Lionel Peer

facing the same issue, running it in a docker container `https://hub.docker.com/r/lionelpeer/sumo-rl`

Hi @jafarinia 👋 glad you are enjoying lightly! I will try to dissect a bit, which also helps myself figuring out if we have an issue at hand here, the...

@jafarinia A PR that allows dispatching to both implementation would of course be highly welcome. However, in that case we should also consider updating the `tests/loss/test_msn_loss.py` and try to test...

Hi @siemdejong, That looks like a sensible approach! What other arguments that are not exposed did you find limiting? Would you be willing to create a PR for it?

Thanks @siemdejong ! I am not sure yet if we really want to pack all possible parameters (`interpolation`, `antialias`, ...) into the constructors of those transforms or if we instead...

Hey @siemdejong! Sorry for the delayed response: I created an internal issue when you submitted this, however unfortunately your request is currently not at the top of the priority list....

Hi @isaacrob-roboflow! We now also support distilling DINOv3 into RF-DETR backbones.

Hey @ROYUNSW! With LightlyTrain you can leverage knowledge distillation from DINOv3 together with RF-DETR or any other vision backbone: https://github.com/lightly-ai/lightly-train

Hey @cgebbe, We fixed this in the PRs from #1717 to #1723 , such that `torchvision.transforms.v2` is used if it is installed. Since this will be available in the next...

Hi @swamidass! That looks like a really promising paper indeed, thanks for making us aware of it. We'll go over it internally and see what we can do about it.