Tim Meinhardt

Thank you for providing those subsets. I hope you can answer another question cause from the code it looks like you used the model trained on _pseudo key-reference frame pairs_...

I just checked out the _ruamel_ package and it indeed supports loading/dumping YAML comments. Here is a small example of how to load a YAML configuration and access its comments:...

This is actually an interesting idea for a normalisation of the observations. The resulting mean and standard deviation would then be used as starting points for the running mean and...

Can I suggest [Sacred](https://github.com/IDSIA/sacred) as an experiment tool which allows to store and organise experiments. In particular the functionality of loading configs from YAML files and injecting them with the...

hi @ranahanocka unfortunately my code in relation to this repository is private and can not be published yet. However, I use Sacred in another [project](https://github.com/michimoeller/liftingLayers/blob/master/python_experiments/src/train_denoising.py) of mine. I have some...

I suggest to combine testing and an automated generation of new pretrained models for the changed codebase (see #71).

Matching the baseline setup is of course a good argument. Maybe they thought about this. But it is definitely something that one should keep in mind. I will post the...

Step 4 [here](https://github.com/timmeinhardt/trackformer/blob/main/docs/INSTALL.md).

These masks specify where padding was added to the image and are used to ignore those regions during the attention computation. If you do not have padding just initialize masks...

I have not resolved the issue. But checkout the new SeqFormer [repo](https://github.com/wjf5203/VNext/tree/main/projects/SeqFormer) maybe they resolved the issue over there.