Ma Zerun

Sorry, we don't support ray by now. To train models in multiple devices, we support using slurm, which refer to the https://github.com/open-mmlab/mmclassification/blob/master/tools/slurm_train.sh

The docstring is broken. 

> Need to add test to show case the new features Done

> Sounds good, I will try it.

> > ```python > _base_ = './base.py' > list_variable[1] = 5 > x, y = tuple_variable > dict_variable = dict( > a=dict(third=3, _delete_=True), > b=dict(third=3) > ) > ``` >...

Any updates on this problem? There are too many noisy warnings during debugging. Or any method to suppress these warnings temporarily?

Same issue.

Please try to calculate the confusion matrix according to the comment https://github.com/open-mmlab/mmclassification/issues/598#issuecomment-992064599

Please add it to the `model-zoo.md` and `models.rst`.

Looks the `WeightRandomSampler` is a good feature and maybe you can create a PR to add it to mmcls