No module named 'mmselfsup.engine.runner'

[Syed05]

After I run the command below:

python tools/train.py configs/selfsup/base/mae_vit-base-p16_Test.py

It get the error below:

return _bootstrap._gcd_import(name[level:], package, level)
File "", line 1014, in _gcd_import
File "", line 991, in _find_and_load
File "", line 973, in _find_and_load_unlocked
ModuleNotFoundError: No module named 'mmselfsup.engine.runner'

How to solve it?

Thanks.

My config file is:

`base = [
'../base/models/mae_vit-base-p16.py',
'../base/datasets/imagenet_mae.py',
'../base/schedules/adamw_coslr-200e_in1k.py',
'../base/default_runtime.py',
]

train_dataloader = dict(batch_size=2, num_workers=2)

optimizer = dict(
type='AdamW', lr=1.5e-4 * 4096 / 256, betas=(0.9, 0.95), weight_decay=0.05)
optim_wrapper = dict(
type='OptimWrapper',
optimizer=optimizer,
paramwise_cfg=dict(
custom_keys={
'ln': dict(decay_mult=0.0),
'bias': dict(decay_mult=0.0),
'pos_embed': dict(decay_mult=0.),
'mask_token': dict(decay_mult=0.),
'cls_token': dict(decay_mult=0.)
}))

param_scheduler = [
dict(
type='LinearLR',
start_factor=1e-4,
by_epoch=True,
begin=0,
end=40,
convert_to_iter_based=True),
dict(
type='CosineAnnealingLR',
T_max=360,
by_epoch=True,
begin=40,
end=400,
convert_to_iter_based=True)
]

train_cfg = dict(max_epochs=400)
default_hooks = dict(
logger=dict(type='LoggerHook', interval=100),
# only keeps the latest 3 checkpoints
checkpoint=dict(type='CheckpointHook', interval=1, max_keep_ckpts=3))

randomness = dict(seed=0, diff_rank_seed=True)
resume = True`

[HelloSZS]

i encounter this problem too

[HelloSZS]

And i now switched to 1.0.0rc6 and installed, the problem fixed.

[fangyixiao18]

And i now switched to 1.0.0rc6 and installed, the problem fixed.

#746
The pr will fix this problem

[1234532314342]

I also encountered the same problem, it is useless to change the version, how to solve it?

[HelloSZS]

And i now switched to 1.0.0rc6 and installed, the problem fixed.

#746 The pr will fix this problem

i found that the contents of registry.py in the main branch
https://github.com/open-mmlab/mmselfsup/blob/main/mmselfsup/registry.py

are different from that in the v1.0.0rc6 branch
https://github.com/open-mmlab/mmselfsup/blob/v1.0.0rc6/mmselfsup/registry.py

[fangyixiao18]

And i now switched to 1.0.0rc6 and installed, the problem fixed.

#746 The pr will fix this problem

i found that the contents of registry.py in the main branch https://github.com/open-mmlab/mmselfsup/blob/main/mmselfsup/registry.py

are different from that in the v1.0.0rc6 branch https://github.com/open-mmlab/mmselfsup/blob/v1.0.0rc6/mmselfsup/registry.py

It is because mmengine update the some logic in registry, so we need to update the this part of code

[fangyixiao18]

I also encountered the same problem, it is useless to change the version, how to solve it?

please git pull the main branch to update your code, the pr is merged

[Syed05]

Thanks, after git pull main branch its working now. However, can anyone please let me know how to structure the custom data folder?

My folder structure is:

data
------meta
------- train.txt (contains the image paths)
------train
------- .jpg images

when I run the config I am getting following error:

ValueError: class EpochBasedTrainLoop in mmengine/runner/loops.py: class CustomDataset in mmcls/datasets/custom.py: not enough values to unpack (expected 2, got 1)

My config file is:

`# dataset settings
dataset_type = 'mmcls.CustomDataset'
data_root = 'data/'
custom_imports = dict(imports='mmcls.datasets', allow_failed_imports=False)

view_pipeline = [
dict(
type='RandomResizedCrop', size=224, scale=(0.2, 1.), backend='pillow'),
dict(
type='RandomApply',
transforms=[
dict(
type='ColorJitter',
brightness=0.4,
contrast=0.4,
saturation=0.4,
hue=0.1)
],
prob=0.8),
dict(
type='RandomGrayscale',
prob=0.2,
keep_channels=True,
channel_weights=(0.114, 0.587, 0.2989)),
dict(type='RandomGaussianBlur', sigma_min=0.1, sigma_max=2.0, prob=0.5),
dict(type='RandomFlip', prob=0.5),
]

train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='MultiView', num_views=2, transforms=[view_pipeline]),
dict(type='PackSelfSupInputs', meta_keys=['img_path'])
]

train_dataloader = dict(
batch_size=32,
num_workers=8,
drop_last=True,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
collate_fn=dict(type='default_collate'),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file='meta/train.txt',
data_prefix=dict(img_path='train/'),
pipeline=train_pipeline))

model = dict(
type='MoCo',
queue_len=65536,
feat_dim=128,
momentum=0.999,
data_preprocessor=dict(
mean=(123.675, 116.28, 103.53),
std=(58.395, 57.12, 57.375),
bgr_to_rgb=True),
backbone=dict(
type='ResNet',
depth=50,
in_channels=3,
out_indices=[4], # 0: conv-1, x: stage-x
norm_cfg=dict(type='BN')),
neck=dict(
type='MoCoV2Neck',
in_channels=2048,
hid_channels=2048,
out_channels=128,
with_avg_pool=True),
head=dict(
type='ContrastiveHead',
loss=dict(type='mmcls.CrossEntropyLoss'),
temperature=0.2))

optimizer = dict(type='SGD', lr=0.03, weight_decay=1e-4, momentum=0.9)
optim_wrapper = dict(type='OptimWrapper', optimizer=optimizer)

param_scheduler = [
dict(type='CosineAnnealingLR', T_max=200, by_epoch=True, begin=0, end=200)
]

train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=200)

default_scope = 'mmselfsup'

default_hooks = dict(
runtime_info=dict(type='RuntimeInfoHook'),
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=10),
sampler_seed=dict(type='DistSamplerSeedHook'),
)

env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)

log_processor = dict(
window_size=10,
custom_cfg=[dict(data_src='', method='mean', windows_size='global')])

vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='SelfSupVisualizer', vis_backends=vis_backends, name='visualizer')

log_level = 'INFO'
load_from = None
resume = False `

[kalikhademi]

Thanks, after git pull main branch its working now. However, can anyone please let me know how to structure the custom data folder?

My folder structure is:

data ------meta ------- train.txt (contains the image paths) ------train ------- .jpg images

when I run the config I am getting following error:

ValueError: class EpochBasedTrainLoop in mmengine/runner/loops.py: class CustomDataset in mmcls/datasets/custom.py: not enough values to unpack (expected 2, got 1)

My config file is:

`# dataset settings dataset_type = 'mmcls.CustomDataset' data_root = 'data/' custom_imports = dict(imports='mmcls.datasets', allow_failed_imports=False)

view_pipeline = [ dict( type='RandomResizedCrop', size=224, scale=(0.2, 1.), backend='pillow'), dict( type='RandomApply', transforms=[ dict( type='ColorJitter', brightness=0.4, contrast=0.4, saturation=0.4, hue=0.1) ], prob=0.8), dict( type='RandomGrayscale', prob=0.2, keep_channels=True, channel_weights=(0.114, 0.587, 0.2989)), dict(type='RandomGaussianBlur', sigma_min=0.1, sigma_max=2.0, prob=0.5), dict(type='RandomFlip', prob=0.5), ]

train_pipeline = [ dict(type='LoadImageFromFile'), dict(type='MultiView', num_views=2, transforms=[view_pipeline]), dict(type='PackSelfSupInputs', meta_keys=['img_path']) ]

train_dataloader = dict( batch_size=32, num_workers=8, drop_last=True, persistent_workers=True, sampler=dict(type='DefaultSampler', shuffle=True), collate_fn=dict(type='default_collate'), dataset=dict( type=dataset_type, data_root=data_root, ann_file='meta/train.txt', data_prefix=dict(img_path='train/'), pipeline=train_pipeline))

model = dict( type='MoCo', queue_len=65536, feat_dim=128, momentum=0.999, data_preprocessor=dict( mean=(123.675, 116.28, 103.53), std=(58.395, 57.12, 57.375), bgr_to_rgb=True), backbone=dict( type='ResNet', depth=50, in_channels=3, out_indices=[4], # 0: conv-1, x: stage-x norm_cfg=dict(type='BN')), neck=dict( type='MoCoV2Neck', in_channels=2048, hid_channels=2048, out_channels=128, with_avg_pool=True), head=dict( type='ContrastiveHead', loss=dict(type='mmcls.CrossEntropyLoss'), temperature=0.2))

optimizer = dict(type='SGD', lr=0.03, weight_decay=1e-4, momentum=0.9) optim_wrapper = dict(type='OptimWrapper', optimizer=optimizer)

param_scheduler = [ dict(type='CosineAnnealingLR', T_max=200, by_epoch=True, begin=0, end=200) ]

train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=200)

default_scope = 'mmselfsup'

default_hooks = dict( runtime_info=dict(type='RuntimeInfoHook'), timer=dict(type='IterTimerHook'), logger=dict(type='LoggerHook', interval=50), param_scheduler=dict(type='ParamSchedulerHook'), checkpoint=dict(type='CheckpointHook', interval=10), sampler_seed=dict(type='DistSamplerSeedHook'), )

env_cfg = dict( cudnn_benchmark=False, mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0), dist_cfg=dict(backend='nccl'), )

log_processor = dict( window_size=10, custom_cfg=[dict(data_src='', method='mean', windows_size='global')])

vis_backends = [dict(type='LocalVisBackend')] visualizer = dict( type='SelfSupVisualizer', vis_backends=vis_backends, name='visualizer')

log_level = 'INFO' load_from = None resume = False `

Did you find any solution for this as I am facing the same problem?

[leonidas-timefly]

To sum up, the solution for this issue is:
Step 1: install mmselfsup version 1.0.0rc6
pip install 'mmselfsup>=1.0.0rc6'
Step 2: go to the directory "mmselfsup/" and change to the main branch with:
git branch
and
git checkout main
Step 3: cover the old version with:
git pull origin main
Then this issue is solved.

However, if you still encounter this issue, you can try to reinstall mmselfsup from the source first, then follow this repair tutorial steps 2-3.

[hadhoryth]

The problem is the registry, for some reason the location of mmselfsup.engine.runner is not properly updated. The quick fix would be edit the registry.py file and remove locations argument.

RUNNERS = Registry("runner", parent=MMENGINE_RUNNERS)
RUNNER_CONSTRUCTORS = Registry("runner constructor", parent=MMENGINE_RUNNER_CONSTRUCTORS)
LOOPS = Registry("loop", parent=MMENGINE_LOOPS)