Ma Zerun

Yes, we support multi-label tasks, You can implement your dataset by referring to the `VOC` dataset. And the head of model should be `MultiLabelLinearHead`.

Can you provide your `configs/resnet/resnet50_8xb32_in1k.py` file?

A little strange, can you execuate the below command in Python interpreter, and provide the output ```python >>> from mmcv import Config >>> from mmcls.apis import init_model >>> cfg =...

Ok, please print the `cfg` variable

Can you provide your raw `configs/resnet/resnet50_8xb32_in1k.py`? Looks `mmcv.Config` failed to read your config file. Maybe an mmcv issue.

What about `configs/_base_/models/resnet50.py`？What's the content of it?

Please try to set ```python torch.backends.cuda.matmul.allow_tf32 = False torch.backends.cudnn.allow_tf32 = False ``` At the beginning of the script and test again on FP32.

Sounds like a good feature. We will review it recently. In addition, MMCLS is now developed based on the `dev` branch, and I have switched this branch's base to `dev`....

Hello, I have modified some APIs to support the entire multi-task design. More details can be found in #808 In the design, we use a dict instead of a tuple...

Looks like you don't have dataset folder