When I ran the script pruning/get_channel_units.py, I got this error: TypeError: expected Tensor as element 0 in argument 0, but got tuple.

[HePengguang]

Describe the bug

A clear and concise description of what the bug is.

When I ran the script pruning/get_channel_units.py, I got this error: TypeError: expected Tensor as element 0 in argument 0, but got tuple.

To Reproduce

The command you executed.

I just used the recommended conmand: python tools/pruning/get_channel_units.py \ configs/pruning/mmseg/dcff/dcff_pointrend_resnet50_8xb2_cityscapes.py \ -c -i --output-path=configs/pruning/mmseg/dcff/resnet_seg.json

Post related information

1. The output of pip list | grep "mmcv\|mmrazor\|^torch" mmcv 2.0.0 mmrazor 1.0.0 d:\allcode\mmlab\mmrazor-main torch 1.13.1+cu117

2. Your config file if you modified it or created a new one. I changed the script dcff_pointrend_resnet50_8xb2_cityscapes.py base = [

   TODO: use autoaug pipeline.

   'mmseg::base/datasets/feces.py', # from cityscapes to feces (my own mmseg dataset) 'mmseg::base/schedules/schedule_160k.py', 'mmseg::base/default_runtime.py', './pointrend_resnet50.py' ]

feces.py

dataset settings

dataset_type = 'Feces' data_root = 'D:\Allcode\mmlab\mmsegmentation-dev-1.x\data\data'

img_scale = (1024, 1024) crop_size = (512, 512) train_pipeline = [ dict(type='LoadImageFromFile'), dict(type='LoadAnnotations'), dict( type='RandomResize', scale=img_scale, ratio_range=(0.5, 2.0), keep_ratio=True),

dict(type='RandomMosaic', img_scale=img_scale, prob=0.5, pad_val=0), # batch_size是2，不知道能不能成功使用，因为它需要4张图片

dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='PackSegInputs')

] test_pipeline = [ dict(type='LoadImageFromFile'), dict(type='Resize', scale=img_scale, keep_ratio=True),

add loading annotation after Resize because ground truth

# does not need to do resize data transform
dict(type='LoadAnnotations'),
dict(type='PackSegInputs')

] img_ratios = [0.5, 0.75, 1.0, 1.25, 1.5, 1.75] tta_pipeline = [ dict(type='LoadImageFromFile', backend_args=None), dict( type='TestTimeAug', transforms=[ [ dict(type='Resize', scale_factor=r, keep_ratio=True) for r in img_ratios ], [ dict(type='RandomFlip', prob=0., direction='horizontal'), dict(type='RandomFlip', prob=1., direction='horizontal') ], [dict(type='LoadAnnotations')], [dict(type='PackSegInputs')] ]) ] train_dataloader = dict( batch_size=2, num_workers=4, persistent_workers=True, sampler=dict(type='InfiniteSampler', shuffle=True), dataset=dict( type='RepeatDataset', times=40000, dataset=dict( type=dataset_type, data_root=data_root, data_prefix=dict( img_path='images/training', seg_map_path='annotations/training'), pipeline=train_pipeline))) val_dataloader = dict( batch_size=1, num_workers=2, persistent_workers=True, sampler=dict(type='DefaultSampler', shuffle=False), dataset=dict( type=dataset_type, data_root=data_root, data_prefix=dict( img_path='images/validation', seg_map_path='annotations/validation'), pipeline=test_pipeline)) test_dataloader = val_dataloader

val_evaluator = dict(type='IoUMetric', iou_metrics=['mDice', 'mIoU', 'mFscore']) test_evaluator = val_evaluator

4. Your train log file if you meet the problem during training. d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\builder.py:36: UserWarning: build_loss would be deprecated soon, please use mmseg.registry.MODELS.build() warnings.warn('build_loss would be deprecated soon, please use ' d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\losses\cross_entropy_loss.py:235: UserWarning: Default avg_non_ignore is False, if you would like to ignore the certain label and average loss over non-ignore labels, which is the same with PyTorch officia l cross_entropy, set avg_non_ignore=True. warnings.warn( Traceback (most recent call last): File "tools/pruning/get_channel_units.py", line 84, in main() File "tools/pruning/get_channel_units.py", line 48, in main model = MODELS.build(config['model']) File "G:\softwares\path\Anaconda3\envs\mmseg\lib\site-packages\mmengine\registry\registry.py", line 570, in build return self.build_func(cfg, args, kwargs, registry=self) File "G:\softwares\path\Anaconda3\envs\mmseg\lib\site-packages\mmengine\registry\build_functions.py", line 232, in build_model_from_cfg return build_from_cfg(cfg, registry, default_args) File "G:\softwares\path\Anaconda3\envs\mmseg\lib\site-packages\mmengine\registry\build_functions.py", line 121, in build_from_cfg obj = obj_cls(args) # type: ignore File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\algorithms\pruning\dcff.py", line 62, in init super().init(architecture, mutator_cfg, data_preprocessor, File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\algorithms\pruning\ite_prune_algorithm.py", line 137, in init self.mutator.prepare_from_supernet(self.architecture) File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\mutators\channel_mutator\channel_mutator.py", line 113, in prepare_from_supernet units = self._prepare_from_tracer(supernet, self.parse_cfg) File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\mutators\channel_mutator\channel_mutator.py", line 311, in _prepare_from_tracer unit_configs = tracer.analyze(model) File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\task_modules\tracer\channel_analyzer.py", line 103, in analyze path_list = self.tracer.trace(model) File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\task_modules\tracer\backward_tracer.py", line 187, in trace pseudo_loss = self.loss_calculator(model) File "d:\allcode\mmlab\mmrazor-main\mmrazor\models\task_modules\tracer\loss_calculator\sum_loss_calculator.py", line 23, in call pseudo_output = model(pseudo_img) File "G:\softwares\path\Anaconda3\envs\mmseg\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl return forward_call(input, **kwargs) File "d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\segmentors\base.py", line 98, in forward return self._forward(inputs, data_samples) File "d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\segmentors\cascade_encoder_decoder.py", line 136, in _forward out = self.decode_head[i].forward(x, out) File "d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\decode_heads\point_head.py", line 127, in forward x = torch.cat([fine_grained_point_feats, coarse_point_feats], dim=1) TypeError: expected Tensor as element 0 in argument 0, but got tuple

5. Other code you modified in the mmrazor folder. nothing

Additional context

Add any other context about the problem here.

[here]

[LKJacky]

Please use "Fxtracer" to replace "BackwardTracer". We cannot maintain BackwardTracer anymore.

[HePengguang]

Please use "Fxtracer" to replace "BackwardTracer". We cannot maintain BackwardTracer anymore.

Thanks for your reply, do you mean "MMFxTracer" or "FxTracer"? I tried to modify the "BackwardTracer" to "MMFxTracer" or "FxTracer" in some files. But there is still that error: File "d:\allcode\mmlab\mmsegmentation-dev-1.x\mmseg\models\decode_heads\point_head.py", line 127, in forward x = torch.cat([fine_grained_point_feats, coarse_point_feats], dim=1) TypeError: expected Tensor as element 0 in argument 0, but got tuple,

I think it's not the problem about "BackwardTracer", so I still do not know what should be modified. Could you please supply a more detailed answer? Thank you so much! I chaned these lines:

[HePengguang]

Please use "Fxtracer" to replace "BackwardTracer". We cannot maintain BackwardTracer anymore.

https://github.com/open-mmlab/mmrazor/tree/main/configs/pruning/mmseg/dcff I refer to this recommendation command.

[HePengguang]

Please use "Fxtracer" to replace "BackwardTracer". We cannot maintain BackwardTracer anymore.

Is it possible that your documentation does not match the repository code? As someone with some coding experience, I have no idea how to use mmrazor to prune my mmseg model. Could you consider improving the documentation? Thank you so much for your hard work.