Linus Härenstam-Nielsen
Linus Härenstam-Nielsen
@mbanani thanks for the info! In fact, I just trained the `dinov2_l14` model with 4 GPUs and batch size 2 (so in total the same batch size as my other...
Same issue here, downgrading jupyter-server didn't work either edit: not sure why but changing the port from 8888 to 8885 made the warnings go away
As a temporary fix I found that calling `cv2.imshow` before `import av` prevents the freeze: ```python import numpy as np import cv2 cv2.imshow('ffmpeg fix', np.array([1], dtype=np.uint8)) cv2.destroyAllWindows() import av red...
@yiyiliao that would be great! I'm looking into implementing something similar-ish, so would be helpful to have it as a starting point
I am wondering this as well. Are evaluation scripts for replicating the paper tables available anywhere? If not, any plans to release them?
@kbmufti kind of, I moved on to using the pytorch version, where the pre-processing is applied automatically if you pass in a .xyz file [see here](https://github.com/mabaorui/NeuralPull-Pytorch/blob/master/models/dataset.py#L19) I should say the...
I followed the instructions here https://github.com/mabaorui/NeuralPull/issues/2#issuecomment-880653009