Justin Ruan

Convert the .xlsx file to csv can solve the issue.

You can find it from [here](https://github.com/YOUSIKI/PyTorch-AlexNet/releases/download/v0.1/lrn-epoch-099.pth).

I removed some of the weights to fit the partial AlexNet. To fix this issue, I had uploaded a weight file. You can try it first, thanks.

Not sure why this happened. Did you check the name of your column? Maybe any invisible space exists before or after the column name.

Now your "Onset" string contains now only the filename but also your full path. Try extracting only the filename by: ```python os.path.basename("") ``` to avoid errors.

Hi, not sure why the problem occurs, all my tries are also collapsing to label 3.

Try [initialising the weights](https://github.com/huggingface/pytorch-image-models/blob/3d05c0e86f2f4c57bf495468aa3f8a7d3487c986/timm/models/vision_transformer.py#L504) properly. (Sorry for not testing this before)

@cheind thanks for pointing out that, but we need to compute different threshold to obtain the $\text{HOTA}_\alpha$ before computing the final $\text{HOTA}$. Is this possible with cache?

Thresholds are fixed and should be `[0.05, 0.1, ..., 0.95]`. The current PR supports all the needed predifined functions (added to `metrics.py`). However, the current iimplementation is inefficient and not...

@cheind i tried using the idea from COCO, but this might take more time to ensure this won't break all the others. It is ok to be merged but the...