intel(R) MKL-DNN v1.7.0 (Git Hash 7aed236906b1f7a05c0917e5257a1af05e9ff683)
OpenMP 201511 (a.k.a. OpenMP 4.5)
NNPACK is enabled
CPU capability usage: AVX2
CUDA Runtime 11.1
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_61,code=sm_61;-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;-gencode;arch=compute_37,code=compute_37
CuDNN 8.0.5
Magma 2.5.2
Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.1, CUDNN_VERSION=8.0.5, 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 -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-variable -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=1.8.0, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON,

TorchVision: 0.9.0
OpenCV: 4.5.4-dev
MMCV: 0.5.1
MMDetection: 2.0.0+unknown
MMDetection Compiler: GCC 7.5
MMDetection CUDA Compiler: 11.1

报错信息：

Traceback (most recent call last):
File "tools/train.py", line 176, in
main()
File "tools/train.py", line 172, in main
meta=meta)
File "/home/lyc/projects/decoupled-feature/mmdet/apis/train.py", line 487, in train_detector
runner.run(data_loaders, cfg.workflow, cfg.total_epochs)
File "/home/lyc/下载/mmcv-0.5.6/mmcv/runner/runner.py", line 383, in run
epoch_runner(data_loaders[i], **kwargs)
File "/home/lyc/下载/mmcv-0.5.6/mmcv/runner/runner.py", line 282, in train
self.model, data_batch, train_mode=True, **kwargs)
File "/home/lyc/projects/decoupled-feature/mmdet/apis/train.py", line 79, in batch_processor
losses = model(**data)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/parallel/distributed.py", line 705, in forward
output = self.module(*inputs[0], **kwargs[0])
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/lyc/projects/decoupled-feature/mmdet/core/fp16/decorators.py", line 49, in new_func
return old_func(*args, **kwargs)
File "/home/lyc/projects/decoupled-feature/mmdet/models/detectors/base.py", line 148, in forward
return self.forward_train(img, img_metas, **kwargs)
File "/home/lyc/projects/decoupled-feature/mmdet/models/detectors/two_stage.py", line 136, in forward_train
x = self.extract_feat(img)
File "/home/lyc/projects/decoupled-feature/mmdet/models/detectors/two_stage.py", line 76, in extract_feat
x = self.backbone(img)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/lyc/projects/decoupled-feature/mmdet/models/backbones/resnet.py", line 594, in forward
x = self.conv1(x)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 399, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/home/lyc/anaconda3/envs/openmmlab-cuda111/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 396, in _conv_forward
self.padding, self.dilation, self.groups)
TypeError: conv2d(): argument 'input' (position 1) must be Tensor, not DataContainer

我运行的命令是python -m torch.distributed.launch --nproc_per_node=1 tools/train.py --gpus 1 --launcher pytorch --validate --work-dir /home/lyc/projects/decoupled-feature/results/faster50 --config configs/faster_rcnn/voc_faster_rcnn_r50.py

ruiningTang · 2021-11-10T07:38:48Z

你运行的是多卡的训练，单卡直接python tools/train.py --validate --work-dir /home/lyc/projects/decoupled-feature/results/faster50 --config configs/faster_rcnn/voc_faster_rcnn_r50.py

Lhj-cd · 2021-11-12T12:11:48Z

最近仔细看了下您的代码，有几个问题想咨询一下。

我看到您在其他issue里面说分类结果的蒸馏是在代码里没有用到，因为提升特别小。但是论文中的table 5应该就是专门对分类部分的蒸馏做了实验，论文中实验证明能够提高student模型1.5个点左右。那对分类部分进行蒸馏效果究竟如何呢？还是说分类头蒸馏对参数特别敏感？
对于特征kd loss，您采用了decouple的方式来分别计算前背景损失。训练过程中，特征kd loss的值(无论是前景还是背景)都比gt loss (包括loss_rpn_cls, loss_rpn_bbox和loss_cls)大很多，甚至高一到两个数量级，这样不会导致模型更倾向于去侧重于降低losskd而忽略了gt损失吗？
对于前背景kd loss，您代码中是设置了3和12两个不同的权重，请问前背景的kd loss权重设置有什么讲究吗？

感谢您的回答。 @ggjy

shuizaola · 2021-11-15T12:43:30Z

你好我想问下，怎么在MMCV目录进行编译呢还是个小白怎么按作者的进行mmcv安装呀？还望指教下谢谢

shuizaola · 2021-11-20T12:33:09Z

怎么样一起编译啊输入什么命令呢

ruiningTang · 2021-11-20T13:05:47Z

先找到对应版本的mmcv，去open-mmlab/mmcv 下git clone，然后把作者的runner_kd.py 拷到mmcv下，再运行官方的编译命令安装mmcv：pip install -e . 可以参考一下这里的从源安装的教程 mmcv v0.5.6

shuizaola · 2021-11-20T13:08:55Z

谢谢你的回复我按你的方法安好了但是现在又报了这种错误
raceback (most recent call last):
File "tools/train.py", line 176, in
main()
File "tools/train.py", line 71, in main
cfg = Config.fromfile(args.config)
File "/home/Disk-2T/DeFeat.pytorch-main/mmcv/utils/config.py", line 165, in fromfile
cfg_dict, cfg_text = Config._file2dict(filename)
File "/home/Disk-2T/DeFeat.pytorch-main/mmcv/utils/config.py", line 84, in _file2dict
filename = osp.abspath(osp.expanduser(filename))
File "/home/caipeng/anaconda3/envs/ENV/lib/python3.7/posixpath.py", line 235, in expanduser
path = os.fspath(path)
TypeError: expected str, bytes or os.PathLike object, not NoneType

1757525671 · 2022-03-28T08:55:17Z

table 5 是ROI损失（使用教师网的proposal在教师网和学生网的rpn特征crop获取ROI，然后再蒸馏两个ROI的分类score），不是检测头损失；
第二个问题不清楚；
对于第三个问题，效果最好的正负样本的比例不是0.05：2么，所以设置了3：12吧；

tongchangD · 2022-12-02T08:11:54Z

您好，冒昧打扰，请问，您说的进行作者自带的mmcv目录进行编译，请问在此处的编译命令是什么

yuhua666 · 2023-02-09T13:53:44Z

您好，为什么我按这个步骤还是报原先的错误？
ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' (/root/miniconda3/lib/python3.8/site-packages/mmcv/runner/init.py)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' #14

ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' #14

Lhj-cd commented Nov 3, 2021

fishlovingcat commented Nov 8, 2021

Lhj-cd commented Nov 8, 2021

ruiningTang commented Nov 8, 2021

fishlovingcat commented Nov 8, 2021 via email

ggjy commented Nov 8, 2021

fishlovingcat commented Nov 8, 2021 via email

ruiningTang commented Nov 8, 2021

ggjy commented Nov 8, 2021

Lhj-cd commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021

ruiningTang commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021 •

edited

Loading

ruiningTang commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021

fishlovingcat commented Nov 10, 2021

ruiningTang commented Nov 10, 2021

Lhj-cd commented Nov 12, 2021

shuizaola commented Nov 15, 2021

shuizaola commented Nov 20, 2021

ruiningTang commented Nov 20, 2021

shuizaola commented Nov 20, 2021

1757525671 commented Mar 28, 2022

tongchangD commented Dec 2, 2022

yuhua666 commented Feb 9, 2023

ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' #14

ImportError: cannot import name 'Runner_kd' from 'mmcv.runner' #14

Comments

Lhj-cd commented Nov 3, 2021

fishlovingcat commented Nov 8, 2021

Lhj-cd commented Nov 8, 2021

ruiningTang commented Nov 8, 2021

fishlovingcat commented Nov 8, 2021 via email

ggjy commented Nov 8, 2021

fishlovingcat commented Nov 8, 2021 via email

ruiningTang commented Nov 8, 2021

ggjy commented Nov 8, 2021

Lhj-cd commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021

ruiningTang commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021 • edited Loading

ruiningTang commented Nov 9, 2021

Lhj-cd commented Nov 9, 2021

fishlovingcat commented Nov 10, 2021

ruiningTang commented Nov 10, 2021

Lhj-cd commented Nov 12, 2021

shuizaola commented Nov 15, 2021

shuizaola commented Nov 20, 2021

ruiningTang commented Nov 20, 2021

shuizaola commented Nov 20, 2021

1757525671 commented Mar 28, 2022

tongchangD commented Dec 2, 2022

yuhua666 commented Feb 9, 2023

Lhj-cd commented Nov 9, 2021 •

edited

Loading